DIGITAL

MARKETING
ANALYTICS
In Theory And In Practice
Second Edition

KEVIN HARTMAN
Copyright © 2020
Kevin Hartman
Digital Marketing Analytics
In Theory And In Practice
 Second Edition
All rights reserved.
No part of this publication may be reproduced, distributed, or transmitted in any
form or by any means, including photocopying, recording, or other electronic or
mechanical methods, without the prior written permission of the publisher,
except in the case of brief quotations embodied in critical reviews and certain
other non-commercial uses permitted by copyright law.
 Kevin Hartman
Printed in the United States of America
First Printing April 2020
First Edition April 2020
Second Printing September2020
Second Edition September2020
ISBN: 979-8638631321
 To my family who has loved me,
my friends who have supported me,
and my colleagues who have
inspired me.
PRAISE FOR DIGITAL
MARKETING ANALYTICS

“Kevin has made science accessible. With the immense

potential Digital holds, I’m confident this book will allow you
to unlock new ways to deepen customer relationships and
fatten your profits!”
Avinash Kaushik Digital Marketing Evangelist - Google,
Author – Web Analytics: An Hour A Day and Web Analytics
2.0

“Kevin combines the experience of a skilled analytics

practitioner with the expertise of an academic. In this book,
he presents a fresh look at the field through the lens of
history: from its earliest beginnings to its modern
incarnation.”
Mike Chapple Director, M.S. in Business Analytics –
Mendoza College of Business, The University of Notre Dame
“Digital Marketing Analytics is an important resource for all
marketers, whether you are early in your career and
learning or a seasoned practitioner who wants to remain
sharp. This book will help you understand the changing
dynamics and apply that thinking to maximize your business
results.”
Karen Sauder Vice President – Google

“This book is an excellent examination of data analytics,

providing both a solid theoretical foundation and a deep
dive into practical application. Readers of all levels of digital
savvy will benefit from Kevin’s expertise across platforms,
tools, and approaches.”
W. Brooke Elliott Associate Dean and EY Professor – Gies
College of Business, University of Illinois
“The power of Digital Marketing Analytics is that it provides
an insightful approach to the art of analytics while
demystifying its science. Whether you’re a big brand or
small company, Kevin’s book is a fantastic resource for any
marketer who wants to do more with data.”
Mike Clarke Director, Research – Facebook

“Kevin has honed wisdom gathered over decades as a

practitioner and lecturer to develop a strong foundation of
analytics. As the evolution of marketing analytics continues
to accelerate, this foundation will prepare you to
successfully tackle any new challenge, approach, or data
source you will encounter throughout your career!”
Stan Balanovskiy Head of Ads Measurement and Insights –
Quibi
TABLE OF CONTENTS

INTRODUCTION
PART 1: The Day the Geeks Took Over
PART 1 | LESSON 1
The ART+SCIENCE Mind
The Three Roles of the Analyst
The Evolution of the Analyst in the Firm
Characteristics of the Successful Analyst
In Summary: The ART+SCIENCE Mind
PART 1 | LESSON 2
The Early History of Data Analytics
Epoch I: ‘Early Maps and Diagrams’
Epoch II: ‘Measurement & Theory’
Epoch III: ‘New Graphical Forms’
Epoch IV: ‘The Beginnings of Modern Graphics’
Epoch V: ‘The Golden Age of Data Analytics’
Epoch VI: ‘The Modern Dark Ages’ of Data Analytics
Epoch VII: ‘The Rebirth of Data Analytics’
Epoch VIII: ‘High-Definition Data Analytics’ (1975-1994)
In Summary: The Early History of Data Analytics
PART 1 | LESSON 3
The Contemporary History of Data Analytics
Epoch I: ‘When Anything Was Possible’
Epoch II: ‘The Bubble and the Burst’
Epoch III: ‘The Seeds of Prosperity’
Epoch IV: ‘The Age of Unicorns’
The Significance of 2014
In Summary: The Contemporary History of Data Analytics
PART 1 | LESSON 4
The Rise of Modern Data Analytics
P&G’s Three-Step Model of Marketing
 The Introduction of the ‘Zero Moment of Truth’ (ZMOT)
Technological Influences
Market Influences
Placing Today’s Data Creation in Context
Rise of the Machines: Machine Learning and Analytics
In Summary: The Rise of Modern Data Analytics
PART 2: Consumer/Brand Relationships
PART 2 | LESSON 1
Online Video
Online Video (OLV) Market Share
Consumer Behavior and OLV
Key OLV Measures
In Summary: Online Video
PART 2 | LESSON 2
Online Search
Search Engine Market Share
Search Engine Management Versus Search Engine
Optimization
The Importance of Search Rank
Consumer Behavior and Search
Key Search Measures
In Summary: Online Search
PART 2 | LESSON 3
Display Media
Display Media Market Share
Evolution of Display Ad Formats
Key Display Media Measures
In Summary: Display Media
PART 2 | LESSON 4
Social Media
Social Media Market Share
Consumer Behavior and Social Media
How Brands Use Social Media
Key Social Media Measures
In Summary: Social Media
PART 2 | LESSON 5
 The Consumer Decision Journey
The Introduction of McKinsey’s CDJ
The Steps of the CDJ
Applicability of the CDJ
How Ad Blockers Affect Digital Measurement
In Summary: The Consumer Decision Journey
PART 3: The Science of Analytics
PART 3 | LESSON 1
Digital Data Infrastructure
Categories of Data
Digital Data Collection: Cookies, Tags and Sign-Ins
Data Availability and Value
Data Collection and the Effect of Privacy Concerns
In Summary: Digital Data Infrastructure
PART 3 | LESSON 2
Digital Measurement
The Four Categories of Digital Measurement
Making the Intangible Tangible: Brand Impact
Measuring Sales Levers: Consumer Outcomes
Understanding What Is Important: Customer Value
Assessing Marketing Effectiveness: Attribution
Getting to True Lift: Incrementality Testing
A Word on Clarity through Measurement Multiplicity
A Word on Digital Measurement Challenges
In Summary: Digital Measurement
PART 3 | LESSON 3
Analytics and Dataviz Tools
Key Categories of Digital Analysis Tools
Evaluation Criteria 1: Data Flexibility
Evaluation Criteria 2: Ease of Use
A Word on SQL
A Word on R and Python
A Word on Tools < The Analyst
In Summary: Analytics and Dataviz Tools
PART 3 | LESSON 4
Digital Marketing Maturity
 BCG’s View of Digital Marketing Maturity
Bain’s View of Digital Marketing Maturity
Deloitte and MIT Sloan’s View of Digital Marketing
Maturity
Benefits of Digital Marketing Maturity
In Summary: Digital Marketing Maturity
PART 4: The Art of Analytics
PART 4 | LESSON 1
Navigating to Your Big Idea
The Four Steps of the Marketing Analytics Process (MAP)
First Step: Plan
Second Step: Collect
Third Step: Analyze
Step Four: Report
‘The Ask Behind the Ask’
In Summary: Navigating to Your Big Idea
PART 4 | LESSON 2
Planning for Your Analytics Expedition
When Objective-Setting Goes Awry
The Six Marketing Objectives
Determining Your Marketing Objective
Prioritizing Objectives
Building an Analysis Plan around Your Objective
A Word on Anchoring Bias
In Summary: Planning for Your Analytics Expedition
PART 4 | LESSON 3
Collecting Data, Data Everywhere
Unstructured Versus Structured Data
Collecting Data
Ensuring Data Quality
A Word on Data Tidying
A Word on Managing Collected Data
In Summary: Collecting Data, Data Everywhere
PART 4 | LESSON 4
Analyzing for Insights
The Five Categories of Marketing Data Analysis
 A Word on the Importance of Context
Finding Patterns in Data
Countering Bias in Analysis
In Summary: Analyzing for Insights
PART 5: Storytelling with Data
PART 5 | LESSON 1
Pictures You See with Your Brain
Visual Perception and the Door Study
Preattentive Attributes
A Word on the Effect of Neuroscience
In Summary: Pictures You See with Your Brain
PART 5 | LESSON 2
Evaluation Framework for the Visual Form
Evaluating the Effectiveness of Dataviz
Connecting Data Visual Elements to the Analysis Journey
Understanding the Components of Visual Form
In Summary: Evaluation Framework for the Visual Form
PART 5 | LESSON 3
Sophisticated Use of Contrast
Size Contrast
Color Contrast
Shape Contrast
Contrived Contrast
Bringing Contrast to Numbers through Context
A Word on the Importance of Sketching Visuals
In Summary: Sophisticated Use of Contrast
PART 5 | LESSON 4
Ensuring Clear Meaning
Properly Titling Your Visual
Highlighting Messages Visually
A Word on Leveraging Emotions
In Summary: Ensuring Clear Meaning
PART 5 | LESSON 5
Refined Execution through Visual Polish
Color Theory Application in Charts
Harmonic Color Themes
 Additional Color Considerations
Elements of Refined Design
Testing to Improve the Visual
A Word on Combining Tools to Refine Visuals
In Summary: Refined Execution through Visual Polish
PART 5 | LESSON 6
On Your Feet and Getting Your Story Across
‘The McCandless Method’ of Data Presentation
The McCandless Method in Practice
A Word on Presentation Style
The Storyteller’s Dash: Dashboarding Data Effectively
A Word on Building Habits over Time
In Summary: On Your Feet and Getting Your Story Across
CONCLUSION
GLOSSARY
ABOUT THE AUTHOR
INTRODUCTION

E ven if you know nothing about digital marketing

analytics, digital marketing analytics knows plenty about
you. It’s a fundamental, inescapable, and permanent
cornerstone of modern business that affects the lives of
analytics professionals and consumers in equal measure.
This book is an attempt to provide the context, perspective,
and information needed to make analytics accessible to
people who understand its reach and relevance and want to
learn more.
The contents of this book reflect my decades of work in the
digital analytics space, with most of that time spent leading
large analytics teams at a major advertising agency and a
global technology company. I’ve also had the privilege of
teaching analytics at the graduate level for nearly 10 years.
This textbook represents the synthesis of those two worlds:
the academic and the actionable, the history and the hands-
on, and the analysis and the application.
I hope readers find this book to be a valuable, enlightening,
and engaging review of the ever-evolving analytics field. I’m
profoundly grateful for the opportunity to share my
experience, knowledge, and insights with you.
Kevin Hartman, September 2020
PART 1: The Day the Geeks
Took Over
Lesson 1 The ART+SCIENCE Mind Lesson 2 The
Early History of Data Analytics Lesson 3 The
Contemporary History of Data Analytics Lesson 4
The Rise of Modern Data Analytics The ubiquity of data
analytics today isn’t just a product of the past half-century’s
transformative and revolutionary changes in commerce and
technology. Humanity has been developing, analyzing, and
using data for millennia. Understanding where digital
marketing analytics is now and where it will be in five, 10, or
50 years requires a holistic and historical view of our
relationship and interaction with data.
Part 1 looks at modern analysts and analytics in the context
of its distinct historical epochs, each one containing major
inflection points and laying a foundation for future
advancements in the ART + SCIENCE that is modern data
analytics.
 PART 1 | LESSON 1
The ART+SCIENCE Mind

Five things discussed in this lesson:

Data analytics affects every aspect of business

Opportunities for data analysts have never been

greater, and expectations of data analytics to
influence business decisions have never been higher

Analysts should focus on skill development in

three important roles: the “Data Strategist,” the
“Techie,” and the “Data Designer”

Analysts’ roles have evolved over time, moving

from focused areas of expertise to broad and
balanced skills

Successful analysts develop deep skills but, just

as importantly, develop a mindset that helps set
them up for success

Digital marketing analytics, in my humble opinion, is the

most vibrant and undiluted blend of art and science in
today’s business world. No other pursuit requires its
practitioners to balance such a depth of scientific
application with a more poignant need to communicate with
audiences who neither understand nor have patience for the
technical underpinnings of that science. The science of
digital marketing analytics presents as verbose and
academic. A purely artistic display of analytics output
suggests the same lack of substance as a charlatan’s
medicine show. This pursuit only works through a harmony
of the two: “ART+SCIENCE.”
Data analytics – the broad science of applying data analysis
to solve business problems – affects every aspect of
business. From finance to operations to human resources to
marketing (and everything in between), it’s understood “the
world’s most valuable resource is no longer oil, but data.” [1]
As a result, everyone in a business today becomes an
analytics consumer. Moreover, the stakes have never been
higher for analysts as the opportunities for data analysis –
and the expectations for their benefits – have reached
unprecedented levels.
Given this predicament for the analyst (whether that word is
in one’s title or not) a broad, yet balanced skill set is
required.

The Three Roles of the Analyst

Analysts should develop skills and expertise in three
functional areas. Think of these functional areas as hats that
successful analysts wear throughout any analysis project or
endeavor. Each area of expertise has always been an
important skill for successful data analysts, regardless of
their industries or titles. These skills are interdependent:
strength in one area will be throttled by weakness in
another. As a result, analysts should seek to balance their
abilities across the three roles.
The ‘Data Strategist’
The first role the analyst must play is that of the “Data
Strategist.” In this role, analysts must bridge the data and
marketing worlds and understand their firms’ business
operations, objectives, and environment. The successful
Data Strategist understands how data can support the firm
and can interpret data in the context of what they mean to
the firm’s performance. In this book, I will provide you with
the frameworks and approaches needed to be a successful
Data Strategist.
The ‘Techie’
The second role an analyst must play is that of the “Techie.”
Here, the analyst is the data owner and must know how to
collect and manage data in ways that ensure quality and
promote efficiency. The successful Techie has the technical
know-how to solve difficult data challenges and understands
a breadth of data tools. Most, if not all, of which will be
discussed in this book.
The ‘Data Designer’
The last role is what we’ll call “Data Designer.” The
successful Data Designer has strong creative skills and can
turn insights into innovative visuals that people remember.
This role requires the analyst to know how to express data
stories and use tools to visualize those stories. Like the
Techie’s data analysis tools, we will discuss a variety of
dataviz tools throughout this book.
The blend of these three functional areas is what I like to
call the “ART+SCIENCE Mind.” When the three skills are in
balance, analysts are perfectly positioned to complete any
analytics task and communicate their findings in interesting,
compelling ways.
We’ll discuss these skills in-depth in this book. First, let’s
look at how these three roles have been staffed over time.
The analyst’s role has undergone a significant and
interesting evolution.

The Evolution of the Analyst in the Firm

In an earlier time, advertisers – particularly, large
advertisers – would build teams of data analysts centered
around each of these functions. The ability of the three
distinct functional teams to work closely with one another
was a key determinant of a company’s success. As a result,
advertisers would invest significant time and energy in
ensuring cross-collaboration through specially designed
information-sharing sessions that would invite analysts from
various functions to experience the rough-and-tumble
exchange of ideas and the unavoidable pain of
collaboration. In other cases, companies created cross-
functional “scrum” teams that would meet frequently to
discuss progress against closely watched project plans and
develop ways to mitigate the risks of handing work from one
group to the next.
Progressive advertisers increasingly expect data analysts to
have demonstrable skills in all of the functional areas. This
expectation undoubtedly had something to do with the
inefficiencies that advertisers with distinct functional teams
likely experienced. Required handoffs between teams in the
old model slowed progress and introduced risk. But another
important reason for the growth in expectations of data
analysts is the natural evolution of a role that has become
increasingly critical for advertisers to compete in today’s
digital economy.
The proliferation of analytics initiatives across industries has
led to more options for good analytical talent than ever
before. Analysts’ opportunities have grown in terms of
geography, expanding beyond the expected hot spots of
major metropolitan areas, to include nearly every area
where businesses operate today. Opportunities for the
analyst have also become more technically diverse. As
industries mature and explore new applications for data, the
need for analysts is emerging in traditional businesses as
well as in so-called “mission-driven roles” that have a
societal or environmental benefit, such as those that focus
on the application of data to advance health outcomes,
energy efficiency, or nonprofit work.
Several trends have expanded analysts’ opportunities.
Data analytics tools are becoming more accessible
The advancement of open-source and vendor-
developed tools has improved accessibility for
challenges that were previously reserved for complex
and rigorous computer science solutions. While the
market’s most sophisticated tools are far from being
“point and click,” they’re evolving quickly and
becoming easier for a wider audience of analysts
(and even nonanalysts) to use.
Analytics know-how is increasingly becoming a
requirement for all roles With many analytics teams
increasing in size and scope, businesses increasingly
face the challenge of connecting technical teams to
other business unit stakeholders. As a result,
companies expect nonanalyst business leaders to
have a greater understanding of data analytics. In
this way, they use their expertise to be
representatives for the quantitative team and free up
data scientists to do what they do best: advanced
technical analysis.
Educational backgrounds are shifting In addition to
the evolution of analytics tools, a fast-growing
number of educational choices are making analytics
training more accessible. These choices include video
tutorials, online courses, “master class” workshops
led by subject matter experts, as well as the
expansion of traditional educational programs to
include analytics-focused curriculums and degrees.
Degrees, certificates, and work portfolios resulting
from such educational alternatives are supplementing
or even replacing the bachelor’s and master’s
degrees analysts hold today.
A perusal of job postings for “analyst” reveals a wide variety
of titles, including “Market Research Analyst,” “Marketing
Analytics Analyst,” “Marketing Strategy Analyst,” “Analytics
Analyst,” “Social Media Analyst,” and many more. While the
name by which the analyst is called has grown more
distinct, responsibilities remain centered around gathering,
analyzing, reporting, and advancing data.
Still, segmenting “Data Scientist” roles from other analytics
professionals makes sense today because of skill set
differences that affect salaries. While there’s no universal
definition of a “Data Analyst” or Data Scientist, the
generally accepted distinction of the two roles is that the
Data Analyst focuses on the “current state” through the
analysis of structured data while the Data Scientist works
primarily with unstructured or streaming data to better
understand the future. While the market values the roles
differently today, both are lucrative professions.
In its 2020 annual survey on salary ranges for the analytics
market, Burtch Works Executive Recruiting collected
compensation and demographic data of 1,742 Data Analysts
(a role Burtch Works refers to as “Predictive Analytics
Professionals”) and 503 data scientists. [2] For Data Analysts
in 2019, Burtch Works reported the median base salary of
individual contributors ranged from $80,000 to $135,000.
Data Scientist individual contributors earned more, with
median base salaries ranging from $95,500 to $165,000.
Managers in both roles earned up to $250,000. [3] These
figures represented increases for Data Analysts and Data
Scientists when compared to previous years.
The divide between traditional analysts and more statistics-
forward Data Scientists is blurring by the day. This is
primarily because requirements for Techie skills have
become more common. Such skills may have once
distinguished subsets of analytics professionals but are less
differentiating today. While Data Scientists have long been
regarded as a specialized subset of the analytics profession,
I would expect fewer companies to employ distinctly scoped
teams.
Over the course of this book, we’ll explore many other
industry trends that are driving the evolution of the analyst
in the firm. Before we do, however, let’s look at the
characteristics of successful analysts.
Characteristics of the Successful Analyst
Being a successful analyst requires more than mastering the
three roles discussed above. It demands a mindset. Indeed,
the successful analyst demonstrates a mindset
characterized by three important traits: curiosity, optimism,
and a healthy dose of boldness.
Curiosity
Data analytics operates on the leading edge of marketing,
advertising, and technology, and that’s exactly what the
analyst will love about it. A new search is always on the
horizon: a new framework to develop, a new challenge to
quantify, a new learning to discover. The analyst should
embrace these challenges with the vigor, zeal, and
inquisitive nature of a “Data Frontiersperson.” Curiosity
compels the successful analyst forward.
Optimism
The challenges data analytics pose aren’t for the weak of
mind or spirit. Analysts should expect to run into more brick
walls than open doors. The arduous process of analyzing
data means that failure is routine. I’ve seen analysts who
abandoned their work when it became too challenging. More
often than not, this was because they were limited in
imagination, resilience, and zeal. Those who succeed do so
because their tenacity and passion outweigh their
frustrations. Perhaps, most importantly, analysts are
successful because their optimism lets them believe
solutions are there to be found.
Boldness
No one is more immersed in the use of marketing numbers
than data analytics practitioners. Those who succeed push
the boundaries of what was thought to be data’s limit.
They’re audaciously daring in the questions they ask and
the answers they seek. They earn the right to be bold. In my
experience, this boldness has driven analysts to focus on
the details while they strive to provide the highest-quality
work. That attention to detail is what separates outstanding
analysis from merely good analysis.
In Summary: The ART+SCIENCE Mind
Data analytics affects every aspect of business – from
operations to investment planning to marketing. While
opportunities for data analysts have never been greater,
they come with expectations that have never been higher.
To be successful in today’s data analytics environment,
analysts should focus on skill development in three
important roles: the Data Strategist, the Techie, and the
Data Designer. These roles have evolved, moving from
focused areas of expertise to broad and balanced skills.
Successful analysts go beyond developing deep skills in
these areas to develop a mindset of curiosity, optimism, and
boldness that helps set them up for success.
In the next lesson, we’ll place the role of the analyst and the
emergence of today’s contemporary practice in the context
of the long and robust history of data analytics.
 PART 1 | LESSON 2
The Early History of Data
Analytics

Five things discussed in this lesson:

Mankind’s experience with data analytics is long

and robust, evolving from rudimentary mapmaking to
the flourishing ART+SCIENCE practice of today

Data analytics has played a major role in

statistics, medicine, politics, and many other fields

Data analytics’ early history can be described

through eight distinct epochs, each boasting
significant events that trace the growth of data
analytics as a practice

Each epoch features a distinct inflection point

that moved the practice forward and ushered in the
beginning of the next epoch

Data analytics’ early history ended when the

creation of digital data introduced the rise of
contemporary data analytics

In the beginning, there was data.

One can look back through 7,000 years of human history
and see examples of mankind expressing ideas through
data. The practice began in the humblest of forms – simple
maps used to document and describe the world around
those early humans – and has evolved to become the
flourishing modern practice of data analytics that we know
today, fanning into statistics, medicine, politics, and other
fields. The discipline evolved over time, adding capabilities,
confronting critical issues and, in the end, emerging as a
balanced blend of art and science.
This long rich history of humans using data to communicate
can be traced back to nearly the beginning of time and can
best be categorized around eight distinct epochs. Each
period had an inflection point that moved us from one point
in this history to another period.
Figure 1.1: The early history of data analytics in visual form

Epoch I: ‘Early Maps and Diagrams’

Epoch I, the “Early Maps and Diagrams” phase, is the period
before 1600. [4] The earliest seeds of visualization arose in
geometric diagrams, tables of the positions of stars and
other celestial bodies, and maps to aid in navigation and
exploration. People used data in a functional way: creating
visual objects to make sense of the world around them.
Displaying information visually allowed people to share their
knowledge, record their environment, and measure critical
matters in their everyday lives.
During this period, the ability to communicate visually was
critical to human development, social interaction, and
humans’ understanding of the world.

Epoch II: ‘Measurement & Theory’

Epoch II spans the 1600s in a period called “Measurement &
Theory.” [5] This century saw significant growth in theory and
the rise of analytic geometry, theories of measurement and
estimation, probability, and the beginning of demographic
statistics and “political arithmetic.” The capability of
measurement expanded as visualizations began to explore
more of the world. An excellent example of this evolved use
of data is Christoph Scheiner’s visual from 1626 that
charted changes in sunspots. This achievement illustrates
humans developing tools to record data – in this case, spots
on the sun collected through a telescope – and then
translating that into a visual form.

Epoch III: ‘New Graphical Forms’

Epoch III was a century-long period in the 1700s. The period
witnessed the germination of the seeds of visualization as
mapmakers began to show more than only geographical
position. As a result, people introduced many graphic forms.
In the period of “New Graphical Forms,” [6] artists began to
explore new ways of visualizing information, expanding the
boundaries of creative forms of communication. This period
is characterized by the development of preattentive
attributes – early examples of shading, size, contrast, and
color – that expressed developing insights and
demonstrated a growing aptitude in the science of data
analysis.

Epoch IV: ‘The Beginnings of Modern

Graphics’
Epoch IV, 1800–1849, has been called “The Beginnings of
Modern Graphics.” [7] With the fertilization of design and
technique innovations, the first half of the nineteenth
century saw growth in statistical graphics and thematic
mapping at a rate not matched until modern times. Real
data visualization took root in this period. William Playfair
introduced the pie chart, bar chart and line chart – all still
very important to data visualization.

Epoch V: ‘The Golden Age of Data

Analytics’
In the following epoch, Epoch V, we move into the “Golden
Age” of data analytics. [8] The late 1800s saw dataviz of
unparalleled beauty and many graphical innovations. Data
artists used visualization to communicate their stories.
Great examples from this period include John Snow's
epidemiological study of cholera in London occurring around
the Broad Street well. This visualization is credited for
revolutionizing human understanding of cholera and its
treatment methods. Florence Nightingale proved the need
for improved sanitary conditions in the British army using
visualization, and Charles Booth visualized poverty in
London using color on a map to demonstrate how the poor
were isolated.
In this epoch, we’re also blessed with possibly the greatest
data visualization of all time: Charles Joseph Minard’s
representation of Napoleon’s march on Moscow. So much
information is packed into this visual, including Napoleon’s
“triumphant” return to Paris. It includes distance and
temperature, and the number of soldiers lost. This visual
communicates a complex story while being simple and
beautiful.
Figure 1.2: Minard’s masterful visual presentation of
Napoleon’s march

During this period, data artists begin to get a little “over

their skis.” New emerging graphical forms inspired people
who made visualizations to create images for visual sake
alone. Several examples from the time are complicated and
overwrought. Graphics printed on top of graphics obscure
the artists’ intended meaning.

Epoch VI: ‘The Modern Dark Ages’ of Data

Analytics
After the careless approach to data visualization triggered
an inevitable backlash, we move into Epoch VI (1900–1949).
If the late 1800s were the “Golden Age” of visualization, the
early 1900s were the “Modern Dark Ages,” [9] as the
previous period’s enthusiasm was supplanted by the rise of
quantification and formality. Scholars, who suspiciously
watched the emergence of data analytics from their
academic quarters, rose in revolt. Their primary concerns
were the need for more accuracy, more controls, and a
standardized language. This movement limited creativity for
the sake of accuracy and introduced what we know today as
the scourge that is “clip art.” Individual icons represented
certain data that were standardized, and frankly, threatened
to knock the path of data visualization off its creative tracks.
We see redeeming introductions during this period,
however. One is the famous Henry Beck map of the London
Underground, which describes a complex system simply and
beautifully. More importantly, Beck’s map doesn’t show the
positions of the Underground stations perfectly. It stood as a
testament to the idea that images can serve a purpose
without exacting accuracy. The popularity of this graphical
representation offered a counterargument to the strict and
rigid practices of data visualization from the societies that
promoted exactness and standardization.
Figure 1.3: Beck’s perfectly imperfect map of the London
Underground

This period also introduced the Mark I, a computer

developed at Harvard University. As the first computer, the
Mark I was the size of a living room, but less capable than
today’s average smartphone. Still, this technological
advancement made it possible to process information in
ways that weren’t possible before. Analysts could visualize
larger volumes of data, generate graphics with greater
efficiency, and discover patterns in data no one could find
before the introduction of the computer.
The introduction of Beck’s map and the Mark I are chief
among the factors that moved the discipline of data
visualization from the Modern Dark Ages into a period of
rebirth.

Epoch VII: ‘The Rebirth of Data Analytics’

Data analysts began to embrace new graphical forms and
capabilities in Epoch VII, “The Rebirth of Data Analytics”
(1950–1974). [10] Still under the influence of the formal and
numerical zeitgeist from the mid-1930s on, data
visualization began to rise from dormancy in the mid-1960s,
spurred largely by technology and data democratization.
Computers played an enormous role in this transformation.
Analysts could bring in more data with greater
sophistication and add elements that weren’t previously
available.

Epoch VIII: ‘High-Definition Data Analytics’

(1975-1994)
We move into the final epoch in the early history of data
analytics, Epoch VIII (1975–1994). The development of
highly interactive computer systems and new methods for
visualizing high-dimensional data ushered in a dataviz
renaissance period that was limited only by human
imagination. We call this period “High-Definition Data
Analytics,” [11] in which computers or applications created
most visuals. They processed vast amounts of information
and leveraged previous data visualization knowledge to
create effective, efficient, and tremendously powerful
visuals. The rapid advancement of data design signaled the
end of the early history of data analytics and a shift to a
contemporary time.

In Summary: The Early History of Data

Analytics
Understanding the evolution of analytics through the eight
epochs described in this lesson is important because they
trace a rich and long history of humans communicating with
data. The journey also shows how we’ve evolved from
simple hand-painted maps to where we are today – using
computer applications that have tremendous power to
generate visuals. Throughout this fascinating history, we’ve
confronted topics such as data scarcity, politics, data
misuse, and other issues to arrive at what is today – in my
estimation – the purest, and most beautiful blend of art and
science that we’ll find. This early history of data analytics
ended when the creation of digital data introduced the rise
of contemporary data analytics, which we’ll discuss in the
next lesson.
 PART 1 | LESSON 3
The Contemporary History of
Data Analytics

Five things discussed in this lesson:

Contemporary history of data analytics saw the

fully formed ideal of the role data and analysis could
play in marketing and advertising

Contemporary history of data analytics is a 20-

year period that began with the introduction of the
banner ad in 1994

Contemporary history of data analytics can be

described as four epochs, each boasting significant
events that trace the growth of data analytics as a
practice

As with the early history of data analytics, the

epochs of its contemporary history feature inflection
points that advance the practice and signal the
beginning of a new epoch

Several events demonstrate the maturation of

digital platforms in 2014, bringing the contemporary
history of data analytics to a close
As we turn our attention to the contemporary history of data
analytics, I’m reminded of a quote from Michael Fassnacht, a
pioneer in the introduction of data analysis in marketing and
my first boss when I entered the world of advertising. In
2006, Fassnacht said, “Over the next few years, geeks in
marketing will become one of the most disruptive forces in a
discipline that traditionally was driven by big creative
personalities.”
Fassnacht was talking about data’s effect on creative and
marketing processes. He was correct in this assessment, but
perhaps a bit too prescient. Indeed, few advertisers had
fully integrated data analytics in their marketing efforts in
the years immediately following 2006. Furthermore, many
advertisers still rely on traditional, comfortable methods to
guide their creative and marketing endeavors.
But Fassnacht knew the opportunity to better understand
consumers in a way that reduced risk for advertisers and
increased accountability for their agency partners had been
gaining steam for a while. Twelve years before he issued his
challenge to the marketing and advertising world, a seminal
event redefined the consumer/brand dynamic forever. That
event was the introduction of the digital banner ad on
October 27, 1994.
Figure 1.4: The world’s first banner ad

AT&T purchased the first banner ad and placed it on

hotwire.com. It earned a click-through rate of 44 percent, [12]
an unheard-of result, especially given that today’s display
ads entice 0.05 percent of consumers to click them. [13] It’s
almost certain the novelty of the experience sparked
curiosity in consumers who wondered what would happen if
they followed the ad’s instructions to “click your mouse
right HERE.”
This humble ad, presented in a haphazard-looking array of
psychedelic colors against a plain black background,
changed the way brands use the web. The ability to
influence behavior and, more importantly, collect
information about behavior, disrupted the economic and
financial models of nearly every company from then on.
The moment also brought a new trajectory for data
analytics, ushering in the next phase of this science –
something I call the “contemporary history of data
analytics.”
Figure 1.5: The contemporary history of data analytics

Similar to the early history of analytics, this period can be

seen as a collection of distinct periods, each with its own
unique characteristics and circumstances. Inflection points
of technological and cultural significance transition us from
one period to the next. In this way, the contemporary
history of analytics is comprised of four epochs.

Epoch I: ‘When Anything Was Possible’

The first epoch is something I will call “When Anything Was
Possible,” and encompassed the time of that first AT&T
banner ad in 1994 to 1999. Some great things happened in
this period. Google opened its first offices. Napster was
invented, which would change the face of music forever.
Other companies founded in this time are less heralded
today, including Flooz.com, Kozmo.com, and Pets.com, but
were just as important to the growth of digital commerce.
At this point, nearly anyone who entered an investor’s office
with a dot-com plan would be taken seriously. No idea was a
bad idea that involved the use of digital platforms to
transform (or create) businesses. The web and how it would
sort itself out wasn’t readily understood, but it was clear
that it had the power to connect people and make
information more accessible than any previous invention. As
a result, investors placed bets on a wide variety of early
internet ventures. Any company left standing on the
sidelines of the “Internet Revolution” felt palpable anxiety
emanating from its board of directors to its junior
employees.
It was the inchoate period of digital “FOMO” (Fear of Missing
Out), without mobile devices and apps. As a young
management consultant who worked in a well-respected
and established firm that had similarly well-respected and
established clients, it seemed as if every day, I watched a
bright, ambitious colleague announce they were leaving the
consultancy to join a startup poised to revolutionize the
[insert industry name here]. In private conversations, my
friends would talk excitedly about being part of the dot-com
experience and, should the stock options pan out as they
hoped, how they would probably be able to retire in a few
years.

Epoch II: ‘The Bubble and the Burst’

That exuberance came to an end with the advent of the
second epoch, “The Bubble and the Burst” (2000–2002).
The NASDAQ stock index, where the vast majority of public
dot-com companies were traded, rose fivefold between
1995 and 2000. But the index tumbled from a peak of
$5,048.62 on March 10, 2000, to $1,139.90 on October 4,
2002. [14] That crash destroyed $2.7 trillion in market value.
More than half of all dot-com businesses founded after 1996
failed to see 2004. [15] Many signs of the unsustainability
that led to the NASDAQ crash were apparent, yet apparently
ignored. Kozmo, one victim of the internet bubble and burst,
demonstrates the limitations of investors’ strategy to forego
profitability in favor of innovation.
A supply chain management company that invested heavily
in innovative technology, Kozmo delivered small consumer
goods its bicycle couriers could accommodate. Kozmo
focused on densely populated urban areas and had a unique
promise: It would deliver any item from its offered product
categories in less than an hour with no recipient delivery
fee. While traditional companies required consumers to
invest their own time and money by shopping at retail
locations or waiting days for their items from costly delivery,
Kozmo promised to “extend (the supply chain) all the way to
the customer’s front door.” [16] The company warehoused
nonperishable items in its delivery areas (with a central
distribution center located in Memphis, Tennessee) and
collected perishable items, including Starbucks coffee and
Gerald Stevens flowers, directly from retailers.
Kozmo’s customer-centric focus meant it sacrificed
efficiency. It carried an awkwardly broad set of 24 product
categories (in line with its mantra of “delivering what the
consumer wants”), including DVDs, books, food, beer,
electronics, and other items that introduced tremendous
complexity to its inventory management and distribution
systems. Kozmo’s need to attract consumers to its new
shopping concept kept a tight lid on prices for consumers,
yet its lack of scale prevented Kozmo from earning
discounts or advantageous pricing arrangements from
suppliers. The mix of low prices and high fulfillment costs
meant Kozmo, like other internet-based delivery services
operating at the time (a list that includes now-defunct
Urbanfetch, Streamline, Webvan, and ShopLink), lost money
on almost every delivery.
Kozmo’s early performance took firm hold of investors’
imagination. At the end of its first year of operations, the
company grew from six markets to 11. Despite posting an
operating loss of $26.3 million on revenues of just $3.5
million, Kozmo continued to attract investment to fuel its
growth. After 18 months in business, the number of Kozmo
customers jumped from 75,000 to 300,000. A $60 million
cash infusion from Amazon in March 2000 provided further
capital and by July 2000, Kozmo had more than 400,000
customers, 2,600 employees, revenues that were doubling
month-over-month, and plans for an initial public offering
(IPO) on the NASDAQ. Just three years after opening a single
delivery shop in New York City, Kozmo featured all the top-
line trajectory the internet age promised.
But for all of its tremendous growth, Kozmo’s underlying
financial viability remained suspect. Its ardent focus on
meeting customer needs led to poor business decisions,
such as its Valentine’s Day offer of jewelry delivery in less
than one hour. Without a clear sense of demand (outside of
anecdotal consumer interest), Kozmo secured inventory and
paid significantly to augment security controls yet managed
to sell only 2 percent of the products it purchased.
Operating mistakes like this led Kozmo to be unprofitable in
all but four of its markets, despite a broadening distribution
network, a homegrown inventory management system that
provided innovative solutions to the company’s unique
delivery challenges, and on-time order fulfillment rates that
hovered around 99 percent. While still attracting
investment, Kozmo calculated an average order size of $5
and an average delivery cost of $7.50. [17]
The NASDAQ crash forced Kozmo to postpone its planned
IPO and re-examine its operating model. As investors’
money dried up, Kozmo made moves to point itself toward
profitability – such as enacting order minimums – but these
moves were too little, too late. The company continued to
burn through cash and layoffs began in mid-2000. Even with
an (unbelievable) $30 million new round of funding raised in
January 2001, Kozmo ceased operations just a few months
later in April 2001.
The Kozmo experience provided a cautionary tale for
investors about the importance of any company’s
underlying business fundamentals. Perhaps more
importantly, it spurred new ventures to re-examine the use
of data and analytics in assessing consumer needs and the
value of collecting consumer data.

Epoch III: ‘The Seeds of Prosperity’

From these ruins emerged new internet-based ventures,
stronger and better-positioned than the companies that fell
victim to the dot-com crash. This brings us to the third
epoch, something I call “The Seeds of Prosperity” (2003–
2010). In this period, a number of new ventures took root,
harnessing the internet’s massive reach. They didn’t sell
products in the traditional sense. Instead, they met
consumer needs that can’t be bought or sold.
These companies include LinkedIn (professional
connections), Twitter (the ability to be heard), Groupon
(collective savings), and, of course, Facebook (community).
They were founded as the web began to recover from the
NASDAQ crash, and each one fulfilled important consumer
needs through the internet.
Epoch IV: ‘The Age of Unicorns’
This brings us to the fourth epoch, “The Age of Unicorns”
(2011–2014). The label, “unicorn,” has been widely
attributed to Cowboy Ventures founder Aileen Lee, who
coined the term to describe pre-IPO tech startups with a
market valuation of $1 billion or more. When she introduced
the concept of the unicorn in 2013, 39 U.S. startups had
topped the $1 billion valuation mark in the previous decade.
[18]

From there, things moved quickly to the point where the

existence of these once-unique startups became the norm.
One year later in 2014, 20 years after the invention of the
banner ad and the year that marks the close of this fourth
epoch, the number of unicorns had doubled to 80, according
to Fortune . [19] The implication of this tremendous growth is
that the tech startup came of age. I would argue that an
underlying explanation lies in the sophisticated use of data
exhibited by each company on the Fortune list.

The Significance of 2014

Between 1994 and 2014, internet use grew dramatically.
When the first banner ad appeared in 1994, fewer than 5
percent of North Americans used the internet. By 2014, that
percentage had begun to level off around 75 percent. [20]
According to Nielsen Online, ITU, PEW Research, and
internet World Stats, the internet’s usage in North America
was a few points below 90 percent in 2019.
Expanded internet access gave rise to new, powerful uses of
digital platforms, representing an unprecedented and
marked shift in the way consumers interacted online.
Several examples signaled that the internet had become
fully mature in 2014.
Ellen’s ‘Oscars Selfie’ Tweet In March 2014, Ellen
DeGeneres’ star-studded “Oscars Selfie,” snapped at
the Academy Awards show, was shared on Twitter
while the show was still filming.
That tweet set a record with 3.4 million retweets. [21] It’s
been estimated that Ellen’s selfie was worth as much as $1
billion in earned media for Samsung, an Oscar sponsor that
year. [22] Such powerful social sharing had not been seen
before.
ALS Ice Bucket Challenge Viral Sensation Four months
later, the amyotrophic lateral sclerosis (ALS) ice
bucket challenge raised more than $100 million as a
viral sensation arose over people videoing
themselves dumping buckets of cold water on their
heads after challenging others to do the same (and
contributing donations to ALS research). [23] Primarily
shared on Facebook, Twitter, and YouTube, these
videos raced through social circles of all sizes.
One of the more popular ALS challenge videos involved
Microsoft founder Bill Gates successfully completing a
challenge from Mark Zuckerberg. That video racked up
nearly 21 million views, demonstrating the connective
power of the internet as a means for raising awareness (and
funding) for social causes. [24]
#Ferguson Digital Activism In August 2014, amid the
unrest that erupted around the police shooting of
unarmed teen Michael Brown in Ferguson, Missouri,
social media became an engine of social activism.
The hashtag, “#ferguson,” was tweeted 21.6 million
times during five days of protests and
demonstrations. [25]
Hollaback! Video
In November 2014, the anti-harassment organization,
Hollaback!, posted a video on YouTube, 10 Hours of Walking
in NYC as a Woman . In clear sound and images, the video
demonstrated the harassment a woman faces walking the
streets of New York and earned more than 6 million views in
24 hours. [26]

In Summary: The Contemporary History of

Data Analytics
The contemporary history of data analytics witnessed the
fully formed ideal of the role data and analysis could play in
marketing and advertising. This 20-year period that began
with the introduction of the banner ad in 1994, created a
novel and significant change in the way analysts could gain
insight into consumer behavior and measure the effects of
their marketing efforts. The contemporary history of data
analytics can be described as four epochs, each boasting
significant events that trace the growth of data analytics as
a practice. Like the early history of data analytics, each
epoch of its contemporary history features inflection points
that advance the practice and signal the beginning of a new
epoch. Finally, we learned that several events demonstrate
the maturation of digital platforms in 2014, bringing the
contemporary history of data analytics to a close.
In the next lesson, we’ll discuss important frameworks
analysts developed to help them make sense of their
consumer relationships and the role data analytics played in
that understanding.
 PART 1 | LESSON 4
The Rise of Modern Data
Analytics

Five things discussed in this lesson:

Data has become the most valuable asset for

those who make decisions or attempt to influence
them

In 2005, Procter & Gamble (P&G) introduced the

concept of a three-step model that detailed the role
of the brand in a consumer’s decision journey

Since P&G introduced its three-step model, the

amount of data consumers and brands have
generated has grown significantly

To adapt P&G’s framework to the changing data

landscape, Google introduced the concept of the
“Zero Moment of Truth”

Several influences drove the explosive growth of

data and new customer touch points, creating new
marketing opportunities that led to the growing
demand for analytics in today’s marketing and
advertising environment
Humankind’s experience with data analytics has been long
and rich. Through periods of cultural, political, and
technological changes, the importance of collecting,
analyzing, and telling stories from data has never wavered.
To the contrary, the critical role of data in nearly every
aspect of marketing has grown exponentially. Data has
become the most valuable asset for anyone who makes – or
attempts to influence – a decision.
From the use of personal computers in the ‘80s, to the
web’s arrival in the ‘90s, to the advent of smartphones in
the 2000s, brands’ interactions with consumers has
changed radically. To make sense of the changing dynamic
in their relationships with consumers, brands frequently turn
to frameworks, including the three-step marketing model
that P&G pioneered.
P&G’s Three-Step Model of Marketing
In 2005, after the internet bubble burst but before Facebook
had fully taken root, P&G introduced a three-step model of
marketing. Although simple in its design, the P&G
framework was powerful in its relevance and insight into
consumer behavior. The model sought to simplify the way
consumers make purchase decisions and explain critical
points along that journey.
‘Trigger’
The model begins with the idea that every consumer’s
purchase path begins with a stimulus. For example, an ad
could spark interest in learning more about a product the
individual decides to purchase. Perhaps the consumer runs
out of a required household staple, learns how wonderful a
product is from a friend, and decides to try it. These stimuli
are examples of what P&G calls a “trigger,” or the first step
toward the purchase of a product.
‘First Moment of Truth’ (FMOT) The next major step in
P&G’s model is called the “First Moment of Truth”
(FMOT). FMOT occurs when consumers are at the
shelf, ready to purchase, and have a number of
options for the products they want. Various brands
boast promises through attention-grabbing packaging
and compelling price discounts. Consumers may even
receive samples that entice them to “try before they
buy.” With FMOT, consumers must choose brands
based on their knowledge of the various offerings.
Importantly, this was the point P&G identified as the
first battleground for brands because they have to
fight for attention.
‘Second Moment of Truth’ (SMOT) The final step in
the P&G model is called the “Second Moment Of
Truth” (SMOT) and is realized after consumers buy
products and bring them home. They appraise
products against the expectations they developed
during their evaluation process. When products live
up to SMOT expectations, brands can win customers
for life. If products falter, however, the results can be
catastrophic for brands. According to P&G, fulfilling
consumers’ expectations was the second critical
moment that brands must win.
P&G’s purchase path moments are as important to brands
today as they were in 2005. Brands must stand out from
their competitors and persuade consumers to choose them
instead of rival products.
The need to set and fulfill consumers’ expectations hasn’t
diminished and is arguably more important today than ever
before. The world, however, isn’t quite the same as it was
then.

The Introduction of the ‘Zero Moment of

Truth’ (ZMOT)
As access to information grew with the increasing
availability of mobile devices and brands’ maturing use of
the internet, the way consumers shopped underwent a
dramatic change. Online information quickly became a vital
source of intelligence for consumers as they evaluated
products.
For that reason, in 2011, Google introduced the “Zero
Moment of Truth” (ZMOT). ZMOT is the time between a
stimulus, still obviously relevant, and the First Moment of
Truth. Still true to the P&G framework, a brand doesn’t win
unless a consumer chooses it. Typically, consumers have
many brands from which to choose. With ZMOT, Google
captured the concept that the shopping dynamic had
changed considerably. Now, consumers were going to the
shelf while armed with significantly more information about
the products and the brands they could purchase.
Such information could be earned from product reviews
found on any number of internet sites. It could be collected
from a friend’s experience with the product and posted to
Facebook. It could be found in a tweet from a celebrity
extolling love for products and brands. The information
could also come from some variety of the estimated 4,000
to 10,000 ads most Americans are exposed to each day. [27]
These sources often work collectively to inform and educate
consumers during ZMOT, and are joined by many others,
including television, friends and family, online destinations,
print media, and brick-and-mortar stores.
ZMOT isn’t a neat picture. It’s messy, with many things
going on and is, in no way, linear. Consumers can hop
between sources of information and back again as quickly
as their browsers are able to load internet pages.
Furthermore, consumers move between online and offline
worlds fluidly – much like how they live. All of these
channels are part of the pre-shopping experience in-home
and in-store, ultimately leading to P&G’s First Moment of
Truth.
For brands, ZMOT represents a critical opportunity to
influence purchase behavior. Each ZMOT touch point
represents an opportunity to interact with consumers and
learn their reactions to specific brands and products. These
new touch points provide opportunities for brands to
influence consumers which requires brands to understand
consumer needs – insights that could be gleaned through
the collection of profile data, and online consumer
behaviors. As a result, ZMOT offers advertisers many
opportunities to influence consideration, and learn whether
their products satisfy consumers’ needs. For marketing
analysts, ZMOT represents the most important opportunity
to understand consumers. It’s where consumers reveal
significant information about themselves.
Delivering influence required brands to understand
consumer needs – insights that could be gleaned through
the collection of profile data, and online consumer
behaviors. As a result, ZMOT offers advertisers many
opportunities to influence consideration, and learn whether
their products satisfy consumers’ needs. For marketing
analysts, ZMOT represents the most important opportunity
to understand consumers. It’s where consumers reveal
significant information about themselves.
Optimizing the effect brands had on consumer choice
required brands to test messages and track results. In this
way, the growth in advertiser opportunities was among
many factors that drove increased demand for analytics.
Technological Influences
One driving force behind the growth of data analytics has
been the ceaseless advance of technology. As technology
has evolved to meet the needs of consumers and
businesses, new opportunities to create and collect data
have arisen. Two of the more important technological
influences are consumers’ swift adoption of mobile devices
and the growth of marketing technology to help businesses
create value from the data they collect.
Swift Adoption of Mobile Devices The swift adoption
of mobile devices has dramatically increased the
volume of consumer data available to analysts.
Devices facilitate the creation of consumer data in
two ways. First, mobile devices provide access to
online sites and apps that create data (e.g., mobile
Facebook and mobile search), making it easier for
consumers to generate data. Second, mobile devices
produce passive data (e.g., location data) in great
volumes.
Mobile device growth was rapid, as illustrated in
photographs comparing the installations of Pope Benedict in
2005 and Pope Francis in 2013, seen in Figure 1.6. The
ability of device users to instantly share thoughts, feelings,
and experiences (as well as their locations) improved
analysts’ ability to understand consumers and contributed
directly to the rise of analytics.
Figure 1.6: Comparison of Pope Benedict (2005) and Pope
Francis (2013) installations [28]
Growth of Marketing Technology Space The rapid
growth in companies in the marketing technology
(martech) space increased the value of analytics
while driving the demand for analysts. This
phenomenon was illustrated through snapshots of the
martech industry from ChiefMartec’s Scott Brinker.
In August 2011, Brinker charted the logos of providers
operating in the nascent business of “marketing
technology” as a way to sort out the market. There were
150.
Figure 1.7: Brinker’s martech landscape in 2011 [29]
The infographic was so well-received, Brinker decided to
visit the martech space a year later and produce a second,
updated industry snapshot. When Brinker recorded his
second analysis in September 2012, the number of
marketing technology firms had grown to 350. When he
updated his analysis in 2014, the number of firms had
climbed to 1,000. A year later, in 2015, the number
increased to 2,000.
Brinker’s March 2016 analysis revealed an astonishing 3,500
firms in the space – almost twice the number from 14
months prior. In 2017, the number of companies in the
marketing technology space reached 5,000, increasing more
than 40 percent from 2016’s total. 2018’s graphic charts
6,829 marketing technology solutions. While that represents
“only” 27 percent growth over the previous year’s
landscape, the increase belies the absolute scale of this
space.
At the writing of this book in 2019, the number of
companies in the marketing technology space surpassed
the 7,000 mark to 7,040. Brinker refers to his most recent
infographic as the “MarTech 5000” because the nickname,
in Brinker’s words, “kinda rhymes with the Fortune 500 and
the Inc 5000.” [30] The extraordinary growth in the number of
firms that offer services to analysts has opened their minds
to the (nearly) unlimited power of data, while introducing
extensive confusion as they attempt to navigate these new,
roiling opportunities.
Figure 1.8: Brinker’s martech landscape in 2019 [31]

The number of firms in the space has led to innovation and

technological advancement, thereby increasing the value
analytics offers the business world.
Market Influences
In addition to the force of technology growth, market
influences also had a profound effect on the demand for
data analytics. By reducing uncertainty and creating
accountability, today’s digital-based business models and
data analytics have played an important role in industry.
Investors’ Demand Post-NASDAQ Crash In the early
go-go days of the mid- to late-1990s, flashy business
ideas light on analytical rigor still received funding
from investors who hoped to cash in on the internet
boom. But once the internet bubble burst in 2000,
businesses that didn’t demonstrate sound financial
structure surrendered their value. Companies that
emerged from the wreckage of the NASDAQ crash
and attracted investment in the post-bubble
environment could demonstrate sound business
models backed by thorough financial analysis. This
heightened examination of the numbers side of
internet business increased the demand for analytics.
Rise Of FAMGA
Following the NASDAQ crash in 2000, FAMGA – Facebook,
Apple, Microsoft, Google, and Amazon – understood the
value of digital data better than others. The data FAMGA
collected, analyzed, and aggregated have dramatically
affected the way consumers shop for products and how
businesses interact with consumers.
Since 2015, these companies have been one of the most
important driving market forces behind the U.S. economy.
How investors evaluate FAMGA is tied to the value of their
data (e.g., Facebook’s key performance indicator [Daily
Average Users]/[Monthly Average Users]). Maximizing the
value of FAMGA’s data has fueled the rise of analytics.

Placing Today’s Data Creation in Context

An extraordinary amount of digital data is generated every
day. It can be easy, however, to lose context for how much
data is produced. A simple infographic like the one
displayed below can help us regain that context.
Figure 1.9: Comparison of daily data-creating human activity
(all figures in billions) [32]
Represented in this graphic are the volumes of several
consumer activities that generate important data. Each box
represents a single activity, such as liking images on
Instagram or tweeting on Twitter, and is scaled to its relative
size. The infographic makes it possible to introduce other
data-producing behaviors that add context to the digital
activities represented in the visual.
Large, popular activities, including watching YouTube videos
(6.2 billion each day), liking things on Facebook (5.8 billion),
and searching Google (5.6 billion), stand out as vast
amounts of data are created daily. [33] Those numbers
appear even larger in the context of the brown box in the
upper right, which represents the cups of coffee made each
day. Although one can find coffee anywhere on Earth, the
number of cups made pales in comparison to the activity on
YouTube, Facebook, and elsewhere, despite the fact that
data generated there comes from approximately 60 percent
of the world’s population that has internet access. [34]
The volume of internet data generated each day becomes
even more impressive when compared to two boxes in the
graphic’s lower right corner. In stark contrast to the hulking
number of videos viewed and like buttons clicked daily, are
the votes cast in the 2016 U.S. presidential election
(represented in blue at approximately 100 million), and in
red, the votes cast earlier that year in the UK’s European
Union referendum (approximately 30 million to decide the
issue of Brexit).
Without a doubt, today’s digital platforms generate an
enormous amount of data.

Rise of the Machines: Machine Learning

and Analytics
The explosion of digital data has been a catalyst for
innovation. The growth of technologies to manage this
volume of data and analytics tools to make sense of it have
been critical developments for the analyst. The next big
force for change – machine learning – is quickening the pace
of innovation in several fields. Machine learning has a long
and varied history, however.
IBM’s Arthur Samuel introduced machine learning in 1959 to
describe a computer that could be programmed to learn. [35]
In the simplest sense, machine learning is the use of
computer-powered algorithms to analyze data for patterns.
What makes machine learning such a transformative
approach to analysis is that the volume of data that can run
through the algorithms is limitless. Should the computers
behind the algorithms reach their data volume capacity, an
infinite amount of computing power can be added to
increase that capacity.
A subset of artificial intelligence (AI), machine learning
separates itself from AI because AI algorithms depend on
human-created rules. In contrast, computers that use
machine learning programs can effectively create rules for
themselves as they “learn.” Learning in this sense refers to
computers’ attempts to optimize outputs from their
algorithms. In other words, a machine learning algorithm
designed to maximize the percent of consumers who click
on a display ad (the algorithm’s target “output”) will churn
through thousands, millions, or perhaps billions of instances
when consumers saw the ad. It will then correlate what’s
known about those consumers and the various elements of
the ad (the algorithm’s data “inputs”) to determine the
“right” conditions for consumer engagement. While human
programmers code some of the initial parameters in the
machine learning algorithms, the computers take it from
there to constantly test and reveal insights. The evolving
field of “Deep Learning,” a further subset of machine
learning, provides the computer even more latitude to adapt
through the use of deep artificial neural networks.
Machine learning’s modern applications are greatest in the
fields of data analysis and pattern recognition, and machine
learning is playing an increasingly important role in
marketing. A machine learning program can find patterns in
a dataset using regression models and clustering techniques
to map inputs to outputs based on example input:output
pairings (referred to as supervised learning). The program
can also find patterns in a dataset, drawing inferences
without reference to a labeled outcome (unsupervised
learning). The program can do this quickly, efficiently, and
without bias.
Machine learning is used in spam filters in your email inbox
and recommended videos on YouTube. Computer-aided
medical diagnoses that can improve disease recognition,
patient care, and outcomes depend on machine learning.
Self-driving cars, which can reduce the risk of automobile
accidents, are powered by sophisticated machine learning
programs. Anheuser-Busch has used machine learning to
optimize delivery routes, improving efficiency and driver
satisfaction. [36] Indeed, machine learning can improve many
aspects of everyday life.
The modern applications of machine learning are felt more
profoundly in digital marketing than anywhere else, given
the vast amounts of accessible data in the field. Its primary
function is determining consumer segments based on
observed behaviors to improve and customize the
consumer’s journey. For example, machine learning
algorithms working on consumer data can match this data
with products or services that consumers who have similar
interests selected. Retailers like Walmart, Alibaba, The North
Face, Amazon, and eBay have all used machine learning to
cater messages to individuals more precisely than ever
before. Perhaps most famously, Target employed a machine
learning algorithm developed by statistician and leading AI
expert Andrew Pole to analyze women’s purchases in an
attempt to identify which shoppers were likely to be
pregnant. The algorithm revealed key indicators like the
purchase of certain supplements, unscented lotion, and
other products that were highly correlated with subsequent
baby registry enrollments. Target used these insights to
send coupons to consumers the model identified, including
a 16-year-old, revealing her unintended pregnancy to her
father. Target adapted its strategy to mix other offers in with
the pregnancy-focused promotions after finding that
customers were uncomfortable with this degree of
personalization. While expanding ethical concerns and
growing limitations in data collection have affected the
robustness of machine learning algorithms, programs built
on publicly available information can still effectively tailor
advertisements a person receives.
Personalized ads have long been viewed as a key to
marketing effectiveness and the health of the relationships
brands maintain with consumers. In a 2016 study Demand
Metric conducted in partnership with Seismic, “Content
Marketing’s Evolution: The Age Of Hyper Personalization And
Automation,” 80 percent of respondents said personalized
content was more effective or much more effective than
generic content. [37] More recently, a June 2019 Infogroup
study revealed that 90 percent of consumers found
messages from companies that aren’t personally relevant to
them “annoying.” [38] But machine learning has been shown
to dramatically increase the relevance and personalization
of ads by analyzing consumer behavior to build highly
customized audiences around brand-specific awareness
measures. Through a combination of survey and response
data, machine learning algorithms can categorize
consumers successfully based on their positions in the
consumer journey: those requiring greater awareness, those
who are working through the stage of consideration, and
those who exhibit purchase intent. Audiences built in this
way provide brands with the opportunity to move
consumers seamlessly through the consumer journey by
matching messages to their personalized needs.
The future of machine learning in marketing is exciting. In
addition to better consumer understanding, other
applications of machine learning are evolving quickly.
Chatbots, for example, can function on the same level as
human customer service representatives. Some of the most
exciting advancements are in content curation and creation.
A sophisticated algorithm could eventually be instrumental
in the creation of digital media, whether making suggestions
on end-to-end strategies for content marketing, providing
content topics, or even creating content.
That said, adopting machine learning techniques in the
marketer’s repertoire makes the analyst’s role even more
important. While algorithms powering the machine learning
programs can identify unseen patterns, it’s the analyst who
interprets those patterns. These tools should be seen as
enablers – not replacements – for analysts. As a result, it’s
imperative that analysts build their Techie skills, so they can
lead the use of machine learning in their organizations. It’s
even more important that analysts hone their Business
Strategist skills, so they can place machine learning insights
in the context of the advertiser’s broader business and
make smarter, more informed decisions.

In Summary: The Rise of Modern Data

Analytics
Data has become the most valuable asset for anyone who
makes or attempts to influence decisions. To help make
sense of the consumer journey and the role data could play,
P&G introduced the concept of a three-step model that
detailed the role of the brand in a consumer’s decisions in
2005. The amount of data consumers and brands generate
has grown significantly since then. As a result, Google
adapted P&G’s framework to the changing data landscape
by introducing the concept of the Zero Moment of Truth.
Google’s ZMOT remains durable through the explosive
growth of data in today’s marketing and advertising
environment. Several influences drove the explosive growth
of data and new customer touch points, creating new
marketing opportunities that led to the growing demand for
analytics in today’s marketing and advertising environment.
 PART 2: Consumer/Brand
Relationships
Lesson 1 Online Video Lesson 2 Online Search
Lesson 3 Display Media Lesson 4 Social Media
Lesson 5 The Consumer Decision Journey The
methods that brands use to build relationships with
consumers – online video, search, display ads, and social
media – give analysts a wealth of data about behaviors on
these platforms. Knowing how to assess successful
consumer/brand relationships and understanding a
consumer’s purchase journey requires a useable framework
for parsing this data.
In Part 2, we explore each digital channel in-depth, including
a discussion of key metrics and measurements, how
consumers interact with brands on each platform, and ways
of organizing consumer data that enable actionable insights.
PART 2 | LESSON 1
 Online Video

Five things discussed in this lesson:

Online video is a very important way for brands

to engage with consumers

Online video’s blend of sight, sound, and motion

provides brands with unique opportunities to craft
consumer experiences

YouTube is the world’s largest online video

platform, but the market is quickly evolving to
include streaming services from traditional video
creators and new entrants

Consumer behavior has seen a significant shift

in the way video content is enjoyed, its role in
entertainment choices, and the influence it has on
purchase decisions

Measuring successful online video campaigns

requires the use of new metrics, as well as traditional
video measures adapted for online video delivery

Online video (OLV) is a very important way for brands to

engage consumers through multisensory experiences –
sight, sound, and motion – and its emergence has
revolutionized the entertainment industry, as well as the
way people around the world use the internet. The 2016
announcement that OLV stars are eligible for Emmy
nominations demonstrates this shift and is a testament to
the fact that short-form digital content can be high-quality
entertainment, regardless of where people view it.
By developing content that resonates with the interests and
passions of viewers today, OLV creators have created
authentic connections with their fans and amassed large
followings. Stars such as PewDiePie, Smosh, and Jenna
Marbles are frequently considered more popular and
influential than many mainstream celebrities. In fact, a
recent Variety study found that YouTube sensation
PewDiePie was more popular than basketball great LeBron
James among Generation Z males. [39]

Online Video (OLV) Market Share

OLV is comprised of “user-generated content” models such
as YouTube and Facebook, but also includes full episode
players (FEP) and over-the-top (OTT) streaming services,
including Hulu, Netflix, and Disney+. YouTube remains the
largest online video platform, reaching 90 percent of the
U.S. population, according to AudienceProject. [40]
Figure 2.1: Comparative reach of OLV platforms today [41]

Platfor U.S. Notes

m Reach
YouTube 90% World’s largest video-sharing site owned
Faceboo 60% by Google World’s largest social media
k 35% site used for video reach Facebook’s
Instagra 21% image-based network offering 15- to 60-
m 18% second videos Social network where
Twitter 17% videos earn quick bursts of reach Social
Snapch app that offers video content but no
at archive Video-based news site AT&T’s
WarnerMedia owns
CNN

comScore reports that 85 to 90 percent of internet users in the

U.S. view online video. Brands are a critical component of the
OLV experience, as marketing strategist Mehmood Hanif
estimates the average internet user is served 11,250 ads per
month. [42]

Consumer Behavior and OLV

As OLV increasingly becomes more important, the way people
watch it is changing. Google documented several interesting
trends in 2019. These trends are affecting consumer behavior
in ways beyond how people access entertainment. In fact, OLV
plays a significant role in shopping behaviors: ➔ Watch time
for shopping-related videos on YouTube grew in the U.S. by
more than five times over the past two years. [43]
➔ Over the past three years, the number of YouTube channels
with more than 1 billion views has grown by five times. [44]
➔ More than half of shoppers say online video has helped
them decide which brand or product to buy. [45]
➔ More than 90 percent of people say they discover new
brands or products on YouTube. [46]
➔ More than 40 percent of global shoppers say they’ve
purchased products they discovered on YouTube. [47]
Such shifts in consumer behavior have been a poignant
illustration of the power of digital platforms. Moreover,
changes in consumer behavior have left traditional linear
television companies scrambling to reimagine the way they
deliver entertainment. The culmination of these influences has
led marketers to rethink their approaches to advertising. The
implications for analysts are far-reaching.
Television has been one of the most influential creations since
its inception in 1925. But its influence is waning today.
According to eMarketer, the average amount of time U.S.
adults watch TV has been on a steady decline. Time spent
watching live, DVR, and other prerecorded television (such as
video downloaded from the internet but saved locally) reached
a peak of 4.6 hours per day in 2012 but has fallen to 3.8 hours
in 2020 – a drop of more than 17 percent. [48] eMarketer
projects that slide to continue.
Conversely, time spent engaging with digital content, and
specifically digital video content, has been on the rise. Over
the same 2012 to 2020 period, the number of hours U.S.
adults spent on digital platforms increased nearly 80 percent,
reaching 7.5 hours per day. [49] Time spent watching digital
video increased an astonishing 240 percent in that period,
reaching more than two hours per day. [50]
From streaming services, such as Netflix, to media platforms
like YouTube, it’s undeniable that video viewing has trended
away from conventional television. The average number of
viewers throughout 2019 for the four largest TV networks, CBS
(7.1 million), NBC (6.3 million), ABC (5.2 million), and Fox (4.6
million), is a combined 23.3 million Americans. [51] This total is
undoubtedly exaggerated as it ignores that many of those
people likely overlap. While the number of Americans
watching network television is not insignificant, it’s a far cry
from Netflix’s 73 million U.S. subscribers (182.2 million
globally) [52] and nowhere near the 175.5 million Americans
who visit YouTube each month [53] (with more than 2 billion
people globally viewing videos on YouTube monthly [54] ). The
overwhelming size of digital video platforms is a clear
indication that consumers look favorably on the breadth of
content choices, ease of use, and simple access. It’s important
for analysts to understand the motivations behind such shifts
in consumer behavior. By doing so, they can help guide
decisions on where brands should place content to reach
consumers.
Who is watching traditional and digital video is an equally
important question. Apparent demographic differences among
consumers are driving the shift from traditional television to
digital video. Younger viewers, frankly, don’t watch TV like
past generations. According to eMarketer, Generation Z
consumers (under the age of 23) spent less than one-third as
much time in front of the television as boomers (age 56 and
older). Millennials (age 24 to 39 in 2020) and Generation X
consumers (age 40 to 55) watch less TV than boomers as well.
[55]
The trend toward older viewers watching television was
clear in a 2018 AdAge analysis: “If the broadcast networks
were living, breathing human beings, they’d be closer to the
tomb than the womb. Last season, the median age of the
primetime CBS viewer was 61 years old, or a few candles
north of the display that illuminates the birthday cake of the
average NBC (57 years) or ABC (55) enthusiast. With a median
age of 51 years, Fox is now eligible for AARP membership, and
even the hip, teen-friendly CW is just a few years shy of aging
out of the 18-to-49 demo.” [56]
Digital video, on the other hand, exhibits greater usage among
younger consumers. More than 90 percent of U.S. teens and
adults under the age of 45 used digital platforms (including 98
percent of Americans age 18 to 24) while fewer than 70
percent of boomers visit such sites. [57] Understanding the
demographic makeup of consumers can inform analysts’
recommendations on which platforms are ideal for reaching
consumer segments.
OLV’s influence on shopping behaviors and entertainment
choices offers opportunities for brands to create content that
influences consumers. Doing so effectively requires
understanding how to measure the success of those efforts.

Key OLV Measures

Brands can evaluate success in using online video platforms
through a blend of traditional methods that are used to
measure television (the original video platform) and new
metrics that measure OLV’s unique capabilities.
Reach
Reach is a measurement of the size of the audience that
watches an ad. This is a classic metric used as a key measure
of success for most traditional media, particularly television,
and is calculated as Reach (%) = Gross rating points (%) /
Frequency.
Frequency
Frequency is the number of times the ad was exposed to an
average person or household during a given period. For
example, if the ad reached 30 million households in an
audience or region that has 10 million households, its
frequency would be three. Frequency is also used to assess
television and other traditional media.
X+ Reach (Effective Reach) Effective reach is a target
detailing the number of people who see an ad with a
frequency at or above the number of times deemed to
be most effective. For example, “1+ Reach” means
everyone who saw the ad at least once. This is also a
measure used extensively to track traditional media
channels.
Rating Points Expressed as a percentage, a Rating Point
is 1 percent of the potential audience. For example, if
25 percent of all targeted viewers saw an advertiser’s
commercial, the advertiser has achieved 25 rating
points. Rating Points were conceived to track the
effectiveness of traditional television campaigns but
are used increasingly to track OLV.
Gross Rating Points (GRPs) GRPs measure the total of all
Rating Points an advertiser earns during a campaign. The
measure is calculated by simply summing the Rating Points of
all ads run during the campaign. Alternatively, GRPs can be
calculated by multiplying a campaign’s reach by its average
frequency. For example, if a campaign reached 30% of the
market and the consumers reached saw the ad an average of
4 times (i.e., a frequency of 4), the campaign earned 120
GRPs. As you can see, this metric can be greater than (even
much greater than) 100.
In-Demo Gross Rating Points (also known as TRPs) In-
demo GRPs represent the percentage of the total GRPs
earned from an audience that fits the advertiser’s
target (e.g., women aged 35 to 64). For example, if an
advertiser’s media plan reached 10 percent of its in-
demo audience, it has 10 in-demo GRPs. This metric is
also used to gauge television effectiveness.
Completed Video Views (or Rate) Completed video
views is the number of times a consumer watches a
video to completion. This metric can also be expressed
as the percent of consumers who are served the video
and watch it all the way through (i.e., completed video
view rate). Consumers who watch the video to
completion, as well as those who don’t, give brands
important insights into their videos’ effectiveness.
Engagement Rate Often, brands want consumers to do
more than watch videos. Engagement rate – the
percentage of consumers who engage in additional
behavior (e.g., sharing the video, clicking a link in the
video, etc.) – can provide insight into the video’s ability
to encourage those desired behaviors.
Share of Audience (SOA) Measuring a brand’s relevance
to consumers provides a sense for awareness, interest,
and many other key brand attributes. Share of
audience (SOA) is one way to quantify relevance. To
calculate SOA, divide the number of subscribers to an
OLV channel by total subscribers for brand channels in
the category. SOA is inexact – after all, the same person
can subscribe to any number of OLV channels – but it
can provide a directional idea of a brand’s popularity.
In Summary: Online Video
OLV is an important way for brands to engage with
consumers. OLV’s blend of sight, sound and motion provides
brands with unique opportunities to craft consumer
experiences in a rapidly evolving marketplace. While
YouTube is the world’s largest online video platform,
streaming services from traditional video creators and new
entrants are adding new content choices for consumers.
These services seek to realize benefits created by the
significant shifts in consumer behavior in the way video
content is enjoyed, its role in entertainment choices, and
the influence it has on purchase decisions. Measuring
successful online video campaigns requires new metrics, as
well as traditional video measures that have been adapted
for online video delivery.
In the next lesson, we’ll turn our attention to what could be
the greatest marketing channel – online search.
PART 2 | LESSON 2
 Online Search

Five things discussed in this lesson:

Online search provides insights into consumer

intent like no other marketing platform

Google is the largest search engine in the

market, contributing more than 90 percent of
consumer searches conducted worldwide

Brands’ management of online search marketing

typically has one goal: improving search rank for
critical consumer searches

Recent increases in “near me” searches indicate

new online search opportunities for brands

Measuring the success of online search

marketing typically depends on a handful of critical
metrics

Online search could be the greatest marketing channel of all

time. Where else is an advertiser given the opportunity to
“talk” with consumers one-on-one and provide answers to
uniquely personal questions? Think of the last time you were
in a store and a sales associate approached you. How
guarded were you? Did you withhold information to maintain
negotiation leverage? Did you withhold information because
you didn’t feel comfortable discussing it with a stranger?
Now, compare that to the last time you visited an online
search engine. Most likely, the difference in transparency is
significant.
The degree of transparency that consumers show when
searching online is one of the clearest signals of their intent.
Incorporating that intent into an advertiser’s marketing
efforts requires a thoughtful approach to online search
advertising.
Search Engine Market Share
A significant number of search engines operate online, each
working in the same general way: People show up with
questions, and the search engines find the answers they see
as most relevant, and then present the answers.
Each search engine has its own methods for collecting
information from the web and determining relevance, which
can vary significantly. But one thing that’s consistent in the
world of search engines is that most people choose Google
to conduct their searches. Recent StatCounter data show
Google had the largest search engine market share
worldwide in 2019:
Figure 2.2: Search engine share of page referrals [58]

Engine Global Notes

Share
Google 92.8% World’s most popular search site
2.6% founded in 1998
1.6% Search engine that Microsoft owns
0.9% and operates Search network that
0.5% Verizon Media owns Dominant search
0.4% engine in the Chinese market Largest
search engine in the Russian market
today U.S. search engine that
emphasizes user privacy
Bing
Yahoo!
Baidu
Yandex Ru
DuckDuck
Go

Presented visually, Google’s large share of page referrals is

even more pronounced: Figure 2.3: Graphical view of search
engine share of page referrals [59]

Before we go too deep into online search as a marketing

platform, we need to talk about two important concepts:
search engine optimization (SEO) and search engine
management (SEM). Each concept is very important for brands
and means different things for them.

Search Engine Management Versus Search

Engine Optimization
Search engine management is about brands investing in paid
search to reach consumers. Although the placement of paid
search results changes as search engines frequently redesign
their search engine results pages (SERPs), paid search ads can
generally be seen at the top and alongside search results and
are distinguished through a Federal Trade Commission-
mandated symbol. [60] Here in Figure 2.4 are those ads and
symbols: Figure 2.4: Sample SERP from Google

How paid ads are returned is a complicated system based on a

number of factors. Hal Varian, Google’s chief economist,
published a video in 2014 that expertly simplified that
process. [61] Varian described the elements that go into
determining the order of paid search results on the SERP as: ➔
The advertiser’s bid in Google’s ad auction.
➔ The click-through rate (i.e., the percent of people who see
the ad and click on it) that Google expects that result to earn.
➔ How well the page linked to the ad (i.e., the “landing page”)
is organized to create a good consumer experience.
➔ The relevance of that ad to the question the searcher asked.
➔ How that search ad uses the various ad formats (e.g.,
additional links, descriptions, etc.) that Google offers
advertisers.
An advertiser’s management of these factors across search
engines is SEM.
Meanwhile, SEO is related to organic results. Those organic
results are presented on the SERP below the ads, which
generally land at the top of the results. Search engine
algorithms trigger organic results similar to the way they
present paid search results; however, organic results don’t
have a bid component. In other words, brands can’t use
money to influence how and where search engines display
organic content, which effectively levels the playing field for
big and small content owners.
SEM and SEO differ significantly in the way they promote
internet content, but they have similar objectives. Advertisers
want their content to be where consumers are when they’re
searching for something relevant to advertisers’ brands.
The Importance of Search Rank
In 2007, InquireO published a study that demonstrated the
importance of brands being at the top of the SERP. By
overlaying a Google search page with a graphic of where
consumers' eyes fell on the page, InspireO visualized this idea
perfectly. The majority of consumers look at the first couple of
results before clicking a link.
A recent study from Travel Tripper makes this clear. The online
travel site and technology provider, which recently merged
with Pegasus, worked with research company Sticky to test
consumers’ interaction with a SERP for a hotel-related query. A
sample size of 100 users had 15 seconds to look at each page.
Sticky recorded eye tracking and the results can be seen here:
Figure 2.5: Sticky’s Google SERP eye tracking results [62]
Some studies say consumers expect the best brands to be at
the top of that list, whether through organic results or paid
search engine management advertising. If a brand’s objective
is to be relevant to consumers when they’re searching for
something the brand can offer, a high search rank – the
position the brand’s website or page is returned on the SERP –
is important to success.
Consumer Behavior and Search
The way consumers engage with online search engines is
evolving. Recent trends in queries demonstrate shifts in
consumer interests, needs, and intent. Google reports that
near me searches are increasing across almost every
category, with people combining locally sensitive searches
with other intent signals.
In 2018 and 2019, “where to buy” + “near me” mobile queries
increased more than 200 percent.
Examples include “where to buy flowers near me,” “where to
buy stamps near me,” and “where to buy pumpkins near me.”
[63]
During that time, mobile searches for “store open near me”
(“grocery store open near me” and “auto parts store open
near me”), and “on sale” + “near me” (“tires on sale near me”
and “houses on sale near me”) increased more than 250
percent year over year. [64]
Key Search Measures
As we’ve seen, understanding what consumers look for as
they search online is an important component to a brand’s
success. But how can the brand understand what works (and
what doesn’t) in search marketing? Every brand online should
track a few critical metrics: Click-Through Rate
Click-through rate is a very important metric that identifies the
percent of consumers who saw a search ad and clicked on it.
The number can help gauge the effectiveness of search
strategies that target consumers, as well as the design’s
effectiveness in engaging consumers.
Conversion Rate
Conversion rate is the percent of consumers who, after
clicking on a search ad, take action on the linked page.
Tracked actions include making a purchase on the page,
downloading a file, or any number of activities the brand
deems to be positive.
Cost Per Action
Another metric that’s important to assessing search marketing
success is cost per action (CPA). This measure takes the cost
invested in a search campaign and divides that cost by the
number of a tracked consumer action. Those actions could be
clicks – cost per click, or CPC – or any other conversion actions
the brand deems important.
Impression Share
Impression share is the number of impressions a search ad
received divided by the number of impressions the ad was
eligible to receive. Effectively, impression share measures the
percent of times an ad showed up in a SERP when a consumer
entered a search on which the advertisers placed a bid.
Eligible impressions are estimated using many factors,
including targeting settings, approval statuses, and quality.
Impression share is a good way to understand how well an ad
has been optimized to achieve campaign goals (i.e., whether
an ad might reach more consumers if its bid or budget was
increased, its targeting was set a different way, etc.).
Share of Search (SOS) Measuring a brand’s relevance to
consumers’ online searches provides a sense for
awareness, interest, and many other key brand
attributes. Calculate share of search (SOS) by dividing
the number of searches for a brand by total searches
conducted for brands in the category. For example,
auto manufactures can gain great insight into the
popularity of their vehicles by comparing the searches
conducted for one of their sedans against all searches
for sedans. Analysts can conduct SOS at a company
level, brand level, or product level, and can compare
the item being searched against the entire market or a
select group of key competitors.
In Summary: Online Search
In this lesson, we saw how online search provides brands with
insight into consumer intent like no other marketing platform.
Google is the largest search engine in the market, with more
than 90 percent of consumer searches worldwide. Brands’
management of online search marketing typically has one
goal: improving search rank for critical consumer searches. A
recent increase in near me searches indicates new online
search opportunities for brands. Finally, measuring the
success of online search marketing typically depends on a
handful of critical metrics.
In the next lesson, we’ll take a look at how brands can use
display media to connect with consumers at various points
along their decision journeys.
PART 2 | LESSON 3
 Display Media

Five things discussed in this lesson:

Display ads have evolved significantly since the

format debuted in 1994

Digital display ads come in many formats,

including banners, video, rich media, and
sponsorships

Each category of display ads grew consistently

and steadily from 2015 to 2019

Display ads have evolved primarily in their

ability to target specific consumers across the
internet, but privacy concerns have cast doubt on
how effective display ads will be in the future

The effectiveness of display strategies,

campaigns, and creative is measured similarly as
search media

The display ad of today hardly resembles the banner ads of

25 years ago.
Introduced in 1994, display ads are one of the oldest forms
of digital advertising and signaled a change in the way
brands could interact with consumers online. Display ads
contributed to the rise of data analytics as we know it today.
But the static ads of yesterday have been updated to
include video, dynamic content, and other forms of
immersive experience that are designed to engage
consumers at key moments in their purchasing journeys.
Coupled with remarketing technology, the display ad can
play a critical and efficient role in a brand's advertising and
marketing efforts.
Display Media Market Share
Digital display ads come in many formats. The primary
types of digital display ads include banners and similar
executions, video ads that aren’t in a player, rich media ads
that integrate streaming activity (often through the use of
Flash or JavaScript to allow users to interact with content),
and sponsorships in which advertisers pay for custom
content and/or experiences presented in a display format.
[65]
According to data from Invesp CEO Khalid Saleh, each
category of display ads grew consistently and steadily from
2015 to 2019.
Figure 2.6: Estimated U.S. display ad spending by format
($Billions) [66]
Ad 2015 2016 2017 2018 2019
Format
Banners
and Others $11.5 $13. $14.7 $16. $17.6
39 17
7 4 8
Video
$7.46 $9.5 $11.4 $13. $14.7
9 05
3 7
Rich Media
$5.44 $7.4 $9.17 $10. $12.1
2 69
9
Sponsorshi
ps $1.68 $1.7 $1.86 $1.9 $2.06
7 6

Placing digital display media has increasingly shifted to

programmatic purchases because of the format’s
straightforward nature. Programmatic media buying is the
data-intensive algorithmic purchase and sale of advertising
space in real time. Software automates the buying, placement,
and optimization of media inventory via a bidding system. The
efficiency of programmatic media buying has had a positive
effect on display ad spend.

Evolution of Display Ad Formats

Display media has evolved with new formats and features
aimed at increasing consumer engagement. Video, rich media,
and custom-constructed sponsorships are examples of formats
that build on the static banner concept by introducing
attention-grabbing motion, sound, and content. Display ads
have primarily evolved, however, in how they can target
specific consumers on the internet.
You’ve probably experienced display ads’ targeting
capabilities. After visiting a retailer’s page to check out a pair
of shoes, you likely noticed those shoes popping up in display
ads on nearly every page you subsequently visited. This is
because digital data placed on your device when you checked
out the shoes (in the form of digital cookies, which we’ll
discuss later in this book) can inform the bidding strategies of
sophisticated display advertisers, ensuring they remind you of
the items. A recent study by Washington Post technology
columnist Geoffrey A. Fowler found that in one week of
browsing the internet with Google Chrome, more than 11,000
tracking cookies were deposited into his browser. [67]
For the analyst, tracking cookies can help identify consumers
who exhibit an intent to purchase products and services based
on behavior across digital properties. As Fowler’s report makes
clear, however, the use of tracking cookies tied to display
advertising is at the center of the privacy issue being debated
in the digital marketing industry today. We’ll explore the
intricate topic of privacy in Part 3 of this book. At this point, it’s
more important to understand that display ads’ ability to
target consumers in the future – and therefore, their
usefulness to analysts – is less than certain.

Key Display Media Measures

As an interactive media unit, display ads are measured by
metrics that gauge the effectiveness and efficiency of
consumer engagement. Typically, analysts would assess
display media using the same metrics found in a search
campaign.
Click-Through Rate
As with search ads, click-through rate is an important metric
for display media. Tracking the percent of consumers who saw
a brand’s display ad and clicked on it gauges the effectiveness
of display strategies that target consumers, as well as the
effectiveness of the display ad’s creative to engage
consumers.
Engagement Rate
Similar to the metric used to gauge video engagement, the
engagement rate of a display ad tracks the rate by which
consumers interact with the ad.
Conversion Rate
Conversion rate is calculated as the percent of consumers who,
after clicking on a display ad, take action on the linked page.
Tracked actions could be making a purchase on the page,
downloading a file, or any number of activities the brand
deems to be positive.
Cost Per Action
Cost per action (CPA) takes the cost invested in a display
campaign and divides that cost by the number of tracked
consumer actions. As with search, the action could be clicks on
the display ad (cost per click, or CPC), or any other action the
brand deems important.

In Summary: Display Media

Display ads have evolved significantly since they debuted in
1994, adding a number of features to increase consumer
engagement. Today’s digital display comes in many formats,
including banners, video, rich media, and sponsorships. Each
category of display ads grew consistently and steadily from
2015 to 2019. Display ads have primarily evolved in their
ability to target specific consumers across the internet, but
privacy concerns have cast doubt on how effective display ads
will be in the future. Measuring the effectiveness of display
strategies, campaigns, and creative can be done in much the
same way as with search media.
In the next lesson, we’ll discuss how brands can use social
media to develop consumer relationships and earn important
insights into consumer behavior.
PART 2 | LESSON 4
 Social Media

Five things discussed in this lesson:

Social media usage continues to grow, with

usage set to hit 3 billion consumers worldwide by
2021

Which social media platform a consumer uses

varies significantly with age and need

While each major social media platform offers

unique positioning to consumers, Facebook is the
largest player

Brands most frequently use social media

through sponsored posts that seek to influence
consumer decisions authentically and organically

Measuring the success of a brand’s social media

efforts requires metrics that are unique to each
platform and those that apply across a brand’s
marketing channels

Since its founding in 2004, Facebook has been the dominant

social media platform in the world. It’s changed the way
people communicate, maintain personal relationships, and
investigate brands for the 3.2 billion users of social media
today. [68] Behind Facebook and its image-based social media
property, Instagram, stands platforms that include Pinterest,
Twitter, Snapchat, and LinkedIn. Combined, these six
players represent the majority of social media engagement,
and each has carved a unique niche in the social media
landscape.
Despite the recent privacy and data challenges social media
platforms experienced (Facebook, in particular), social
media remains a critically important channel for brands.
Analysts can mine tremendously robust information and
deep insights from social media.
Social Media Market Share
Social media use is increasing around the world. Without a
doubt, it’s one of the most popular online activities for
people of all ages. According to Yapmedia, in 2019, 3.2
billion social media users were active, and that number is
growing worldwide. [69] That equates to about 42 percent of
Earth’s population.
Facebook dominates the social media landscape (see
market share data as of November 2019 from StatCounter
in the table below). Young Americans, especially those ages
18 to 24, stand out for embracing a variety of platforms and
using them frequently. Who else uses social media? Active
users include 90.4 percent of millennials, 77.5 percent of
Generation X, and 48.2 percent of baby boomers.
Figure 2.7: Worldwide share of social media traffic [70]

Networ Global Notes

k Share
Faceboo 61.3% World’s largest (and original) social
k Twitter 14.5% network Unique value proposition due to
11.5% intensity of sharing Image-based social
8.4% site with robust integration to Web App-
focused photo and video site that
Facebook owns
Pinteres
t
Instagra
m
Consumer Behavior and Social Media
The social platform a user chooses typically depends on need.
For business connections, a person turns to LinkedIn, while
someone who wants to connect with others on a personal level
is more likely to fire up Facebook. Age also affects platform
choice. Approximately 78 percent of 18- to 24-year-olds use
Snapchat and a sizeable majority of these users (71 percent)
visit the platform multiple times per day. [71]
Furthermore, social media has matured to become a strong
influencer in buying decisions. Social media channels give
customers an easy way to contact brands. Customers can
learn about brands’ organizations and browse their friends’
opinions. In late 2019, Lyfe Marketing documented three facts
that support the power social media has on the consumer
decision journey: ➔ 81 percent of customers make buying
decisions based on their friends’ social media posts (Market
Force). In other words, when followers engage with an
advertiser’s page, they not only connect with those
consumers, but also potentially influence all of their
connections.
➔ 71 percent say social media referrals make them more
likely to purchase an item (HubSpot). When people refer a
brand on social media, the advertiser boosts its sales potential
with everyone who sees that post.
➔ 133 percent increase in conversions when shoppers on
mobile see good reviews about a brand before they buy an
item (Bazaarvoice). Great reviews on Facebook and other
social media sites can help convince people to try a business.
How Brands Use Social Media
A sponsored post integrated into a user’s feed is the most
common ad product social media platforms sell. These posts
can feature all types of digital advertising mentioned thus far:
display ads, video ads, and even search ads from a social
media platform’s search page.
Social media ads tend to gravitate toward common
appearance. “Sameness,” in this sense, is by design as the
platforms monitor one another for innovations and
optimizations. Snapchat – with its young consumer segment –
has produced less traditional (perhaps more innovative?) ads
than other platforms, although the benefit is questionable. In a
similar (albeit, antithetical) way, LinkedIn has crafted
advertising that serves specific purposes for specific consumer
segments. Other social media platforms tend to copy
innovations they believe will have a positive effect on
consumer behaviors.
Advertisers can use each social media network to realize a
unique objective, despite the commonality in ad formats.
Facebook’s enormous reach and consumer data allow brands
to connect to a great number of consumers in a highly
targeted, personalized way. Twitter’s connectiveness and
unique format (i.e., limited character content delivered in
short bursts) enables brands to reach targeted segments of
consumers quickly and efficiently through rapid sharing.
Pinterest’s use of “image pinning” reveals internet browsing
behavior. Instagram works in conjunction with Facebook’s
social network, allowing advertisers to connect with
consumers in different ways while complementing
simultaneous Facebook campaigns.
Guiding an advertiser in the successful use of social media –
including the leading networks mentioned here and those that
weren’t discussed – requires a clear understanding of each
network’s position in the social media market. Through this
understanding, the analyst can design objectives that fit each
network while carefully weighing its pros and cons.
Key Social Media Measures
How brands measure return on social media investments is a
function of brand focus and the chosen social media network’s
measurement offerings.
Audience Growth When tracking audience growth,
analysts look at how many people are connected to a
brand on social media. “Followers” are a common
designation on social media platforms for people who
have agreed to receive content updates from brands
(e.g., Twitter followers). Picking two points in time and
measuring the increase in followers provides audience
growth. Clearly, the faster a brand increases its
audience and the bigger the audience is, the better for
that brand.
Amplification Rate Amplification rate measures the
advocacy that consumers show toward content on a
brand’s social media platform. For example, if
consumers share videos or comments from a brand’s
social media feed on their social channels, that’s
amplification. Amplification rate tracks the volume of
“shares” for a piece of content, or the rate at which
those shares are collected (i.e., the number of
consumers who shared the content / the total number
of consumers who saw the content). Amplification is
important for many reasons – namely, when brands talk
to consumers, brands want consumers to relay their
messages to others. A brand gains credibility when
consumers share the brand’s words. In addition,
consumer endorsements are perceived as being more
credible than marketing messages. The higher the
amplification rate, the better for the brand.
Applause Rate Applause rate is different from
amplification rate. Applause rate measures the degree
to which consumers react positively to a brand’s work.
Positive reactions can come from a “Like” on Facebook,
a favoriting of a tweet on Twitter, or saving an image to
a Pinterest account. Applause rate is calculated as the
percent of users who see the content that has the
desired positive reaction. Applause is not sharing the
content, so this measure stops short of quantifying
brand advocacy. Applause rate can still be an effective
measurement for identifying the content that does or
doesn’t work for consumers.
Click-Through Rate As with search and display ads,
click-through rate is an important metric for social
media ads and campaigns when a consumer can click
on an ad and be sent to another destination, such as
the advertised company’s website. Tracking the percent
of consumers who saw a brand’s ad and clicked on it
gauges the effectiveness of display strategies that
target consumers, as well as the effectiveness of the
display ad’s creative to engage consumers.
Engagement Rate In the context of social media,
engagement rate measures the percentage of
consumers who engage in additional behavior (e.g.,
comments on Facebook, re-pins on Pinterest, brand
mentions on Twitter, etc.). Similar to the way this
metric works with OLV, engagement rate can provide
insight into the video’s ability to encourage those
desired behaviors.
Reach
As with OLV, reach, in the context of social media, is a
measurement of the size of the audience that sees an ad.
Reach applies to video ads, sponsored posts, and any other ad
types on a social media platform.
Share of Voice (SOV) Similar to SOS (see Lesson 2:
Online Search), brands can measure their relevance by
conducting a share of voice (SOV) analysis. To calculate
SOV, divide the number of social mentions for a brand
by total social mentions for brands in the category. SOV
can be conducted at a company level, brand level, or
product level, and can compare the item being
mentioned by consumers against the entire market or a
select group of key competitors.
In Summary: Social Media
The role social media plays in the consumer/brand relationship
has evolved significantly since Facebook’s founding in 2004.
Social media usage continues to grow, with more than 3 billion
users worldwide. The emergence of other social media
platforms has offered consumers additional choices. Each
platform plays a unique role in the market and, therefore,
enjoys a unique role in creating consumer relationships for
brands. Which social media platform a consumer uses varies
significantly with age and purpose. While each major social
media platform offers unique positioning to consumers,
Facebook and Instagram are the dominant players in the
market. Brands most frequently use social media through the
use of sponsored posts that seek to influence consumer
decisions authentically and organically. Measuring the success
of a brand’s social media efforts requires metrics that are
unique to the social media and those that apply across a
brand’s marketing channels.
In the next lesson, we will explore a framework analysts can
use to organize and make sense of the consumer data
collected from digital platforms, McKinsey’s Consumer
Decision Journey.
PART 2 | LESSON 5
The Consumer Decision Journey

Five things discussed in this lesson:

Consumers’ digital lives are complicated

Making sense of consumers’ digital data

requires a framework that organizes the data and
makes it easier to understand

McKinsey & Co.’s “Consumer Decision Journey”

(CDJ) is an effective framework for understanding
how consumers research and buy products

Each CDJ step triggers new questions that

brands can investigate to improve their
understanding of consumers

Data can provide answers to those questions

The Zero Moment of Truth (ZMOT), the period from when

consumers experience a stimulus that produces the
recognition of a product need (i.e., trigger) until they stand
at a shelf and choose a product, is a vitally important time
for analysts. In that ZMOT, consumers transmit significant
information that would be beneficial for brands to collect.
The problem is that the data is messy.
Consumers are bombarded by information from all types of
sources online. Product reviews can shed insight into others’
experiences with items that consumers are thinking about
buying. Discussion groups can offer an intricate web of
opinions and frustrations. Advertisements in the form of
display ads or sponsored social posts can detail product
benefits, while unboxing videos can reveal true quality.
In turn, consumers create their own information. When they
ask questions on social media, conduct searches, watch (or
skip) video ads, and engage with brands on their websites –
including filling out forms – consumers generate rich digital
data. Brands would love to collect this information to learn
about consumers and their interests.
Collecting data points for one consumer may seem
manageable. A recent Luth Research study found that
consumers who perform research for a prospective
automobile purchase – a significant investment for anyone –
experience a few hundred digital interactions before making
a decision. [72] But when one considers that most large
brands would need millions of consumers to reach a
meaningful scale for a product, those hundreds of
interactions quickly number in the billions. Analysts need a
framework to turn the data that flies around at that moment
into useful information for analysis.

The Introduction of McKinsey’s CDJ

In 2009, consulting firm McKinsey & Co., introduced the
Consumer Decision Journey (CDJ) to help analysts make
sense of the contemporary consumer path to purchase. The
CDJ offered a view into the consumer experience as an
alternative to the traditional funnel idea that the path to
purchase is a straight line that begins with a great number
of brands and progressively pushes brands out of
consideration until one brand (i.e., the one purchased) is left
standing.
I’ve found McKinsey’s CDJ to be a great framework for
organizing this data and creating actionable insights. I
learned the framework at the consulting firm I joined after
graduate school. After I left consulting for an advertising
agency, I was delighted to find the agency’s planners and
data analysts used the CDJ extensively. When I joined
Google, I found that it, too, had embraced the McKinsey
framework. The CDJ is a broadly used approach, and
chances are, you’ve worked at a firm that used it.
McKinsey’s CDJ illustrates the variety of influences on
consumers through the purchase process. It identifies a
number of critical moments that consumers experience
before they buy and enables advertisers to use digital
analytics to improve how they position and sell brands and
products.
The Steps of the CDJ
The CDJ is comprised of six elements, inclusive of the initial
action that kicks off the journey. Each step represents a
distinct phase of the decision-making process and is
traversed in the same logical order, no matter what product
a consumer is evaluating. At each step, brands have a
different set of questions they can ask to gain insights into
their consumer/brand relationships, as well as data that
analysts can use to provide answers. [73]
Figure 2.8: McKinsey’s CDJ framework in visual form
Trigger
The CDJ begins with a trigger, similar to P&G’s three-step
model of marketing. It’s the point when consumers realize
they need to begin shopping for a new home, or they see an
ad that makes them want to buy a pizza, or any stimuli that
starts a consumer on a purchase path.
At the point of the trigger, brands want to know what
creates that consumer need. Why do consumers suddenly
feel like they need their product or a competitor's product?
Was an ad the impetus? Was it an environmental issue that
popped up? The answers to those questions can provide
insights into when brands should interact most urgently with
consumers, and how they should position messages in later
stages of the decision journey.
‘Initial Consideration Set’
That trigger is followed immediately by the “Initial
Consideration Set” (ICS). The set includes the relevant
brands that pop into consumers’ heads when they’re
considering products. For example, a consumer who’s
preparing for an upcoming triathlon realizes their shoes are
too worn for training and identifies prospective
replacements: Nike, Adidas, Brooks, or any brands that
immediately spring to mind comprise this athlete’s ICS.
Being on the list doesn’t guarantee brands that the
consumer will pick their shoes. It simply gives those brands
a head start.
Questions around the Initial Consideration Set for brands are
straightforward: “Am I part of that list? Do customers
recognize my brand? Are they aware of me and my
products?” To get on that ICS, brands must build awareness,
recognition, and trust. Being a brand that a consumer
recognizes as a poor provider of the needed product may
land that brand on the ICS but doesn’t bode well for
converting that recognition into a purchase.
‘Active Evaluation’
A period of “Active Evaluation” follows the ICS. Essentially,
this time is Google’s Zero Moment of Truth (ZMOT). It’s
when consumers evaluate what products to buy, what
brands to invest in, and where to purchase them. Significant
amounts of information flow to and from consumers during
this point. It’s a critical part of the decision journey and a
key battleground for brands.
All is not lost for brands that don’t make a consumer’s ICS
because during Active Evaluation, they have the chance to
get into the consumer’s head. To do so, it’s important for
brands to understand consumers’ needs and how their
products satisfy those needs. A brand should answer the
question, “Am I positioned in a way that makes my products
attractive to consumers?” Understanding what consumers
are evaluating during that process is vitally important and a
way for brands to win during ZMOT.
‘Moment of Purchase’
Next is the “Moment of Purchase.” This is the time when a
consumer makes a decision and buys a product. McKinsey’s
Moment of Purchase equates to P&G’s First Moment of Truth.
At the Moment of Purchase, brands will ask a number of
questions, including, “Are my sales efforts resulting in wins
for my brand?” Brands should determine if the things they
do at the shelf, in the store, and during checkout facilitate
sales, so they can win at the Moment of Purchase.
‘Postpurchase Experience’
The “Postpurchase Experience” is the next step in the
buying journey. It occurs when a consumer takes a product
home and uses it. Effectively, this period is P&G’s Second
Moment of Truth. As with SMOT, the brand’s challenge is to
live up to the expectations it sets with the consumer.
In the Postpurchase Experience, the primary question
brands ask is, “Does the experience I deliver fulfill the
expectations that consumers have for my product?”
Understanding how the brand can help consumers is an
important insight this phase can reveal. This is where
customer service comes in. In addition, it helps to have a
solid product, and a reputable name behind it.
The ‘Loyalty Loop’
The final component of the CDJ is the “Loyalty Loop.” This is
where every brand wants to be. The Loyalty Loop is
basically a shortcut from a trigger to the Moment of
Purchase. When traveling along the Loyalty Loop, a
consumer experiences a trigger, and rather than going
through an Initial Consideration Set and Active Evaluation,
invests in a trusted brand straightaway.
An example of the Loyalty Loop in action is my experience
with Apple Inc.’s Macintosh laptops. My last several
computers were Macs. The next time my Mac laptop dies, I
will – as I have done in the past – willingly enter this Loyalty
Loop and buy a Mac immediately. I don’t have to think about
it, consider other brands, or evaluate my options. I trust
Apple, and the brand is an important part of my life.
The Loyalty Loop’s benefits are obvious for a brand. Bain &
Company research shows it can be five to 25 times more
expensive for a brand to earn a new customer than to keep
a current one. [74] But the benefits to the consumer –
expedited decision-making from not having to do research,
an absence of stress that’s often associated with buying a
brand for the first time, and enhanced service through
brand-sponsored rewards programs – can be just as
powerful.
Regarding the Loyalty Loop, an important question brands
need to explore is, “Do customers advocate for my brand?”
Often, people in a brand’s Loyalty Loop are highly motivated
supporters of the brand – known as brand evangelists – who
influence others toward the brand. When consumers
advocate for brands, it can indicate the brand is well-
positioned to earn a customer’s loyalty during the next
trigger
Applicability of the CDJ
One of the things I love about the CDJ framework is that it
takes something that’s extremely complicated and makes it
simple and flexible. While the time it takes consumers to
make decisions and the number of sources they use during
their evaluations may vary, the CDJ accurately describes the
decision-making process for consumers who are buying a
car, or standing in line at a grocery store and looking at
candy bars on display. The highly involved category of car
shopping may fit the CDJ more intuitively, but the
framework also applies to consumers who eye candy bars.
What can seem like a knee-jerk decision can more
accurately be broken into several important steps. Standing
in the checkout line, consumers may notice the candy and
reason that they could go for a treat. That moment of
recognition is the trigger. The candy that immediately
comes to mind is their Initial Consideration Set. As they look
at the candy and weigh their options, they’re engaged in
Active Evaluation. Choosing a treat, dropping it on the
conveyor belt, and paying for it is the Moment of Purchase.
And what Postpurchase Experience could be better than the
first bite of a candy bar? A consumer who eats only one kind
of candy bar will go from the trigger of being hungry for a
candy bar through the Loyalty Loop to a specific brand
before making that purchase.
The CDJ applies to small products as well as it does to big
purchase decisions. The only variable is the amount of time
consumers will spend in their journeys. The decision journey
that ends in a checkout aisle candy bar may be completed
in a few moments, while recent research suggests
consumers who shop for automobiles take an average of 96
days to make their purchases. [75] In either case, the CDJ is
an effective tool for mapping consumers’ thought processes
and important moments for brands.
How Ad Blockers Affect Digital
Measurement
While frameworks like the CDJ can help analysts make sense
of consumer data, technology complications can inhibit
these models. One such disruptive technology is the ad
blocker.
Ad blockers rank among the most common categories of
installed software in the United States with approximately
30 percent of consumers having some form of the
technology installed on their computers or phones. First
introduced in 2002 as an element of Mozilla’s Firefox
browser, the major ad blockers today (Adblock, Adblock
Plus, and Ublock) are all based on browser extensions. In
general, these blockers prevent ads from appearing by
inspecting each element of a webpage as it loads, cross
referencing each element against rules that include a list of
known ads, and preventing known ads from appearing. Ad
blockers typically use default ad identification rules from
easylist.to and have historically blocked display ads and
pop-ups from rendering on webpages and social media
feeds. More advanced technologies like Adblock Plus claim
to be able to block video ads on platforms like YouTube. [76]
The history of ad blocking is an interesting game of cat-and-
mouse, as ad revenue dependent digital platforms and
advertisers work to find tactics that undermine the ad
blocker and allow ads to show. Ad blockers have reacted
quickly to shore up their rules when a workaround is
discovered, and the pattern continues.
Ad blockers’ implications are far-reaching. Digital
advertising, as we’ve seen, is a fundamental element of the
internet, and ad blockers pose a serious threat to digital
marketers. Google, Facebook, Twitter, Snapchat, Tik Tok, and
nearly every digital platform operating today depend on the
revenues earned from serving digital advertisements to
consumers. It’s been estimated that a platform like
Facebook generates 98 percent of its revenue through the
ads. [77] Digital advertising is the reason visitors to these
platforms aren’t required to pay for each video they watch,
or search they conduct, or message they post. Yet, ad
blockers impede this revenue stream for platforms.
Ad blockers have the further consequence of complicating
an analyst’s ability to measure the effect of digital
advertising as they threaten to create a class of consumers
who cannot be reached through digital advertising. While
advertisers often view ad blockers with ambivalence (after
all, when ads are blocked in most models, it means the
advertiser isn’t charged) this growing segment of “digital
unreachables” should give advertisers pause and motivate
marketers to produce creative advertising consumers want
to view. Achieving this objective places even greater
emphasis on the role of analysts to test and optimize digital
advertising. For more about analysts optimizing digital ads,
see the discussion of Creative Optimization in Part 3, Lesson
3.

In Summary: The Consumer Decision

Journey
Consumers’ digital lives are busy and complicated. To make
sense of consumers’ purchase journeys, analysts needs a
framework that will allow them to organize associated data.
McKinsey’s CDJ is an effective framework. At each step of
the CDJ, brands have relevant questions and unique ways
they can use data to answer them. The CDJ applies to all
consumer purchase decisions and is an invaluable tool for
analysts. Technology complications can inhibit these
models. One such disruptive technology is the ad blocker,
which introduces even greater need for the analyst who can
account for and work to counter this disruption.
 In the next part of this book, we’ll explore the technical
ways brands collect data that feed frameworks such as the
CDJ.
 PART 3: The Science of
Analytics
Lesson 1 Digital Data Infrastructure Lesson 2
Digital Measurement Lesson 3 Analytics and
Dataviz Tools Lesson 4 Digital Marketing Maturity
Part 3 focuses on understanding digital data creation, how
brands use that data to measure digital marketing
effectiveness, and the tools and skill sets analysts need to
work effectively with data. While the contents are lightly
technical, this section veers into the colloquial as I dive into
multitouch attribution models, media mix models,
incrementality studies, and other ways analysts conduct
marketing measurement today.
Part 3 also provides a useful framework for evaluating data
analysis and visualization (“dataviz”) tools and explains the
critical importance of digital marketing maturity to analysts
and the companies for which they work.
PART 3 | LESSON 1
 Digital Data Infrastructure

Five things discussed in this lesson:

First-, second-, and third-party data are the

three primary categories of data brands use

Digital cookies, tags, and sign-in requirements

have provided the technical foundation of digital data
collection for decades

Data value was inversely related to data

availability in the internet’s early days, but that
relationship has changed in modern data analytics

Recent changes in data privacy regulations and

browser strategies have affected data collection and
use

Analysts can still make informed decisions based

on available data, even if that data isn’t as robust as
it was previously

We’ve seen the explosive growth in consumer-generated

data and the way brands can make decisions from that data
using helpful frameworks like McKinsey’s CDJ. But how is
data collected?
The basic tools that brands use to collect consumer data
have been around for more than 25 years. Brands have
come to depend on these data collection tools, whether
they collect data themselves or work with partners, such as
comScore, IRI, and Nielsen.
These tools are experiencing great change today as a result
of regulatory reform triggered by privacy concerns. The
future for these long-used tools is uncertain. But before we
investigate industry changes that threaten their viability,
let’s look at the basic categories of data and the data
collection tools that help brands understand consumers.
Categories of Data
Brands obtain three primary categories of data to help their
marketing efforts. Typically, a brand will collect data in all
three categories to learn more about consumers who
purchase their products and those who don’t.
First-Party (1P) Data
An organization collects first-party data through a direct
relationship with a consumer. This data could be earned
through the company’s sales process (e.g., sales data), its
own digital properties (e.g., web traffic data), its offline
connections with consumers (e.g., in-store data a retailer
collects and imports online), or other systems that record
the company’s direct exchange of data with consumers
(e.g., customer relationship data). First-party data usually
includes consumers’ personally identifiable information (PII)
and is well-guarded by the data owner.
Second-Party (2P) Data
When an organization owns and shares data, the data it
shares is considered second-party data to the recipient.
Examples of second-party data are survey responses a
survey vendor collects and shares with a brand or data a
brand collects through its website analytics tool (e.g., Adobe
Analytics, Google Analytics, etc.). These approaches allow a
brand to collect data through means that are outside its
areas of expertise or technical capabilities. A brand typically
obtains second-party data through a data-sharing
agreement.
Third-Party (3P) Data
Firms collect third-party data when they don’t have direct
relationships or agreements with the consumers who
generate the data. For example, third-party data can be
data scraped from public websites, purchased from the
original data owners, or inferred (i.e., modeled) from past
behavioral data. This data is aggregated across many
different sources, matched to a specific consumer and
offered for sale. Third-party data often consists of rich
behavioral or demographic data that expands beyond a
brand’s first-party data.
Brands need a system to store, sort, and analyze data
before it becomes useful, given the vast amount of data
available to marketers across the categories mentioned
above. This is where a data management platform (DMP)
can be valuable. Examples of DMPs include The Trade Desk,
Adobe Audience Manager, and Oracle DMP (formerly
BlueKai). A data management platform can help brands use
data to target specific consumer audiences and can
measure campaign performance across segments and
channels.

Digital Data Collection: Cookies, Tags and

Sign-Ins
How do brands (or systems working on behalf of brands)
collect consumers’ digital information? Historically, the
three important tools are cookies, tags, and sign-in
processes. These are probably familiar terms so I will be
brief in my descriptions.
Digital Cookies
Digital cookies are text files that sit in an internet browser’s
cache on your device, typically a desktop computer or
laptop, and allow websites and servers to identify you
through a unique cookie ID: a string of characters that is
associated with the browser storing the cookie. [78] Think of a
digital cookie as a name tag. Lou Montulli, a 24-year-old
programmer, invented cookies in 1994 and digital cookies
have played a central role in the collection of digital data
since then. They are particularly relevant in the collection of
third-party data.
Tags
Tags are strings of code that initialize when someone
performs an action in a browser, such as loading a webpage
or clicking an object displayed on that page. When
activated, tags allow analytics tools or marketing partners
to collect data related to consumer behaviors on the tagged
website. Analytics platforms, such as Google Analytics and
Adobe Analytics; testing tools, such as Adobe Target and
Optimizely; market research companies, such as comScore,
IRI, and Nielsen; and marketing partners, such as Google
and Facebook, use tags. Tags are critical for the collection of
second-party data.
Sign-Ins
Giving consumers the opportunity to create accounts and
sign in is a common practice among online content
providers. From The New York Times to Travelocity to
Facebook, digital accounts are everywhere. Sometimes,
those sign-ins are required and used to control access to
subscriber-only information kept behind a sign-in paywall
(think NYTimes.com’s premium content), sensitive financial
information, such as credit card numbers (think Travelocity),
or private personal data (think Facebook). Other content
providers, including Google, give consumers the option to
create accounts and sign in or access content without those
steps.
By tying activities to accounts, content providers can
connect consumers to behaviors on their platforms without
cookies and tags. Sign-ins are an important tool in the
collection of first-party data.
Data Availability and Value
In his 2004 book, Web Analytics Demystified, [79] Eric
Peterson introduced the “Pyramid Model of Web Analytics
Data.” He believed in an inverse relationship between the
volume of available data and its value. At the bottom of
Peterson’s pyramid are readily available web traffic data
points like “hits.” While easy to collect, this data tells us
little about consumers and is “mostly useless,” he
concluded. At the top of the pyramid are more elusive, yet
more insightful, data that point to “Uniquely Identified
Users.”
The data revolution that’s been underway since the time of
Peterson’s writing has significantly affected the availability
and accessibility of data. The biggest effect has been at the
top of Peterson’s pyramid: data for Uniquely Identified
Users. From social media posts to location histories, data
related to the behavior of identifiable consumers has
exploded in volume. While expanding the usefulness of data
for advertisers, this increased ability to understand
consumer behavior has prompted meaningful data privacy
concerns.

Data Collection and the Effect of Privacy

Concerns
Privacy concerns are popping up in every corner of the
internet. High-profile data breaches, hacker attacks on
corporate data stores, and cases of data misuse by digital
platforms have produced an expected backlash of mistrust.
This mistrust is particularly deep in the area of third-party
data collection. As a result, regulatory bodies have stepped
in to provide protection for consumers. Many companies –
some at the center of consumers’ privacy concerns – have
taken steps to improve data security.
How regulatory reform will play out in the digital space is
unclear. A few things are coming into focus, however, as
new directives around the world are raising the bar for
privacy. For example, the European General Data Protection
Regulation (GDPR) that became effective in 2019 requires
advertisers to clearly identify each party that may collect,
receive, or use data from their sites, apps, and other
properties. They must also obtain consent for that collection
and for personalized advertising. Policymakers continue to
advance legislative proposals around the world, with many
focusing on privacy. I expect action around privacy and data
protection to continue to grow.
In addition, companies in the digital realm have worked to
enhance privacy, reflecting consumer demand, and, often,
taking advantage of the market opportunity that shift in
consumer awareness creates. Web browsers are adding
features to limit digital cookie and tag use, including those
used for personalization and measurement. For example,
the latest Safari browser included “Intelligent Tracking
Prevention” (ITP), which immediately rendered unusable
digital cookies that can track users across multiple sites,
eliminating content providers’ ability to track and measure
consumer behavior. Firefox recently announced plans to
implement an enhanced approach to anti-tracking, and I
expect other platforms will make similar decisions in the
future.
Some common themes are emerging from privacy-related
regulation: Users should know who is collecting data about
them, how that data is being used, and have the
opportunity to opt out. At the heart of the issue is the idea
of user control. If consumers can’t disable data collection
technologies, a claim of users agreeing to allow companies
to collect that data rings hollow. The permission to amass
data buried inside long, nearly incomprehensible user
agreements doesn’t give consumers transparent control
over the collection and use of their personal data. Indeed,
without control, there can be no consent.
Opaque tracking and profiling techniques that don’t provide
user controls, such as the practice of “digital fingerprinting,”
whereby unique attributes of computer/browser/user
behaviors are used to identify consumers, are increasingly
popular as cookie technology has become less reliable.
Browsers, however, are reacting to such insidious tracking
by offering consumers enhanced control. Apple’s Safari
browser limits fingerprinting and other probabilistic methods
of tracking. It also blocks social media “Like” and “share”
buttons and comment widgets from tracking users without
their permission.
The regulatory and market responses to this shift in privacy
awareness have had a dramatic effect on digital
measurement by reducing the availability and accuracy of
collected data (primarily third- and second-party data, but
also first-party data to a degree). Yet, analysts can still make
informed decisions based on available data, even if that
data isn’t as robust as it once was.
Analysts can account for the loss of data and bring more
accuracy to their analysis through a thorough understanding
of privacy limitations’ impact on data collection. Moreover, a
strong understanding of the brand’s operations, objectives,
and environment (think Data Strategist skills) will sharpen
the analyst’s intuition and ability to separate truth from lies
in the data.

In Summary: Digital Data Infrastructure

First-, second-, and third-party data are the primary
categories of data brands use to better understand their
customers and consumers who have yet to purchase their
products. Digital cookies (and mobile device IDs), tags, and
sign-in requirements have provided the technical foundation
of digital data collection for decades. While data value was
inversely related to data availability in the internet’s early
days, that relationship has changed thanks, in large part, to
this important technical foundation. While recent changes in
data privacy regulations and browser strategies have
affected the collection and use of digital data, analysts can
still make informed decisions based on available data.
In the next lesson, we’ll explore the categories of digital
measurement analysts can use to evaluate a brand’s
marketing efforts.
PART 3 | LESSON 2
 Digital Measurement

Five things discussed in this lesson:

Assessing the effect of digital marketing

programs across platforms has long been a challenge
for analysts, but recent regulatory changes have
introduced additional uncertainty

Tools still available to analysts, including

multitouch attribution models, marketing mix
models, and incrementality studies, can help brands
answer marketing effectiveness questions

The degree to which today’s uncertainty has

affected each category of measurement, as well as
each tool and provider, varies

Leading advertisers use the concept of

“Measurement Multiplicity” to clarify their decision-
making

A test-and-learn mentality is also critical for

marketing (and measurement) success

Think of the role measurement plays in marketing as sailing

with a compass. That compass can give analysts the most
efficient, expedient path to their destinations. Although the
compass’s dial may flicker from time to time, analysts can
trust and depend on its guidance. Unfortunately,
uncertainties introduced in the wake of regulation and data
privacy initiatives have smashed that compass. While it’s
still readable, it’s hard to know if its guidance is accurate.
Yet it remains – just as it has always been – easy to know
which way the wind is blowing.
The marketer’s instinctive ability to read the wind and tack
toward a destination has become more important than ever.
Regardless of the challenges in the digital environment, a
variety of measurement techniques can confirm wind
direction with great accuracy, even if their guidance toward
the best path forward must be taken with a grain of salt.
In this context, digital measurement minimizes uncertainty
in evaluating marketing performance and fuels instinctive
decision-making. Specifically, measurement does three
important things for a brand: ➔ Establishes a source of
truth – Ensure stakeholder trust in data reliability across
data sources to measure what matters by defining clear
customer value and key performance indicators.
➔ Allows for insight discovery – Understand consumer
behavior and the effect of marketing efforts in the context
of the customer journey.
➔ Enables the activation of insights – Allocate scarce
marketing resources optimally across channels, initiatives,
and partners.
Analysts’ measurement plays an important role in any
organization by enabling business growth through a
combination of understanding and accountability.

The Four Categories of Digital

Measurement
Digital measurement covers a wide variety of objectives,
platforms, and tactics, making it necessary to use a simple
framework of categorization. The framework presented in
Figure 3.1 classifies digital measurement in four categories:
Figure 3.1: The four categories of digital measurement

Making the Intangible Tangible: Brand

Impact
Companies can use several methods to value what has been
an inherently intangible asset: their brands. Each approach
differs in its perspective regarding value and the inputs in
its calculations, but each can provide valuable insights into
brand impact.
Assessing Attributes This subjective means of
assessment assigns values to attributes, such as
satisfaction, loyalty, awareness and market share,
that are tracked separately or weighted according to
industry. Advertising agency Young & Rubicam has
developed a “Brand Asset Valuator,” which is an
attribute assessment approach that’s based on
differentiation, relevance, esteem, and knowledge.
Other approaches exist, but the concept remains the
same. Such methods often use an assigned value,
rather than a measured value, and are subject to
challenges.
Brand Equity
The brand equity approach combines three elements:
effective market share, which is the sum of market shares in
all segments, weighted by each segment’s proportion of
total sales; relative price, which is a ratio of the price of
goods a brand sells, divided by the average price of
comparable goods in the market; and durability – the
percentage of customers who will buy that brand in the
following year. While thorough in its design, this data-
intensive approach relies heavily on modeling for its
calculation.
Brand Valuation
Brand valuation methods seek to take the most robust
financial data available to model a plausible valuation of a
brand. While the assumptions underlying these methods are
also subject to challenges, they at least strive to create an
objective-as-possible marker or view of a brand’s strength.
Algorithmic
Advertising holding company WPP performs an annual
valuation published as “The BrandZ Top 100 Most Valuable
Brands” report. This approach uses an advertiser’s financial
data, market dynamics, and an assessment of the brand’s
role in income generation to forecast brand value. Other
similar “blended” formulas are used to quantify elements
that most influence a brand’s strengths and risks. Though
certainly detailed, this method provides a once-per-year
snapshot of brand value due to the annual nature of
required data inputs.
‘Royalty Relief’
Brand Finance publishes its annual Global 500 study using a
“royalty relief” approach that calculates the net present
value of the hypothetical royalty payments an organization
would receive if it were to license its brand to a third party.
As a hypothetical approach, the assumptions underpinning a
royalty relief calculation are open to challenge. Yet the
system of royalty payments is well established so
assumptions can be rooted in practical experience and
applications.
‘Net Promoter Score’
A popular measure is “net promoter score,” or NPS, a metric
Fred Reichheld, Bain & Company, and Satmetrix developed.
Its power is its simplicity. It asks customers, “How likely are
you to recommend company/brand/product X to a
friend/colleague/relative?” and scores their responses from
zero to 10. “Promoters” give a nine or 10 score, “passives” a
seven or eight, and “detractors” a zero to six score. The NPS
score is the percentage of promoters less the percentage of
detractors and ranges from −100 to +100. Although largely
qualitative, this method is straightforward.
Each method for quantifying brand impact has strengths
and weaknesses that analysts should consider and manage
accordingly. When analysts apply these methods properly,
the insights can show analysts how their digital marketing
performance complements (or replaces) measurement of
more tangible assets, such as sales lift.

Measuring Sales Levers: Consumer

Outcomes
The desired outcome of nearly every marketing program in
the for-profit world is more sales for the brand. Indeed, the
result of increased sales is a diligently tracked consumer
outcome. Vast networks of point-of-sale solutions and
internal tracking systems exist for the express purpose of
tracking sales. Analysts can mine these outcome data for
insights and, frequently, provide greater context through
sophisticated measurement techniques, such as
incrementality studies (detailed later in this lesson).
Ever-present consumer outcomes stop short of an ultimate
sale, but are valuable, nonetheless. Such outcomes happen
along a consumer’s decision journey and the path toward a
sale and, therefore, should be measured. In this context, the
small presale steps (e.g., visiting an auto dealer website)
are called microconversions, while the final conversion (e.g.,
buying a car) is the macroconversion. Microconversion
outcomes occur in online and offline consumer behaviors.
Analysts can measure a plethora of microconversions in
consumers’ online interactions. As data are created online,
analysts can access and track the information with tools in
Google Analytics. Activities that could qualify as examples of
microconversions from consumer actions on a website
include: ➔ Viewing a page
➔ Watching a video
➔ Commenting on a blog post ➔ Sharing a post through
social media ➔ Creating an account ➔ Signing up for an
email newsletter ➔ Downloading an app
➔ Placing an item in an online shopping cart Analysts can
earn insights that will inform marketing strategies that are
designed to drive these microconversions by assigning
dollar values to each outcome. Typically, the assigned
values should be based on the likelihood that a consumer
fulfills a macroconversion after the microconversion (and, of
course, the calculated dollar value of that
macroconversion). Outcomes more highly correlated to a
macroconversion (e.g., placing an item in an online
shopping cart) should have a higher assigned value than
those that exhibit less correlation (e.g., viewing a webpage).
Measurable offline behaviors can also be thought of as
microconversions. Store visits, which measure consumers
who enter a defined physical location, are an important
microconversion outcome for most advertisers. This is
especially true for brands that interact with consumers
offline (e.g., retail shops, movie theaters, auto dealerships,
etc.) and build online relationships to influence behaviors in
the physical world. Although a store visit outcome is a step
short of a sale, advertisers that value store visits typically
see a high correlation between a consumer who visits a
location and then buys an item at that moment or in the
future.
The mobile location industry began as a way to customize
apps and target ads but has, in the words of The New York
Times’ Jennifer Valentino-DeVries and Natasha Singer,
“morphed into a data collection machine.” [80] As of Valentio-
DeVries and Singer’s report in late 2018, at least 75
companies receive precise location data from hundreds of
apps whose users enable location services for benefits such
as weather alerts. [81]
The New York Times revisited the topic of location tracking
in December 2019 when Stuart A. Thompson and Charlie
Warzel wrote an article that demonstrated how easily they
were able to obtain from a company a database of “50
billion location pings from the phones of more than 12
million Americans as they moved through several major
cities, including Washington, New York, San Francisco and
Los Angeles.” [82] Data such as this is at the heart of the
consumer privacy concerns discussed in the previous
lesson. Thompson and Warzel make this clear, as the issue
of consumer consent is central to their investigation.
Typically, companies collect, use, store and sell location
data to help advertisers, investment firms and others. Apps
most popular among data companies are those that offer
services keyed to people’s whereabouts – including weather,
transit and travel – because users are more likely to enable
location services on them.
Location data can be tremendously important to advertisers
who seek to optimize their marketing strategies (and, more
specifically, their media bidding strategies) to consumers’
locations. For example, my local frozen custard shop in
Chicago’s Roscoe Village may want to answer a “best ice
cream” query from someone searching in the neighborhood,
but chances are, the shop is less inclined to bid on a
consumer searching from Pittsburgh. Location tracking is
the data that enables such differentiated strategies.
Perhaps even more importantly, location data helps analysts
assess advertisements’ effectiveness. Online retail
advertisements should be assessed by the degree to which
they drive consumers to stores. Used in this way, location
data provides transparency for investing advertisers while
holding platforms, agencies, and others involved in media
sales accountable for ad performance.
Digital ad platforms, including Google and Facebook, offer
important store visit measurements to advertisers. As The
New York Times noted, however, a growing number of
independent third-party companies offer this data as well.
Analysts should carefully consider the use of location data
as a means of measuring marketing effectiveness, assuming
providers can ensure consent and privacy, and limit data
quality issues.

Understanding What Is Important:

Customer Value
Scott Kirby, formerly president of American Airlines, gave
Wall Street analysts a peek into the lopsided economics of
his airline on an October 25, 2015, conference call, saying
half of the company’s 2014 sales came from 87 percent of
customers who flew on the airline once. [83] What this
means, of course, is that a paltry 13 percent of travelers
contributed 50 percent of American Airlines’ annual
revenue.
The idea of differentiated customer value isn’t unique to the
airline industry. Every business earns different economic
value from different consumers. By understanding what
their valuable customers look like (and what they don’t
resemble), advertisers can affix differing levels of value to
nearly every consumer they could pursue in the market.
The key to unlocking this insight is understanding valuable
attributes for prospective customers. Advertisers can build
“best customer” profiles based on analyses of demographic,
psychographic, and behavioral data they collect. “Look-alike
analyses,” that is, identifying consumers who demonstrate
similar attributes, can reveal new high-value targeting
opportunities.
The importance of understanding valuable consumer
attributes goes beyond acquisition opportunities. Applying
relative values to each characteristic, attitude, and action
that comprises the value calculation can deepen customer
understanding while providing the foundation for marketing
strategies that target consumers, engage with messages,
and incentivize behaviors.

Assessing Marketing Effectiveness:

Attribution
Measuring lift on a marketing campaign seeks to isolate the
effect the campaign had on consumer behaviors. Online
behaviors are easier to measure than offline actions, but
online and offline conversions are important to lift
measurement. At its core, this pursuit seeks to answer
whether an intervention (e.g., the ad campaign) compelled
a consumer to act, or would the consumer have acted
anyway.
Well-designed studies can help answer this question, but
they don’t address an advertiser’s complete marketing
investment, which can be spread across a number of
campaigns and platforms. Indeed, one of an analyst’s most
critical challenges is measuring total marketing
effectiveness by determining which media investments
drove sales, calculating returns, and optimizing future
investments.
The rapid pace of change leaves analysts reeling from
reshaped consumer behaviors, data, and partnerships. And
yet, no single digital solution measures the full journey a
customer will take to evaluate brands and make an ultimate
purchase. Facebook, for example, remains the only solution
that rigorously measures consumer behavior on Facebook
pages, as the platform maintains an exclusive list of
partners that receive valuable Facebook data. But when
those consumers are on YouTube, Facebook is just as
unlikely to know their behaviors as Google is to collect
behavior data while consumers are on Facebook.
Analysts can plug gaps in understanding with highly tuned
and data-rich models, but the results are estimates of
consumer behavior. This situation leaves analysts in the
uncomfortable position of feeling tremendously empowered,
but unable to answer questions related to digital campaign
performance. Several solutions available today, however,
can help them piece together an assessment of marketing
effectiveness.
Multitouch Attribution Multitouch attribution models
(MTAs), also known as marketing attribution
solutions, promise to provide actionable insights by
collecting real-time customer behavior data. Through
sophisticated models based on collected data
samples, MTAs measure the relative benefit of
specific digital media to influence consumers. Where
actual data on consumer behavior exists, the models
are termed “deterministic,” while models that plug
gaps in consumer behavior data are termed
“probabilistic.” A blend of deterministic and
probabilistic models forms the inner workings of any
MTA solution.
Traditional MTA solutions attempt to measure consumer
interactions across the full spectrum of media channels
(e.g., from TV to radio to out-of-home to all digital ad
exposures). In contrast, digital platforms offer single media
channel attribution solutions to measure and optimize
consumer interactions exclusively on their own sites. Each
attribution solution faces challenges in today’s
measurement environment.
Figure 3.2: Comparison of single channel and cross-channel
attribution solutions

As with any measurement system that relies on third-party

cookies, MTAs are limited in their ability to measure and
reach users because of changes in privacy and browser
technology. These solutions will see reduced effectiveness in
the form of limited match rates or even sample biases for
surveys and testing. The lack of insight into consumer
behavior has forced MTAs to be more probabilistic in their
modeling, making it difficult to validate findings and base
decisions on MTA results.
Content providers’ collection of consumer behavior data on
their own digital properties has been largely unaffected by
moves to enhance privacy. As we’ve seen, these content
providers can rely on user sign-in as a mechanism to
connect behaviors to consumer profiles, rather than digital
cookies. As a result, the attribution solutions these providers
offer – single media channel attribution solutions – can be
seen as reliable predictors of how marketing efforts affect
consumers on those platforms. Advertisers can use
attribution solutions on their media platforms to help
optimize investment.
However, single media channel attribution solutions have
drawbacks that are important for analysts to understand: ➔
No cross-platform view – Single media channel
attribution solutions are good at helping advertisers
optimize media within the channel but have serious flaws
when it comes to evaluating holistic advertising efforts.
➔ Bias – Single media channel attribution solutions can only
see the media placed on that channel, and hence, no matter
which attribution calculation (last click, position-based, or
data-driven), these solutions are biased toward themselves,
taking more credit than they deserve.
➔ Data discrepancies – The data each channel collects
fluctuates constantly, as should be expected. As data
accumulates on platforms at different rates, certain media
types are more effective with a product type or audience,
etc. This necessitates fixed weighting to try to match a
single media channel attribution solution to a specific data
source. Otherwise, collected data that’s related to the
platform will be incorrect in different ways at different times
of the year, month, week, or even day.
➔ Varying methodologies – Every single media channel
solution has a different methodology and, thus, can’t be
used for apples-to-apples comparisons. For example, the
way view-through conversions (i.e., a conversion that
follows a consumer simply seeing – and not clicking on – a
digital ad) are valued in Google attribution solutions for
YouTube or display ads is different from how Facebook
attributes credit for view-through. This means that if an
analyst were to use the same approach for every digital
channel where they believe view-through has significant
value, some platforms could come out ahead unfairly based
on methodology versus performance.
Marketing Mix Modeling Marketing mix modeling
(MMM) is a time-tested method for measuring the
effect of marketing and media investments. Many
leading brands use MMMs to determine what’s
working across different channels. The MMM can be a
crucial tool for guiding budget decisions as a periodic
analysis to measure the effectiveness of each media
type and channel (offline and digital). Typically, MMM
studies are run quarterly, but a recent trend has been
for brands to increase the frequency of MMM, up to a
monthly cadence.
MMMs attempt to determine how media spend has affected
key performance indicators (KPIs) – such as brand
perception or product sales – by isolating the broad factors
that can influence consumer behavior to calculate a media
channel’s return on investment (ROI). This is a tremendous
undertaking, as everything from advertising to
recommendations from friends to the weather can affect
nearly any KPI an advertiser adopts as its metric for
success. As a result, MMMs are built upon vast amounts of
data and have been refined over long periods. Many can
trace their roots as far back as the 1960s.
Many MMMs struggle with the nuances of digital advertising
as they attempt to evaluate all channels of consumer
connection through a consistent currency (e.g.,
impressions). Search advertising, for example, poses a
particular conundrum for MMMs as the idea of a search ad
impression betrays the true value of the channel. Search
ads, after all, are to be clicked on and not simply seen.
Furthermore, most of the time, an MMM isn’t set up to
provide actionable insights when it comes to digital
channels. Many brands are stuck treating digital ads the
way they would traditional advertising platforms, including
television, with one ROI for TV, one for search, and another
for online video. On the surface, that might seem fair, but it
doesn’t capture the variety of ad formats and tactics
available in digital marketing.
MMM providers can improve their models by embracing the
nuances of digital measurement. To do so, analysts should
advocate for their MMM providers to follow the five steps
described below to get more accurate and actionable reads
on the effect of their digital efforts.
Figure 3.3: The five steps to more actionable MMM digital

reads
A more detailed look at these steps reveals their importance
to and effect on actionable measurement insights: ➔
Evaluate media by geography and market – Big
television buys may run nationally, but digital ads deliver on
demand locally and offer granular reporting. Collect and
model the data by market to get additional data points that
make the model more representative of reality. Jeff Shatz,
vice president of marketing effectiveness at Nielsen,
explains why this is critical: “The more granular the data,
the more variability the statistical model is able to pick up,
and the greater ability to tease out true drivers. If only
national-level marketing data is used, the model will not be
able to account for critical market-level influences that
impact whether a purchase is made.” [84]
➔ Differentiate video platforms – Ditch the generic
digital buckets that treat all online video impressions
equally. Instead, break out video by platform, so the model
provides an individual read on each. Qualities like watch
time, audibility, and viewability vary widely across video
platforms and, depending on what you’re trying to achieve,
will have a different effect on the effectiveness of media.
Recent research from Nielsen supports this, finding that
when advertisers’ MMMs evaluated video platforms
independently rather than aggregated, return on ad spend
varied by as much as 48 percent. [85]
➔ Consider the various elements of your media plan –
Assess different elements of the media plan by channel,
including ad formats, audience segments, and the
campaign’s reach and frequency. From there, the analyst
can find stronger connections between online efforts and
offline sales. The Hershey Co. marketing team does this by
asking media partners and their media agencies to supply
data directly to their marketing mix modeling measurement
partner, helping ensure the data is accurately broken out by
brand and ad format. On YouTube, this level of granularity
helped the team establish that a specific mix of YouTube
TrueView, Google Preferred, and six-second bumper ads was
most effective in driving sales for Reese’s Peanut Butter
Cups. As measured by marketing mix modeling, the retail
ROI for The Hershey Co.’s portfolio of brands increased 40
percent year over year from 2017 to 2018. [86]
➔ Validate model outputs through experiments –
MMMs are inherently complex models – and they’re not
perfect. Before making major changes to the media strategy
based on marketing mix modeling results, run sales lift tests
or other isolated experiments to test the effect of a single
change in strategy. For example, Frito-Lay North America, a
division of PepsiCo, uses MMM, along with sales lift and
brand lift studies, so it can evaluate three points of
measurement before changing its media strategy. The
results of all three methodologies won’t always align, but
this validation helps guide future media planning and
justifies increased investments in a platform or strategy
over time. [87]
➔ Test your creative tactics – Telling the brand story on
digital is possible in infinite ways. That’s why the savviest
advertisers remove the guesswork from these decisions by
using marketing mix modeling to measure the effectiveness
of their creative. Frito-Lay North America uses MMM –
complemented by isolated experiments – to test the effect
of the more personalized creative it developed for YouTube.
For example, as its marketing team creates customized
videos at scale, they’re seeing indications in their MMM
results that more personalized creative is driving higher
incremental sales than a message designed to have broad
appeal. “We want to provide consumers with the most
relevant content based on what makes them tick,” says
James Clarke, senior director of portfolio media, analytics
and customer relationship management at PepsiCo. Clarke
added that as the company measured the effect of creative
elements on sales, surprises arose. “Sometimes, creative
elements we thought would be meaningful don’t have an
impact; other times, things that seemed trivial really move
the needle.” [88]
With the flexibility and customization that digital marketing
offers, traditional measurement methods must evolve.
Setting up the right data inputs in an MMM at the outset of
campaigns means more actionable measurement in the
output.

Getting to True Lift: Incrementality Testing

Incrementality tests are on-demand experiments that
measure the incremental effect of a specific campaign or
tactic, as needed. Most often, incrementality testing is done
using traditional test and control designs and demonstrates
benefits by comparing observed results to those one would
expect if the stimuli being tested (e.g., a digital
advertisement) hadn’t happened. A well-designed
incrementality test will assess any Marketing Objective
effectively, whether the objective is related to brand impact,
consumer outcomes, customer value, or attribution.
Incrementality tests fall into two categories: experimental
and observational. According to Facebook’s guide for
implementing incrementality tests, these approaches differ
in important ways, including their requirements for upfront
resources and the robustness of the results they produce.
➔ Observational – Begin with an existing set of data that
resulted from exposing people to a certain ad or ad variable,
and then apply a model or statistics to estimate how much
value a treatment may have had. Common methods involve
using synthetic experiments to attempt to replicate a real
experiment by “finding” a control group within a group of
people who were not exposed to the ad or ad variable the
analyst is trying to evaluate. For example, one could
evaluate the effect of a technical issue that only affected
some users by finding a “similar” group of people who were
unaffected. This method doesn’t require upfront work, but it
may be less accurate and subject to bias on unknown
factors. It also requires advanced methods and support from
data scientists later in the process. [89]
➔ Experimental – Begin by developing a hypothesis about
the effect a change in strategy will have. Next, designate a
group (or groups) of people who will be exposed to the
treatment, and a control group that won’t. By isolating the
exposure of a variable, such as creative or audience, and
then comparing it to the control group, an analyst can
understand the true incremental value of the strategy. The
quality of experiments may vary, but they’re still the ideal
and most accurate way to measure incrementality. True
experiments are often the benchmark for other
methodologies. Despite their benefits, experiments require
upfront setup, as well as the opportunity cost of withholding
treatment from the control group. [90]
To determine the most effective way to collect evidence in
the assessment of interventions, we can turn to a field with
a long history of testing: the medical sciences. Clinical trials
that assess the effects of treatment strategies and
pharmaceutical products are a hallmark of modern
medicine. In much the same way marketing incrementality
testing seeks to identify the effect of an intervening
advertisement on consumer behavior, clinical trials in the
medical sciences attempt to isolate the effect of medical
interventions on health and well-being.
In assessing various methods for collecting evidence,
pediatric gastroenterologist Dr. A K Akobeng found a class of
experimental incrementality tests, known as randomized
controlled trials (RCTs), to be the gold standard for
evaluating the effectiveness of interventions. [91]
Figure 3.4: Akobeng’s Hierarchy of Evidence

An RCT is a type of study in which participants are randomly

assigned to one of two groups (most commonly called the
“treatment” and “control” groups). RCTs that use large
geographic locations to assign participants to treatment and
control groups discourage potential hidden biases by
ensuring that fundamental, yet unknown, differences
between samples are balanced among treatment and
control groups. After a representative sample of the
population of interest is randomly allocated to one or
another group, the two groups’ behaviors are observed in an
identical manner for a specified period of time called “the
pre-test period.”
During the pre-test period, a factor can be calculated that
equates the treatment group’s behavior to that of the
control group before the intervention. Once the intervention
has been introduced (i.e., the “test period”), this factor can
then be applied to the observed control group behavior to
estimate how the treatment group should be expected to
behave. This expected behavior pattern from the treatment
group is called “the counterfactual.” Truly measuring the
incremental effect of the intervention means comparing the
observed behavior of the treatment group to the
counterfactual during the Test Period – not the observed
behavior of the control group during that time.
Figure 3.5: Proper incrementality test design

Although incrementality tests can be time- and resource-

intensive, they remain the most accurate way to measure
marketing’s effect.

A Word on Clarity through Measurement

Multiplicity
As we have seen, the present state of digital marketing
measurement is fraught with uncertainty. The vastness of
consumer behaviors online presents an overwhelming
number of important behaviors that should be
conscientiously observed. Data limitations have added
complexity to measurement tools that are already
excessively sophisticated. Traditional measurement
techniques that struggled to evaluate digital media
channels accurately still have this problem.
Given this predicament, no single measurement tool or
technique should be viewed as gospel. Instead, a
combination of tools applied concurrently or in a planned
cadence clarify a brand’s performance and market
environment. This is a practice I call “Measurement
Multiplicity.” Today, leading advertisers apply this idea by
cross-checking brand impact and customer lifetime value
measures with single media channel attribution reads or
supplementing quarterly MMM reports with incrementality
tests run as campaigns launch. In the case of the most
sophisticated advertisers, all available measurement
techniques described in this book are employed, each
providing context and validation of the next.
This approach requires time and considerable investment of
resources. Ensuring that such a robust measurement
program is done efficiently requires the adoption of
something else: a test-and-learn attitude. The test-and-learn
attitude accepts risk but mitigates its effect by allowing for
programs to “fail fast,” and seeks to improve a brand’s
marketing efforts by constantly optimizing targeting,
messages, and other important elements. This attitude
enables the ability to balance a number of inputs and
construct one clear view. Many of the underpinnings of the
test-and-learn attitude depend on the company’s level of
digital marketing maturity, described in Lesson 4.
It’s important for analysts to adopt the practice of
Measurement Multiplicity and a test-and-learn attitude. In
doing so, analysts will ensure the most effective and
efficient approach to measurement.

A Word on Digital Measurement

Challenges
Digital measurement is a challenging endeavor. From the
need for tedious precision in study design (and
implementation) to the trial of accounting for inevitable
irregularities in collected data, digital measurement poses
several problems for the analyst. Obvious issues (e.g., a
poorly designed study, bias in the data, etc.) sometimes
produce the flickering dial of the marketer’s measurement
compass described at the beginning of this lesson. Such
glaring imperfections can be easily rectified.
More frequently, however, intrinsic challenges to measuring
digital activity complicate the analyst’s work. For example,
in the previous lesson, we learned about the use of digital
cookies to track consumer behavior. Cookies operate at a
computer level, however, and a number of people can use
any given computer (think of the shared family desktop
found in many households). Consumers also frequently
access websites from multiple devices: their work desktops
one day, their personal laptops on the next, and their
mobile phones during the time between. In addition, as
consumer identification data passes between a brand and
its marketing partners, the degree to which consumer IDs
are linked across different datasets to the same person (i.e.,
the “match rate”) affects the accuracy of the data and the
brands’ ability to provide consumers with personalized
experiences.
Multiple people using the same computer profile, cross-
device measurement, and cookie matching are some of the
challenges that obscure the clear and consistent picture of
an individual’s digital behavior. Analysts can account for
these challenges by clearly understanding their
measurement objectives and the effects such challenges
have on the data they collect. In this way, analysts can offer
the grain of salt along with insights to bring more accuracy
to the marketer’s instinctive decisions.

In Summary: Digital Measurement

Assessing the effect of digital marketing programs across
platforms has long been a challenge for analysts; however,
many measurement techniques provide data and insights to
guide marketers’ decision-making. Tools, including
multitouch attribution models, marketing mix models, and
incrementality studies, can help brands answer questions
regarding marketing effectiveness. While regulatory shifts
and privacy initiatives have affected each category of
measurement, the effect on precision varies from category
to category. Leading advertisers use a combination of
measurement techniques and a test-and-learn mentality to
bring clarity to their decision-making, a practice and
attitude that analysts should adopt. Finally, analysts must
account for the inherent (and obvious) challenges posed to
digital measurement to ensure the marketer’s decision-
making process is rooted in accuracy.
In the next lesson, we’ll explore the data analytics and data
visualization tools that analysts can use to help them find
the answers brands need.
PART 3 | LESSON 3
 Analytics and Dataviz Tools

Five things discussed in this lesson:

The market for data analytics and visualization

tools is a fast-moving field of established providers
and new entrants

Adopting a way to think broadly about the

market is more important than being up to speed on
the development of new tools

Two evaluation criteria – data flexibility and ease

of use – provide a powerful framework for the
evolving marketplace

Data flexibility and ease of use make it clear

which tools analysts should keep in their toolboxes
and which they should discard

No tool, regardless of its cost, is as important as

the analyst who uses it

Tools play an important part in the analyst’s work. They aid

in data collection, cleanse data of errors, and allow for
sophisticated analysis techniques. They can scale insights
across companies large and small. In a very real sense, the
analyst is enabled by the tools they use.
The market for digital analysis tools is expanding quickly
and changing constantly. The growing need for analytics
solutions, rampant acquisition activity, and the low cost of
tool design have produced a fast-moving field that features
established companies and entrepreneurial startups. A quick
web search reveals page after page of “Top 10 Analytics
Tools” lists, showing the best resources selected from what
must number in the thousands, if not tens of thousands, of
tools.

Key Categories of Digital Analysis Tools

In the vastness of the digital marketing analytics tool
market, analysts will find a few critical categories that are
particularly important. Tools in these categories are required
to unlock insights that are unique to consumers’ online
behavior.
Social Listening Social listening tools connect to
various social media networks to extract consumer
data. These tools provide direct access to content
consumers create and allow marketers to learn about
interests, actions (e.g., likes, favorites, etc.), and
thoughts in consumers’ words.
Some social listening tools, such as Brandwatch
(brandwatch.com , formerly Crimson Hexagon), Sysomos
(sysomos.com ), and Hootsuite (hootsuite.com ), connect to
multiple platforms at the same time. In addition, analysts
can use several free tools to gain insight into how
consumers interact with a brand on social networks. For
example, TweetReach (tweetreach.com ) is a free tool for
analyzing Twitter hashtags, user accounts, and other
activity. Tools that include BrandMentions’ free real-time
social tracking tool, SocialMention
(brandmentions.com/socialmention ), collect publicly
available information about a brand across social media
posts, website blogs, news, videos, articles, and other
content and are particularly valuable tools when conducting
SOV analysis. Also, social media networks often provide
detailed analytics about who’s interacting with content
through native social listening tools, including Facebook’s
analysis tools.
Content Analysis Content analysis tools help analysts use
collected data in meaningful ways by studying digital text,
photos, audio and visual formats of communication. These
tools can reveal otherwise unrecognizable patterns in data.
Term relevance (i.e., the relative number of times a term is
used in a body of text data) and consumer sentiment (i.e.,
the tone of consumer mentions of a company, brand, or
product typically categorized as “positive,” “neutral” and
“negative”) are two of the more popular and useful types of
content analysis. [92]
Many free tools, including the R package Quanteda, provide
quantitative text analysis by analyzing keywords,
representing text visually, applying sentiment analysis
frameworks, and more. Commercial tools, such as Linguistic
Inquiry and Word Count (LIWC), interpret text to reveal
thoughts, attitudes, feelings, personality, and motivations of
the author. Another tool, BuzzSumo (buzzsumo.com ),
provides insights into the types of content that resonate
with specific audience groups and can be a valuable tool for
targeting content and competitive analysis.
Search Trends Internet search trend data reveals the
issues that are on consumers’ minds. Search analysis
tools aggregate and visualize that data to show
analysts the popularity of specific topics, which can
offer insights on demand for new products, consumer
response to marketing campaigns, and brand
awareness.
Google Trends (trends.google.com ) provides free access to
indexed search volume from Google.com, Google Images,
Google Shopping, and YouTube. Using Google Trends,
analysts can collect data about topic and search term
interest over time (ranging from the last hour to 2004) at
various geographic levels (ranging from worldwide to a
specific city) and download the data into raw files for further
analysis. Google Trends is a particularly valuable tool when
conducting SOS analysis. Bing and Yahoo offer services
similar to Google Trends to analyze search behavior on their
properties.
Several other free tools – including Searchvolume.io
(searchvolume.io ) and a search analysis tool found on The
New York Times website (nytimes.com/search ) – offer
simple keyword research through online interfaces, while
commercial services like Moz ( moz.com ) analyze search
volume, report metrics and suggest actions.
Website Analytics Website analytics tools provide
information about visitors to a website, including the
number of visitors and how they behave on the site.
These tools let the analyst gauge traffic and the
popularity of content on the site, which is useful for
market research. Say the analyst wants to improve a
website’s design. Clickstream analysis – tracking how
visitors interact with the site – can reveal which
content visitors believe is most valuable. Metrics,
including the number of times people visited the site
(“visits”), the specific pages people view (“unique
page views”), and the percentage of people who left
the site from the first page they visited (“bounce
rate”) offer valuable insights.
Large, powerful, paid tools, such as Adobe Analytics and
Google Analytics 360, can handle this challenge (along with
a thousand other challenges) for you. But many lighter-
weight tools can perform expert clickstream analysis and
visualization at a fraction of the cost of premium platforms.
For example, Google’s free version of Google Analytics
provides many of the capabilities found in the premium tool,
including clickstream analysis. The disadvantage of these
lighter-weight tools is they can handle less data and have
more restrictions than premium website analytics tools.
A/B Testing
A/B testing, also known as “conversion rate optimization”
(CRO), is another important category of analysis for the
analyst. A/B testing measures the effectiveness of ads,
website design, and other forms of digital content by
producing two versions of the item being tested, presenting
those versions to consumers, and tracking engagement
rates. In doing so, A/B testing helps the analyst determine
which version is more effective overall and for specific
consumer segments.
Online A/B testing is efficient and effective due to the scale
of the internet. A tool, such as Google Optimize
(optimize.withgoogle.com ), offers nearly everything the
analyst needs for free. Other premium tools, such as
Optimizely (optimizely.com ), Visual Website Optimizer
(vwo.com ), and Evergage (evergage.com ), offer a full set
of deep capabilities. In addition, the free and paid versions
of most website analytics tools offer A/B test and
optimization modules.
Creative Optimization Related to A/B testing is the
category of analysis designed to help advertisers
optimize the creative effect of their advertising.
Creative optimization seeks to maximize an
advertisement’s effect by using analysis to determine
the most effective use of the ad’s various elements.
Those elements can include tone, the use of color,
the use of objects, casting, music, and many others.
While creative optimization’s objective fits with any
type of digital advertising, it’s particularly effective in
optimizing digital video ads.
Analysts can use a wide range of measurement tools,
testing techniques, and technologies to conduct creative
optimization. A/B testing optimizes some basic elements of
an ad, such as color or even casting. More complex
elements require more sophisticated approaches to
optimization. One such technique is the use of retention
curves for videos. Retention curves visualize the percent of
viewers exposed to an ad who are still engaged at each
subsequent second through to the ad’s completion. Drastic
drops in retention (i.e., any point when a large percentage
of viewers leave or skip the ad) can reveal elements of the
ad that led viewers to disengage. Analyzing retention curves
by consumer segments can reveal how elements of the ad
appeal to each segment. While retaining all viewers for the
ad is an unrealistic goal, analysis of retention curves can
help configure ads that have the strongest consumer
appeal.
Machine learning techniques can also help to optimize ads.
Algorithms can analyze nearly any conceivable element of
an ad and correlate that element’s effect on any
measurable form of consumer behavior. Algorithms can be
trained to detect broad sets of categories within a video
frame, ranging from logos to animals to the way a room is
decorated to visual effects. This allows for optimization of
viewer retention on a massive scale that’s much larger than
the hand-spun analysis an analyst can perform using
retention curves. Studies conducted in lab settings that
collect dense consumer neurological response data, such as
eye movements and brain activity, are an important source
of inputs to machine learning algorithms. For a broader
discussion of the effect of neuroscience on analysts, see “A
Word on the Effect of Neuroscience” in Part 5, Lesson 1 of
this book. These data can lead to optimizations that would
be imperceptible to the analyst who’s unaided by advanced
data collection techniques and machine learning.
Perhaps more impressively, machine learning algorithms
can analyze ads so deeply that they transcend the “this-or-
that” insights offered by A/B testing to reveal patterns in
storylines and other artistic characteristics of an ad. In
addition to analyzing image attributes in the video frame,
machine learning algorithms can detect elements of stage
direction, including movement, camera angles, and image
cropping.
“Data-driven creative for me no longer means switch this
pair of shoes for that one, update that price, or insert logo
name here,” says Andrew Shebbeare, co-founder and
chairman of Essence. “We can do so much more with this
kind of technology to bring products to life in ways that are
more human. And, in fact, advertising can be more human
when it is more data-driven.” [93]
While some tools found in these categories have enjoyed
long tenures, many more lasted much less time or changed
names and/or directions because of mergers and
acquisitions in the digital analytics industry. For example,
Klout was once the gold standard for measuring social
influence but shuttered in 2018 following the influx of
simpler and more methodologically sound online tools.
Because the digital analysis tool market shifts so rapidly, it’s
more important for us to characterize and segment this
market than it is to stay on top of the newest and “hottest”
tools. I like to do this by thinking about tools in terms of two
variables that can be posed as questions: (1) How much can
I do with this tool (i.e., data flexibility)?; and (2) How easy is
it for me to use the tool (i.e., ease of use)?

Evaluation Criteria 1: Data Flexibility

When we look at the first question, we see clear
demarcation in the tool marketplace. In fact, three distinct
categories of tools emerge: ➔ “Enterprise platforms ,”
which are big, powerful solutions that handle lots and lots of
data and are packed with capabilities; ➔ “Point solutions
,” which typically center on one primary capability, where
they go very deep; and ➔ What I call “analysis gadgets ,”
which offer a single, well-defined capability and sit on the
low-data flexibility side of the scale.
Let’s spend a little time talking about each tool category.
Enterprise Platforms (High Data Flexibility) The brand
names behind today’s enterprise platforms are easily
recognizable. These tools come from some of the
most well-known names in marketing analytics.
Companies, such as comScore, Adobe, IBM, and
Google offer premium tools that feature in-depth
analysis and visualization capabilities. In fact, these
tools give analysts as many capabilities as they could
possibly want. The downside to enterprise platforms
is their large price tags. [94]
It can be a huge benefit if an analyst works for an
organization that has an enterprise platform. The analyst
will have access to a significant amount of information and –
as advertisers that can invest in enterprise platforms
typically operate large, complex business models – the
ability to handle inputs from a variety of data sources.
Benefits aside, not every business has the wherewithal or
the need to invest up to six figures a year in an enterprise
platform solution.
A number of popular free tools – including R and Python –
are included in the enterprise platform solution category as
they offer a broad set of capabilities, placing them high on
the data flexibility scale. R and Python are open-source
tools, with thousands of people contributing to their design
by creating help files, building new capabilities, and offering
their experience to the user community. Unlike the premium
solutions found in this category, however, these free tools
lack the slickly packaged and fully integrated feel of
premium solutions. This makes them more difficult to
navigate and means they may not fit well in every business
situation. Still, analysts should become familiar with R
and/or Python (see “A Word on R and Python” later in this
lesson for insight into the R versus Python debate), given
their popularity and low cost to implement.
Point Solutions (Medium Data Flexibility) Point
solutions offer more affordable, but less robust,
solutions to analytics problems. If analysts have a
single objective or challenge to crack, a point solution
can be a viable option. For all analysis and
visualization challenges analysts face, they’ll find
myriad companies that offer valuable solutions.
Examples of point solutions include online survey platforms
such as iPerceptions (iperceptions.com ), SurveyMonkey
(surveymonkey.com ), ForeSee Results (foresee.com ), and
Google’s Consumer Surveys (surveys.google.com ) – that
offer insights through “voice of the consumer” data. [95]
These tools allow analysts to design surveys, collect
responses from consumers as they browse the internet
(typically, so they can get access to premium content), and
analyze results through visualization modules. Promising
results in as little as three days – compared to months for
traditional surveys – online survey tools can provide
consumer insights that help drive important business
decisions. Their scope is limited to digital surveys, but they
do that job exceptionally well with a deep set of capabilities.
Analysis Gadgets (Low Data Flexibility) All challenges
don’t require the depth of capabilities featured in
enterprise platforms and point solutions. Low- and
no-cost options that I call analysis gadgets are widely
available and fit a wide variety of analytics needs.
The analysis gadget market is an exciting, undulating
collection of quick-to-market tools born from the minds of
experimenting analysts and coders, and community-based
open-source solutions. The one characteristic that unites
them under the banner of analysis gadgets is that they’re
extremely lightweight, and the amount of data they can
handle pales in comparison to that absorbed by point
solutions and enterprise platforms.
These analysis gadgets are available online and usually
don’t need users to download anything onto their desktops.
Some of these tools require users to register and sign in
with email subscriptions, while others require nothing. Some
analysis gadgets include a request for voluntary
contributions to help defray associated hosting or
technology licensing costs while others are free.
An example of an analysis gadget is Wordle (wordle.net ).
When I want to run an analysis of text-based data and need
a word cloud, I head to Wordle.net. Wordle works perfectly,
making easy-to-generate and beautiful word clouds that lift
insights from opaque blocks of words quickly and effectively.
While Wordle doesn’t do anything more than create simple
word clouds, it does the job with such ease, it’s become my
go-to tool.
Figure 3.6: Word cloud based on discussion topics presented
at 2013’s “SXSW” digital conference created using Wordle

Not every analysis gadget is created by a single developer

or even a small company. Google’s Google Trends, as well as
the search analysis tools from Bing and Yahoo, fit in this
category. Cision, the owner of the analysis gadget
TweetReach, is a publicly traded company with annual
revenue nearing $1 billion today.
New analysis gadgets are popping up all the time. A simple
web search is an easy way to find them. Searching on the
term “free digital analysis tools” returns a long list of tools
that analysts can use for free or a nominal fee and start
collecting valuable digital data instantly.
Evaluation Criteria 2: Ease of Use
When we address the second question, “How easy is it for
me to use that tool?” the market begins to separate into
groups of tools we should use for the majority of our
analytics challenges and those tools we should discard. The
answer to the question of ease of use, of course, is a
personal one. A tool that comes naturally to me might not
be easy for another analyst. In addition, an analyst’s skills
will certainly improve, and with training, a tool that’s
difficult to use today may become easier tomorrow. As a
result, how an analyst evaluates tools through this lens
should be thought of as a uniquely personal snapshot in
time.
Regardless of the temporal and shifting nature of these
evaluation criteria, putting together the two questions we’ve
discussed allows for a very useful framework with which to
view the data analysis and visualization tool market. To
illustrate the usefulness of this approach, I’ll offer how I
would look at the marketplace today (I’ve included only free,
publicly available tools in my evaluation). Four quadrants
emerge in this framework. Which quadrant a tool resides in
will become important to the role it plays in our practice.
Figure 3.7: My personal 2X2 assessment of the data analysis
and visualization tool market [96]
Upper Left Quadrant: Tools to Collect Tools in the
upper left quadrant have limited capabilities but are
easy to use. They perform one-off jobs or specific
tasks quickly and efficiently. Analysts should keep
these analysis gadgets in their toolboxes. In fact, the
more easy-to-use tools analysts can master – and,
given the inherently low level of complexity,
“mastering” tools in this quadrant is often not a
challenge – the broader and more flexible their skills
will be.
My recommendation is to get hands-on with these tools.
When analysts find a tool that comes naturally, they should
make it their go-to, with the assurance they won’t need to
keep up on certifications or training. After all, the benefit of
an easy-to-use analysis gadget is akin to learning how to
ride a bike: it doesn’t matter how long it’s been since you’ve
done it – you can pop right on and hit the road.
Upper Right Quadrant: Your Go-to Tool(s) On the other
side of that ledger, in the upper right quadrant, are
powerful and easy-to-use tools. It makes sense for an
analyst to settle on one (or possibly two) here and
use it as the foundation for data analysis and
visualization. The tool(s) analysts choose from this
quadrant will do just about everything they need
easily and efficiently. This should become an
analyst’s go-to tool for the majority of data analysis
and visualization tasks.
I place Tableau in this quadrant. It’s intuitive, proven, and
powerful. In fact, Tableau has, in many ways, become table
stakes for most data analysts in their role as Data
Designers. I use the tools in my upper right quadrant to
complete about 75 percent of my visualization tasks.
Lower Right Quadrant: Your Secret Power Tool(s) In
the lower right quadrant, we find tremendously
powerful tools that have deep capabilities and
potential. The downside is that these tools are a tad
awkward. Learning them will take a bit more effort
and practice. Mastering them can require significant
time. Indeed, their power comes with the price of
being challenging to use.
Analysts shouldn’t avoid these tools, however. In fact, using
a tool from this quadrant is extremely important. The
explosive creation of data gives analysts opportunities to
collect, analyze, and visualize data like never before.
Realizing these opportunities often requires tools powerful
enough to move through large amounts of data. To do that,
a little more sophistication (which often materializes in the
form of a sophisticated tool) is required. One of the tools I
love most in my lower right quadrant is R. Of course, several
other tools in the “High data flexibility/Hard to use”
quadrant can serve many purposes as well as R can and
even surpass some of the things it can do. But R has
become my tool of choice.
The time required to master tools in the lower right
quadrant and their overlapping capabilities makes it
impractical to learn them all. A better approach is to focus
on a single tool in this quadrant. Analysts should master
that tool to complete the sophisticated, data-dense tasks
that their go-to tool in the upper right quadrant struggles
with.
Lower Left Quadrant: Tools to Avoid Analysts should
avoid any tool that’s hard to use and doesn’t have
strong capabilities. Given the plethora of credible,
capable tools that can fit in the other quadrants,
there’s no reason to waste time on difficult tools.
This approach to evaluating data analysis and visualization
tools gives analysts a helpful way to think about the tool
marketplace. As the marketplace evolves, so must this
evaluation. Think of the quadrant exercise as a view into
where the analyst is at the time of the evaluation. Adding
new tools to the framework is key. Analysts should shift
tools from the “High data flexibility/Hard to use” quadrant to
the “High data flexibility/Easy to use” quadrant as their
skills progress. Most importantly, analysts should revisit
their evaluations periodically to ensure they’re using data
analysis and visualization tools efficiently.
A Word on SQL
Structured query language (SQL) is a programming
language designed for managing data held in relational
databases. Donald D. Chamberlin and Raymond Boyce
developed SQL in the 1970s for manipulating, storing, and
retrieving structured data from IBM’s original relational
database management system (RDBMS) called System R.
Today, it’s the most popular language the analyst can use to
interact with a relational database. For the analyst,
collecting and analyzing data at companies that have a
relationship database (including MySQL, a free and open-
source RDBMS designed by Oracle) requires some use of
SQL.
Successful analysts will make it a goal to become very
familiar with SQL as they build their Techie skills. In fact, the
analyst community largely regards SQL skills as “table
stakes,” given the programming language’s ubiquity and
importance.

A Word on R and Python

While the analyst community agrees on the criticality of
SQL, there is less consensus on whether the analyst should
complement SQL with R or Python (or both). Both R and
Python are open-source tools for data analytics that handle
very large datasets. The two have important differences.
These differences can help the analyst determine which is
the right choice.
Data scientists made R for data-oriented projects. It features
a large number of ready-made packages and has built-in
ways to visualize data. In addition, R boasts a very large
community that provides support through mailing lists,
documentation, and blogs. R’s learning curve, however, is
steeper than many languages (including Python), and it’s
less efficient for general computations.
Python has a growing user community that over indexes in
software engineers and programmers. It provides more
opportunities to take advantage of artificial intelligence,
allows analysts to integrate data analysis with website and
mobile apps more efficiently, and can be adapted more
easily for programming tasks besides analyzing data.
Python, however, is less efficient for statistical
computations, features less-appealing data visualization,
and comes with fewer add-on modules and packages.
From my perspective, analysts can boost their productivity
with either tool as each is powerful enough to complete
nearly any analytics task. Learning the tools and
maintaining skills in them can be considerable tasks, so
choosing one or the other is more prudent than trying to
learn both. The choice between R and Python comes down
to personal preference (or the preference of your working
group).
I found that RStudio (R’s integrated development
environment) made programming in R so easy that it
quickly became my favorite. I lacked a programming
background, and Python didn’t have a comparable
development environment when I was beginning to learn R.
As a result, the choice was simple for me. When making the
choice for yourself, consider the pros and cons of each to
find the tool that best fits your needs and abilities.

A Word on Tools < The Analyst

One final thought on tools.
While tools are vital to our work, it’s important to keep the
proper context on that value. As Avinash Kaushik summed
up perfectly, “No tool would be useful unless you had a
Michelle or an Amir or Enrique or Sasha who understands
your business and has the drive to use the tool intelligently
to deliver actionable insights.” [97] A tool won’t solve an
advertiser’s challenges. When used properly, a tool will do a
great job of showing patterns in data that would be
undetectable otherwise.
Yet it is you, the analyst, who must first look for and then
properly interpret those patterns. Never underestimate your
importance and value.

In Summary: Analytics and Dataviz Tools

Data analytics and visualization tools live in a fast-moving
marketplace of established providers and new entrants. As a
result, it’s more important for analysts to adopt a way to
think about the market than be up to speed in the
development of new tools. Two evaluation criteria – data
flexibility and ease of use – provide a powerful framework
for the evolving marketplace and make clear which tools
analysts should keep in their toolboxes and those they
should discard. Remember, no tool is as important as the
analyst who uses it.
In the next lesson, we’ll discuss the importance of an
advertiser’s digital marketing maturity.
PART 3 | LESSON 4
 Digital Marketing Maturity

Five things discussed in this lesson:

Digital marketing maturity is a quantified

measure of an advertiser’s sophistication regarding
digital data and its use

Boston Consulting Group’s view of maturity is

based on a robust assessment across a broad array
of digital capabilities

Bain & Company’s view of maturity focuses on

how savvy an advertiser is in its digital measurement

Deloitte Consulting and MIT conducted research

that found advertisers fall into one of three levels of
maturity

Regardless of the approach to measuring

maturity, advertisers found to be more mature saw
benefits in sales growth and cost efficiency

The number of touch points along the consumer journey has

increased exponentially over the last decade. The tools
available to analysts, as we just discussed in the previous
lesson, have grown in much the same trajectory as more
data created more need for sophisticated analytics and
visualization tools. Meanwhile, have the relationships brands
maintain with consumers changed?
Some areas of the marketplace have undergone astonishing
shifts. An example is how Netflix disrupted the home
entertainment market and continues to use technology to
optimize its efforts. In contrast, other businesses lag and
have yet to realize the full value that digital data can
contribute to their marketing. What separates Netflix from
the laggards, regardless of their business category or the
degree to which their operations depend upon the internet,
is digital marketing maturity.
BCG’s View of Digital Marketing Maturity
According to Boston Consulting Group (BCG), the path to
data-driven marketing maturity is comprised of four stages:
nascent, emerging, connected, and multimoment. BCG’s
research indicates that as of 2018, only a handful of
advertisers – just 2 percent – operate at the most mature
levels by connecting with consumers at multiple moments
across the purchase journey through personalized content.
[98]

Figure 3.8: BCG’s assessment of marketing maturity’s

current state

BCG defined the four phases of digital maturity as: ➔

Nascent – Marketing campaigns are executed, mainly using
external data and direct buys, with limited link to sales ➔
Emerging – Some use of owned data in automated buying
with single-channel optimization and testing ➔ Connected
– Data integrated and activated across channels with
demonstrated link to ROI or sales proxies ➔ Multimoment –
Dynamic execution across multiple channels optimized
toward individual customer business outcomes and
transactions To arrive at its point of view on digital maturity,
BCG examined success factors and capabilities of more than
40 European advertisers across eight industries –
automotive, retail, financial services, travel, consumer
goods, technology, entertainment and media, and fashion
and luxury. The study revealed that digitally mature
businesses share a number of success factors. From an
organizational standpoint, BCG found that successful
advertisers invested in specialist skills, strategic
partnerships, agile teams, and fail-fast cultures. In addition,
technology at these advertisers are integrated and
automated, ensuring the use of connected data and
actionable measurement.

Bain’s View of Digital Marketing Maturity

Bain & Company’s view of digital marketing maturity
centered firmly on measurement. Citing opportunities
created by the collection of customer information and the
technology available to act on it, Bain found measurement
is more important than ever. In collaboration with Google,
Bain assessed the measurement maturity of 600 advertisers
in the U.S., UK, and Canada, and sorted them into four
levels on a maturity curve, ranging from “Foundational” to
“Best in Class.”
Figure 3.9: Bain’s perspective on measurement maturity

Bain’s look at the digital measurement practices of

advertisers found organizations that have realized
measurement advantage did three things better than their
competitors: [99]
➔ Measurement – They had a deep understanding of their
customers and linked marketing activities to business
outcomes ➔ Activation – They used the latest automation
and machine learning tools to reach and connect with
customers at scale and personalized their messaging ➔
Ways of Working – Their agile, customer-centric teams
had budget flexibility across channels to understand, test,
learn, and act on measurement insights Advertisers that
take control of their marketing and advertising data and
technology can respond quickly to customer needs and send
personalized messages at the right moment, Bain’s research
found. This is the ultimate promise of digital marketing and
is unlocked through measurement maturity, according to
Bain.

Deloitte and MIT Sloan’s View of Digital

Marketing Maturity
Following four years of research across three groups of
advertisers (early, developing, and maturing), Deloitte and
MIT’s Sloan School Of Business delivered their view of digital
marketing maturity. Their perspective was based on the idea
that adapting to increasingly digital market environments
and taking advantage of digital technologies to improve
operations are important goals for nearly every
contemporary business. Through their analysis, Deloitte and
MIT found many advertisers are beginning to make the
necessary changes to adapt their organizations to a digital
environment.
Based on a global survey of more than 4,300 managers,
executives, and analysts and 17 interviews with executives
and thought leaders, the research showed the digital
business environment is fundamentally different from the
traditional one. Digitally maturing advertisers recognize the
differences and are evolving how they learn and lead to
adapt and succeed in a rapidly changing market. [100]
Benefits of Digital Marketing Maturity
In the simplest sense, digital marketing maturity helps
advertisers identify, acquire, and deepen relationships with
high-value customers. Advanced technology and analytics
provide a digital measurement foundation to gain a better
understanding of the customer lifetime decision journey.
We’ll delve into how analysts can contribute to customer
understanding in Part 4 of this book. Let’s close the
discussion of digital marketing maturity by reviewing the
quantified benefit of each approach.
➔ BCG found leading businesses that are adding customers,
ROI and competitive advantage focus on adopting a path to
full data-driven marketing and attribution. Those that
succeed are seeing significant benefits – reducing costs by
up to 30 percent and increasing revenue by 20 percent. [101]
➔ Bain ’s research found the 100 most measurement-
mature advertisers were four times as likely to exceed
business goals compared with the 100 least-mature
advertisers. [102]
➔ Deloitte ’s work revealed that highly mature advertisers
are far more likely to develop the leaders they need for the
future. [103]
The benefits of digital marketing maturity are clear in a
philosophical sense – after all, it’s hard to argue against the
benefits of maturity – and, as the research shows, drive
business benefits.
In Summary: Digital Marketing Maturity
Digital marketing maturity is a quantified measure of an
advertiser’s sophistication with regard to digital data
collection and use. Several firms have defined this concept:
Three leading ideas come from BCG, Bain, and Deloitte (in
conjunction with the MIT Sloan School Of Business). BCG’s
view of maturity is based on a robust assessment across a
broad array of digital capabilities. Bain Consulting’s view of
maturity is centered firmly on how savvy an advertiser is in
its digital measurement. Deloitte and MIT’s joint research
found advertisers fall into one of three levels of maturity.
Regardless of the approach to measuring maturity,
advertisers found to be more mature saw benefits in sales
growth and cost efficiency.
 PART 4: The Art of
Analytics
Lesson 1 Navigating to Your Big Idea Lesson 2
Planning for Your Analytics Expedition Lesson 3
Collecting Data, Data Everywhere Lesson 4
Analyzing for Insights Every analyst dreams of coming up
with the “big idea” – the game-changing and previously
unseen insight or approach that gives their organization a
competitive advantage and their career a huge boost. But
dreaming won’t get you there. It requires a thoughtful and
disciplined approach to analysis projects. In this part of the
book, I detail the four elements of the Marketing Analytics
Process (MAP): plan, collect, analyze, report.
Part 4 also explains the role of the analyst, the six mutually
exclusive and collectively exhaustive (“MECE”) marketing
objectives of analytics, how to find context and patterns in
collected data, and how to avoid the pitfalls of bias.
PART 4 | LESSON 1
 Navigating to Your Big Idea

Five things discussed in this lesson:

Getting to a big idea is hard, but following a

process map will help analysts be effective and
efficient

Thorough analysis begins with thorough

planning

Analysts can collect data using three tried-and-

true techniques (although one among them stands
out as the most capable technique)

Finding patterns in data requires analysts to be

purposeful in how they investigate the data they
collect

Reporting results is a critical exercise that

separates big ideas from simply solid analysis

Analysts desperately want that big idea. They want to

uncover the insight that expresses something no one else
sees, gives their organizations a competitive advantage,
and earns them accolades.
The impossible way for the analyst to reach the insight they
seek is to sit alone in a room and think on the problem until
they come up with the solution. Indeed, analytics is not an
individual sport. The Mad Men days of three martini lunches
and big personalities delivering campaign ideas born from
lightning strikes of inspiration are a thing of the past, even if
some advertisers cling to the way things were before. Today,
a methodical planning process that draws insights from data
paves the way for success.
Navigating the analytics journey successfully requires
several important and required elements. My friends, Jamie
Shuttleworth of mcgarrybowen, Karl Turnbull of Cavalry, and
Ross McLean of 20|20 Research, captured this idea very well
in a simple visualization. Shuttleworth, Turnbull, and McLean
developed the image below, which neatly places “YOU,” the
analyst, in communion with your big idea.
Figure 4.1: You sat cozily with your big idea [104]

The image was printed on a folded piece of paper. As the

paper was pulled out of its folded state, all the inputs
required to get you, the analyst, to that big idea were
revealed.
Figure 4.2: The process of navigating to your big idea [105]
Among the items detailed in the image are the knowledge
the analyst will earn, the data the analyst will pull in, the
data the analyst will cast out, the stimuli the analyst will
want, the tension the analyst will create, the outsiders who
will provide context and new perspectives, and the pain that
everyone involved in the process will certainly feel. Any
good insight, and indeed, any good analysis, will involve
some pain. But that pain signifies the growth of ideas.
Luckily for us, we have a map that will guide us along the
way and help us navigate to our big idea.

The Four Steps of the Marketing Analytics

Process (MAP)
The marketing analytics process (MAP) that guides us along
our analytics journey consists of four steps: plan, collect,
analyze and report. The four steps are a circular process and
each time through informs the next journey. Analysts find
data they’re comfortable working with. They hone analysis
techniques through experience. And as analysts prepare to
report their findings, they learn what audiences value and
what they don’t like. The knowledge analysts build through
this process is important. The next time they start a
planning phase, they’ll have understanding, data, and
insights to apply moving forward.
Figure 4.3: The MAP in visual form

Let's take a deeper dive into the MAP’s steps. Analyst start
their journeys with thorough planning.

First Step: Plan

The first item in the plan step is to establish a clear, singular
objective for the analysis. Singular is a key word here. The
analyst can’t tackle multiple objectives with their analysis.
It’s imperative that the analyst fight for this upfront by
getting executive sponsors and stakeholders to align on a
singular objective and commit to its pursuit. Taking on too
many objectives (i.e., more than one) will scatter resources,
analysis, and focus. Being singular in the objective allows
the analyst to bring the full weight of their analysis to a
specific, identified goal. We’ll review the comprehensive set
of Business Objectives available to analysts in the next
lesson.
The second phase in the plan step is defining the key
questions analysts will ask of the data. Approach this by
thinking of the three (or more) questions to answer to
achieve the objective. For example, if the objective of my
analysis is to determine how my company/client can
improve awareness for a new product, I might ask questions
such as: ➔ What ’s the current level of awareness for
products in this category?
➔ Why have some channels been successful while others
haven’t?
➔ How can I use media most efficiently to build awareness
among my target consumers?
Questions such as these will dictate the data the analyst
needs, as well as the sources for that data. Getting those
key questions right is a valuable and important step. We’ll
discuss the art of question design and explore the Plan step
more broadly in Lesson 2.
The final thing the analyst will do in this plan step is to pull
it all together in a planning document (a literal plan) that
will help guide them through the rest of the analysis. The
key elements of this planning document make clear the (1)
objective, (2) key questions, and (3) data and sources used
to find those answers. This document helps keep the
analysis centered as the project works its way through the
remaining complex (and distraction-filled) phases of the
MAP. We will see an example of such a plan in Lesson 2.

Second Step: Collect

In the collect step, the analyst seeks data to answer key
questions. The first step is to find and engage the sources of
required data. Frequently, data sources in the plan step
don’t work out. Data owners could deny access to those
sources, access could be too expensive, or the data could
no longer exist. This is one of the primary reasons we need
to be flexible in our analysis journey. When data proves to
be inaccessible, we must refresh our plan with a new
dataset that can provide the answers we seek. This is where
the experience gained from previous journeys will come in
handy, as quickly replacing an inaccessible data source with
one the analyst used in the past will preserve the analysis
timeline. We will explore data sources in Lesson 3.
Once the analyst identifies and engages sources, the second
phase to the collect step is to use data mining techniques
and other tactics to pull that data out and get it into a
usable form. We’ll discuss these techniques at length in
Lesson 3, but for now, know there are primarily three: (1)
accessing facilitated downloads, (2) tapping application
programming interfaces (APIs), and (3) scraping webpages.
The final phase in the collect step is select a data
management system to house the collected data. Many
systems are available, but the key consideration is
balancing analysis power with simplicity of use. Usually,
those are negatively correlated, as we will see in Lesson 3.

Third Step: Analyze

In this step, we want to uncover powerful, relevant insights
that will compel action. We have an objective, we have
identified key questions we want answered, and we have
collected the data we want to explore. Now, it’s time to get
into the analysis and see where the data takes us.
The first phase to the analyze step is producing an analysis-
ready, tidy dataset. We'll talk about this later in Lesson 4,
but beginning with clean, well-organized data will make that
process much more efficient and effective.
The next phase of the analyze step is to perform analysis.
We’ll talk about a number of techniques in Lesson 4. Each
has a varying level of depth and, therefore, produces
varying levels of insight. Sometimes, a simple analysis with
a simple insight is all analysts need. Other times, it’s
important to go deeper. We’ll discuss how those analysis
techniques produce that depth of insight.

Step Four: Report

The fourth and last step in the MAP is to report the analysis
results. In this step, analysts should feel tension as they
compress the story into a tight, digestible packet.
Compressing this story means no extraneous information
obscures insights that matter, the logical flow of the story is
solid, and the story’s aperture can be opened wide enough
to provide details to low-level stakeholders and focused
enough to provide a high-level narrative to a C-level
audience. Part 5 of this book is dedicated to this topic.
The first phase in the report step is to design a visual story
that reaches audiences effectively and efficiently. The
analyst will do this by leveraging preattentive attributes in
visual perception, which we’ll discuss at length. While that
sounds very scientific, we’ll talk about how a few simple
design tweaks can bring greater contrast to visuals and
influence audience understanding significantly.
The second phase is to ensure recommendations are clear
and concise. Concepts, ideas, and recommendations
included in the analyst’s report must be easy to understand.
We’ll discuss some simple rules to help ensure our message,
and all of its elements, is precise.
The last phase in the report step is to ensure the
presentation is distraction-free by focusing on detail and
sophisticated execution elements. We’ll review a number of
guidelines and tips to help analysts complete this final task
successfully.

‘The Ask Behind the Ask’

Open and frequent communication between analysts and
stakeholders is a hallmark of any successful analytics
journey that arrives at a Big Idea. While communication
throughout the process is important, it’s never more critical
than at its beginning. This is when clear expectations must
be set and understood.
One of the greatest complications for the analyst is the
discussion of needs and requirements for an analytics
project with nonanalyst stakeholders. The analyst is a
strategist who applies a specialized set of tools and skills on
data to solve problems and find new opportunities. Analysts
cannot (and should not) expect their nonanalyst
stakeholders to know the solutions that would solve
stakeholders’ challenges.
Put simply, the analyst is not an order taker. The analyst is
not employed to simply produce on-demand analytics
products that stakeholders request from a neat, well-defined
menu. The sophistication of business today and the
complexity of data analytics projects dramatically reduces
the shelf life of unmalleable analytics projects. What may
make sense for the business today will almost certainly be
out of date within a few quarters (if not weeks). More
importantly, the carefully prescribed analytics menu
constrains analysts and stifles their creativity. This, in turn,
leads to suboptimal outcomes for the business as well as
the analyst’s career.
It’s important for analysts to get to what we call “the ask
behind the ask” from their stakeholders. In other words,
rather than taking a nonanalyst stakeholder’s request at
face value, the successful analyst will seek to uncover what
the stakeholder is attempting to solve. The nonanalyst’s
requested solution is “the ask,” but the real underlying
challenge that has motivated the request (the ask behind
the ask) is what the analyst must understand. By doing so,
the analyst can use their expertise to assess the challenge
and craft an appropriate solution.
I recently witnessed a well-intentioned sales leader
approach an analyst with a lengthy data request. The sales
leader had just received a call from a client during which the
client shared plans for a new campaign to take advantage of
an unexpected spike in consumer demand.
“The client needs help with their approach to consumer
targeting,” the sales leader stated, “so there’s some data I’d
like you to pull.” Without any hesitation, the sales leader
launched into a review of the long list of data they had
hastily scribbled on a notepad. At that point, the analyst
interjected with one simple question: “What are we trying to
solve?”
Those six simple words stopped the sales leader in their
tracks. What followed was a thoughtful and thorough
discussion of the client’s market position, the difficulties
they faced, and the opportunities before them. This led the
sales leader and the analyst to a concise articulation of the
challenge facing the planned campaign. In a few minutes,
the analyst successfully guided the sales leader to an
entirely different analysis path using a different collection of
data that would produce a different set of outcomes. These
outcomes would provide a clearer and actionable solution to
the client’s true challenge. When the analyst finished
describing their proposed approach, the sales leader stood
in amazement.
“I didn’t know we could do that,” the sales leader said
gleefully.
An excited discussion of reasonable turnaround times and
deliverables followed. The analyst then set off to do the
work secure in the knowledge that they were solving the
client’s true challenge while working against a set of
reasonable expectations.
That the sales leader didn’t know the analyst’s proposed
solution was a possibility illustrates a harmonious system
working exactly as it should. In this (as well as any)
scenario, the sales leader cannot and should not be
expected to know the best way for data to be collected,
analyzed, and presented. Such understanding is the
responsibility of the analyst and is precisely the reason why
the analyst who dutifully rushes off to fulfill orders from
nonanalyst stakeholders won’t be successful.
By getting to the ask behind the ask, this analyst applied
their expertise in data analytics to the client’s challenge
rather than pursuing the request of a nonanalyst
stakeholder. More importantly, the analyst saved the team
several rounds of back-and-forth with clients and countless
hours of pointless (and, most likely, ultimately rejected)
analysis.

In Summary: Navigating to Your Big Idea

Getting to a big idea is hard but following a process map
enables analysts to be effective and efficient. That process
involves four steps: plan, collect, analyze, and report. We’ll
use this process to guide us through the remainder of this
book. We'll look at the steps in-depth so that analysts can
understand them and know how to apply them. As we do,
keep in mind how important it is for analysts to get to what
we call “the ask behind the ask” from their stakeholders.
Only in this way will the analyst ensure the open and clear
communication required of a successful analytics journey.
In the next lesson, we’ll focus on the planning phase and
how to set objectives for analysis.
PART 4 | LESSON 2
 Planning for Your Analytics
Expedition

Five things discussed in this lesson:

Identifying an advertiser’s Marketing Objective

is critical for successful analysis, but this process is
frequently poorly executed and often goes awry

Analysts can choose from six primary Marketing

Objectives when planning digital marketing analytics

Finding your Marketing Objective is simple when

using the CDJ and asking key questions

Prioritizing objectives is an important exercise,

as the success of an analysis hinges on the
objective’s clarity and effect

Design a plan for analysis using a modified

scientific research approach

During a November 1957 speech, Dwight Eisenhower

famously told military personnel, “Plans are worthless, but
planning is everything.” [106] This motto applies nicely to how
analysts should think about their approaches to developing
plans for analysis projects.
The uncertainty that analysts face when starting their
journeys means that flexibility is crucial. All too frequently,
data deemed necessary is unavailable, errors render data
unreliable, or patterns revealed in data contradict the
stories analysts believed. Analysts must modify their plans
quickly to keep analyses on track in these situations. The
dynamic and shifting nature of the analysis journey means
that analysts’ carefully laid plans can appear worthless.
The planning process yields insight into an organization’s
needs, the stakeholders invested (or not invested) in the
analysis, and other elements that affect projects and can be
taken as important truths. Analysts can use these as “North
Stars” to anchor their analyses and guide projects through
inevitable twists and turns. One such truth is the objective
that acts as a fixed point in the distance for analysts and
their analysis journeys. The rigor needed to identify that
objective successfully – in Eisenhower's perspective, the
“planning” – can keep analysts centered, regardless of how
thoroughly they’re forced to abandon elements of their
plans.

When Objective-Setting Goes Awry

Marketing tends to go sideways for a few recognizable
reasons. Too much complexity in a brand’s marketing leads
to confusing messages that make it difficult for consumers
to understand how a product will fulfill their needs. At the
other end of the spectrum, marketing without refinement
leads to broad messages that lack the punch needed to
pierce the consumer’s consciousness. More often than not,
such debilitating errors enter the marketing process at early
stages. While errors are typically small at the time of
introduction, they infect other elements of an analyst’s
work, rendering a campaign ineffective.
One of the most damaging errors analysts make happens at
one of the earliest stages: establishing an objective
improperly. Should that objective be off in some way, the
effect on the work it guides will ripple through every step. In
my experience, three common errors occur during the
critical stage of objective setting that an analyst must fight
to avoid:

1. Making the objective too broad

2. Confusing an objective with a result
3. Making the objective too complex
Let’s take a deeper look at each of these errors and how to
correct them through proper objective setting.
Breadth
Focus when setting objectives is critical for success. A well-
defined, narrow objective is more attainable than one
spanning multiple outcomes. Work hard upfront to ensure
the objective is singular. One easy way to make sure this
happens is by forbidding the use of the word “and” in the
documented objective. For example, the broad objective
“build awareness for our new product AND influence
consumers to choose our new product” becomes two
separate, singular objectives (i.e., “(1) Build awareness for
our new product,” and “(2) Influence consumers to choose
our new product”) that must be tackled by two separate
analysis paths.
Objectives Versus Results
Too often, analysts incorrectly identify objectives as results.
No marketing analysis can yield an insight for an advertiser
to “drive sales.” Sales are not an objective, but rather the
result an advertiser realizes when achieving a different
objective. For example, an advertiser whose sales suffer
from a lack of awareness must focus on increasing
exposure. Therefore, “building awareness” is the
appropriate objective that will ultimately “drive sales.” In
the next section of this lesson, we’ll see that the objectives
marketing analysis should seek to improve are what we’ll
call “Marketing Objectives” while sales, volume, and profit
are “Business Objectives” that should be viewed as results.
Complexity
When attempting to rally an organization behind an
objective, message clarity and conciseness are critical.
Likewise, straightforward messages are important when
centering stakeholders on the objective of an analysis. To do
this, analysts must simplify their approaches to objective
setting. Analysts can feel confident in the soundness of their
objectives when selecting from one of six well-known and
tested Marketing Objectives.
The Six Marketing Objectives
I submit that six Marketing Objectives can serve as the
focus for analysis and the objective of any campaign. Those
six objectives are: stimulate demand, build awareness,
influence consideration, improve the sales processes,
reposition the brand, and increase loyalty.
The six Marketing Objectives simplify the underlying
complex systems of consumer/brand connections. They’re
true objectives that marketing can influence (i.e., they
aren’t results but are marketing levers that will end in
results), and – most importantly – they’re mutually exclusive
from one another. Marketing Objectives, and their relative
position to other categories of important objectives, are
presented here in Figure 4.4:
Figure 4.4: A taxonomy of objectives

The objectives presented in this visual are defined as

follows: ➔ Business Objectives – Broad C-level goals the
company wants to achieve, stated in the company’s own
language and easily measurable. These objectives are the
result of the company achieving its Marketing Objectives.
➔ Marketing Objectives – Measurable department-level
goals that support Business Objectives and can be
influenced by marketing. Marketing Objectives are the result
of the company achieving its Media Objectives.
➔ Media Objectives – Broad goals affected by various
media executions and tactics employed to present
consumers with the brand’s messages. Media Objectives are
reached when Campaign Objectives are successful.
➔ Campaign objectives – Key performance indicators
(KPIs) through which campaign success is measured.
The most important layer of objectives for marketing
analysts to understand is the line of Marketing Objectives.
These objectives will result in revenue, volume, or profit
growth that’s critical to business performance. In addition,
these are the highest level of objectives that marketing
performance can influence. As such, the analyst can most
affect the organization by conducting analysis to solve
challenges to Marketing Objectives.
There are a few important things to note about the
taxonomy of objectives related to the interplay between
lines of objectives and their relevance.
First, this visual isn’t intended to be read in a vertically
linear, one-to-one fashion. In other words, while each layer
of objective drives the layer above (and is, in turn, driven by
the collection of objectives layered below) nearly any
objective on a line can be used to stimulate one of the
objectives above. Only the company’s unique mix of
resources, market opportunities, and competitive dynamics
will determine which lower-level objective will move an
objective above. For example, while increasing reach and
frequency may be the unlock to build awareness for one
company, reach and frequency could help another company
realize success needed to influence consideration.
Second, an advertiser will most likely face multiple problems
across any of these objectives. Advertisers that don’t need
to build more awareness, influence consideration more
effectively, and increase loyalty are rare. Nearly every
marketing executive I’ve met would love to improve all six
Marketing Objectives simultaneously. As discussed earlier in
this lesson, however, the analyst must focus their analysis
to deliver insights related one Marketing Objective at a time.
Let’s take a deeper look at each of the six Marketing
Objectives now.
Stimulating Demand
Stimulating demand is a curious Marketing Objective. At its
root is the idea of convincing consumers they need an
advertiser’s product. Of course, this idea is not unfamiliar to
advertisers. Every brand in the world recognizes the need
for consumer demand. The challenging thing about pursuing
this objective lies in the need for one critical element:
authenticity.
Consumers – and especially consumers of today, with their
limitless access to information – can sense inauthentic
brand messages from miles away. A brand that simply calls,
“Please! Buy my product!” to the market will undoubtedly
face consumer apathy or, even worse, distrust. After all, as
we’ve seen in today’s consumer/brand relationship, it’s the
brand that must fit in the consumer’s life – not the other
way around. Consumers have too many choices for an
inauthentic message to drive demand. Brands can use
marketing, however, to effectively stimulate demand for
their products in a few limited ways.
One way a brand can achieve this Marketing Objective is by
providing clarity around how the product is used or the need
it fulfills. If consumers are confused about the function or
purpose of a product, demand for that product will be hard
to earn. By clearing up this confusion – typically through
demonstration – the advertiser can reveal the ability of the
product to satisfy a need when, often, the consumer didn’t
know they had that need. When I think of brands that must
stimulate demand by offering more clarity around use,
myriad products in infomercials (particularly those with the
most bizarre appearance) come to mind.
Figure 4.5: I have only one question about the Naväge nose
cleaning machine: Why?
A second way brands can stimulate demand is through a
high-risk and (sometimes) high-reward approach: creating
scarcity. By announcing a dwindling supply of products,
brands can try to stimulate demand by effectively
manipulating market forces. This approach can have
disastrous results for a brand when the message is viewed
as a spurious attempt to “trick” consumers. In contrast,
creating scarcity can be effective with consumers who are
aware of the products and demonstrated affinity for them
previously. An example of a brand that successfully
stimulated demand through marketing scarcity is Knob
Creek’s mid-2009 “Thanks for Nothing” campaign.
Figure 4.6: Knob Creek’s expertly crafted campaign to
stimulate demand

Bill Newlands, president of Knob Creek owner Beam Global,

discussed the predicament that led Knob Creek to advertise
its low product stock at the time of the campaign’s launch.
Knob Creek bourbon must age nine years to produce a
sufficiently full-bodied taste, Newlands explained.
Unfortunately, the company did a poor job of predicting
supply in 2000. More accurately, the company hadn’t seen
the signs of the global economic meltdown that struck in the
latter half of the 2000’s first decade. The hard times
consumers experienced drove unexpected “double-digit
growth” for Knob Creek. [107] Anyone who works in the spirit
category knows that when times are tough, sales of spirits
soar.
Knob Creek ran full-page newspaper advertisements
apologizing to its brand devotees, mailed empty bottles to
journalists, and even handed out T-shirts bemoaning the
“drought of 2009.” It also secured demand for the product
with a significant portion of its base for years to come.
Stimulating demand is a well-known objective that can be
measured easily through market share and rates of sale. In
fact, this Marketing Objective is so obviously woven into the
fabric of marketing, there’s no need to continue its
discussion. For the rest of this book, we’ll focus on the five
remaining Marketing Objectives outlined below.
Building Awareness
Awareness is the percent of consumers who are familiar with
an advertiser’s brand or product. Awareness can be
measured by the percent of the total consumer population
that claims familiarity with the item being measured or
against a more specific target population (e.g., awareness
among demographic groups like “women 35+” or
psychographic groups like “frequent gamers”). General
awareness is nice to have, but awareness among key
consumer groups is always more important.
There can also be degrees of awareness. “Aided awareness”
means the percent of consumers who recognize an
advertiser’s brand or product when they see or hear either
name. “Unaided awareness” means consumers voluntarily
mention the brand or product when prompted by a generic
question such as, “What brands of automobile have you
heard of?” Unaided awareness clearly reveals a deeper
consumer/brand connection than aided awareness.
Consumer surveys are typically the best way to gauge
awareness. Survey measurement is particularly helpful for
brands that launch new brands, subbrands, products, or
features. Tracking awareness can help advertisers optimize
their approaches to targeting midcampaign toward
consumer groups (e.g., demographics) with the highest
awareness lifts.
Influencing Consideration Consideration is the
percent of consumers who have a positive perception
of an advertiser’s brand or product. Like awareness,
consideration can be measured in terms of the
general population or against specific segments of
the consumer base. It can also be measured in an
aided way (“Which of the following products would
you consider purchasing?”) or in an unaided way
(“Which automobile brands would you consider
purchasing?”). Consideration goes beyond awareness
to measure consumers who are more than simply
aware of the brand or product: They would purchase
the item or would consider buying it. Consideration
implies a deeper consumer/brand relationship than
simple awareness.
To that end, consumers have several layers of consideration.
Each represents a different degree of affinity for the brand
or product being measured. Basic consideration (“Would you
consider buying this product?”) indicates the lowest level of
consumer commitment to the brand. Favorability (“Which
athletic shoe brand do you like most?” – also known as
“preference”) implies greater brand affinity. Purchase intent
(“The next time you buy a personal computer, which brand
are you most likely to buy?”) measures the deepest degree
of the consumer/brand relationship.
Improving the Sales Process The sales process occurs
when a consumer exchanges cash for an item. This
can be seen clearly in a digital sense as the online
checkout steps that follow placing an item in a digital
shopping cart. Offline, the definition of the sales
process can blur the lines between improving the
sales process and the previous Marketing Objective of
influencing consideration. Sales associates in many
retail settings are now able to discuss product
benefits, close sales, (i.e., influencing consideration)
and check out consumers through the use of
handheld point-of-sale (POS) systems (i.e., the sales
process). Visiting an Apple Store anywhere will bring
this concept to life, as Apple has been outfitting sales
associates with handheld checkout devices for years.
To keep the Marketing Objectives we’re discussing clear and
mutually exclusive, we’ll define the offline sales process to
include only the exchange of money. In other words, from
the moment a consumer readies a credit card or other form
of payment (e.g., entering a checkout line).
Assessing the sales process is a study in measuring
efficiency. In the online sense, cart abandonment rate (i.e.,
the percent of consumers who place items in their digital
shopping carts, but then remove the items or cancel the
sales process before completion) is an insightful way for a
brand to reveal challenges in its sales process. Hidden
delivery fees, exorbitant taxes, and the inability for the
company to deliver the product by the time a consumer
needs it are all reasons why a brand’s sales process can be
disrupted. Other challenges, such as confusing or frustrating
checkout steps, can also affect a brand’s sales process.
Repositioning the Brand Repositioning the brand is a
relevant Marketing Objective when a brand fails to
live up to the expectations its advertising sets.
Infrequently, repositioning the brand can be an
important pursuit when the product provides an
unexpected benefit. For example, Pabst Blue Ribbon’s
sudden and dramatic popularity with the hipster
crowd in the 2000s may have been unexpected
among the members of PBR’s marketing team, but
the pivot revived a brand that had reached an all-
time low in 2001.
Postpurchase consumer response provides the best insight
into the need for repositioning and is trackable through
consumer surveys and product return rates.
In rare cases where brands or products have caused harm
and the existing brand positioning is no longer viable,
repositioning the brand can be worthwhile. Often, it’s the
only option available without abandoning the brand.
Following the Valdez oil spill in Prince William Sound, Alaska,
on March 24, 1989, Exxon, the tanker’s owner, had no
choice but to reposition its brand in a way that
demonstrated the company’s care and consideration for the
environment and sustainable energy practices.
Growing Loyalty
Consumer loyalty for a brand or product is a fundamental
component of a brand’s success. Measures of loyalty can
range from the simple (e.g., the percent of people who own
your product and buy that product again, or repeat
purchase rate) to the sophisticated (e.g., the highly-tuned
and complex calculations of lifetime customer values) to the
apparent (e.g., social media mentions from people who
experience the brand or product). Assessing loyalty for a
brand seeks to quantify brand advocacy: the all-important,
yet elusive, support for a brand that turns product owners
into authentic word-of-mouth advertisements.
Determining Your Marketing Objective
How can analysts determine whether their organizations
struggle with one of the Marketing Objectives described
above? Some simple questions and a bit of analysis will
reveal relevant and irrelevant objectives.
Figure 4.7: Questions to help determine your Marketing
Objective

Do You Need More Awareness?

Effectively, the awareness a brand or product earns is
reflected in the percent of consumers who place that item
on their Initial Consideration Set when the need is triggered.
Asking whether consumers are aware of the brand or
product will quantify the need for greater awareness.
The brand can approach these questions by collecting data
from a variety of sources: ➔ Aggregated consumer search
trends from Google Trends or other search trend tools ➔
Market share data from industry groups, investor services,
and market research companies ➔ Product and brand
awareness data from voice of the consumer survey tools ➔
Social media mentions from social listening or social media
platform tools Do You Need to Better Influence Consumers?
For influence and consideration, analysts should investigate
whether a brands’ products satisfy consumers’ needs. If the
products don’t, this means consumers are choosing
competitors’ products, revealing the need to better
influence consideration. The analysis of nuanced data,
typically collected through market surveys, can point to
required solutions. These solutions can include more clear
communication of product benefits, better understanding of
consumers’ needs, and a stronger “challenger brand”
position to steal market share away from a dominant brand
in the category (think of the “Pepsi Challenge” head-to-head
taste choice affront to market leader Coke).
The assessment of a brand’s need to influence consumers
comes down to measuring shoppers who experience the
brand during the active evaluation step of their decision
journeys. The brand can approach these questions by
collecting data from a variety of sources: ➔ A/B testing data
to provide insight into consumer preferences from A/B
testing or CRO tools ➔ Clickstream analysis (focused on how
consumers navigate a website to obtain insight into key
content as well as content deemed less valuable) from the
advertiser’s website analytics or clickstream analysis tool ➔
Consumer needs data from voice of the consumer survey
tools ➔ Product and brand consideration, favorability, and
purchase intent data from voice of the consumer survey
tools ➔ Social media sentiment data related to a brand from
social listening, content analysis, or social media platform
tools Do Your Sales Efforts Help or Hinder?
For the Marketing Objective, “Improve Sales Experience,”
analysts must focus on the Moment of Purchase by
investigating the sales process for leaks. As mentioned
earlier in this lesson, cart abandonment rate is an effective
way to quantify the need for changes to a brand’s online
sales process. In-store customer surveys, observing
consumers on their shopping trips (i.e., shop-alongs), and
consumer focus groups are among the best ways to assess
the efficiency and effectiveness of offline sales processes.
The brand can approach these questions by collecting data
from a variety of sources: ➔ A/B testing data to provide
insight into the effectiveness of various versions of checkout
process designs (e.g., messaging, number of steps, page
layout, etc.) from A/B testing or CRO tools ➔ Checkout
process data (e.g., cart abandonment data) from the
advertiser’s e-commerce system or e-commerce partner ➔
Checkout satisfaction survey response data for consumers
who buy the products and those who don’t using voice of
the consumer survey tools ➔ Clickstream analysis (focused
on experiences of consumers who convert and consumers
who don’t convert) from the advertiser’s website analytics
or clickstream analysis tool ➔ In-store customer surveys,
shop-alongs, and focus group data for offline Moment of
Purchase insights Do You Deliver on Expectations You’ve
Set?
After consumers make their purchases, follow-up is typically
the best way to provide the truest view into a need for
repositioning the brand. Tracking product returns,
monitoring social media, and other measures can also show
how well a brand delivered on the expectations set during
the decision journey.
The brand can approach these questions by collecting data
from a variety of sources during the Postpurchase
Experience: ➔ Customer satisfaction survey response data
for consumers who buy the products using voice of the
consumer survey tools ➔ Net promoter score (NPS) data
from NPS partner or market research companies ➔ Product
return/refund data from the advertiser’s internal systems or
an e-commerce partner ➔ Social media sentiment data
related to an advertiser’s products (focused on consumer
sentiment of posts that mention the products, implying they
own the products or have experience with them) from social
listening, content analysis, or social media platforms tools ➔
Other data collected related to the customer’s response to
an advertiser’s products (e.g., customer service
representative and help desk phone logs) Do Customers
Advocate for Your Brand?
Investigating consumers who advocate for a brand will
reveal whether a brand can be satisfied with its level of
loyalty. Advocacy quantified through repeat purchase rates,
positive social media mentions, and other means can help
analysts assess whether the Marketing Objective, “grow
loyalty,” represents a need.
The brand can approach these questions using many of the
same data and sources collected to assess the Postpurchase
Experience: ➔ Customer satisfaction survey response data
from surveys sent to consumers who buy the products using
voice of the consumer survey tools ➔ Social media
sentiment data related to an advertiser’s products (again,
product-level posts imply consumers own the products or
have experience with them) from social listening, content
analysis, or social media platform tools ➔ NPS data from an
NPS partner or market research companies ➔ Repeat
purchase rate from the advertiser’s internal systems or a
CRM partner
Prioritizing Objectives
As mentioned previously, most advertisers will have needs
across many of these Marketing Objectives. Since we must
focus on a single Marketing Objective, it’s imperative that
we prioritize a brand’s relevant Marketing Objectives. An
analyst must identify where need exists and find the
Marketing Objective that will have the greatest effect on
brand performance. One way to do this involves McKinsey’s
CDJ framework.
The questions we’ve discussed around the five critical
Marketing Objectives layer in well to the CDJ. Brand
awareness fits with the Initial Consideration Set. The need to
influence consideration can be assessed during the CDJ’s
Active Evaluation step. The sales process aligns with the
Moment of Purchase, just as the need for repositioning ties
to the Postpurchase Experience, and the need to increase
loyalty is attached to the durability of a brand’s Loyalty
Loop. A simple econometric model that quantifies the
number of consumers at each step will quickly point to
areas of critical need (as well as those of less-pressing
concern).
Assigning values to each CDJ step allows the analyst to
modify inputs at each step to calculate the overall effect of
that Marketing Objective. For example, if aware consumers
place a brand on their Initial Consideration Set following a
trigger, analysts can make assumptions about the percent
of consumers who will purchase the product for a certain
price. Analysts can, therefore, quantify the value of
awareness. Testing different assumption values (i.e.,
conducting sensitivity analysis) will reveal the incremental
effect of driving awareness. Analysts can perform similar
analyses at each step of the CDJ. Comparisons of those
model outputs across the CDJ steps, informed by the brand’s
appetite and ability to achieve the assumed levels of
success, help to identify the cost/benefit of each Marketing
Objective.

Building an Analysis Plan around Your

Objective
With the Marketing Objective in hand, the final phase in the
plan step is to create a clear, concise document that can
help guide analysis journeys. The document format matters
little. Analysts should adopt the visual framework that works
best for them. However, it’s important that the document: ➔
Clearly identifies the Marketing Objective – This “North
Star” truth will guide the analysis and must be succinct in
the analysis plan.
➔ Details the connection between the Marketing
Objective and key questions – Identifying key questions
that will logically advance the brand toward its Marketing
Objective when answered is critical. These questions should
be singular (i.e., eliminate the word “and” from all key
questions) and distinct from one another.
➔ Clearly identifies the data to be collected and the
source of those data – While the data and sources aren’t
foolproof, setting them out as areas for exploration is
important so that analysts use their time efficiently.
In my experience, designing a plan in a pyramid layout
works best to meet these criteria, while allowing for quick
modifications to plan elements later on. Analysts can ensure
the key questions they pursue explore the full gamut of
business challenges tied to their objectives by designing
three types of questions: What? why? and how?
Answers to “what” questions (e.g., “What is our brand’s
current level of unaided awareness?”) will typically provide
straightforward descriptive facts that can set the foundation
of understanding an analyst needs to begin a more robust
analysis. An analyst can get to the heart of the matter by
answering a “why” question (e.g., “Why did some of our
marketing efforts drive awareness while others didn’t?”)
that provides the context, texture, and nuance required of a
deeper analysis. Finally, solutions to “how” questions (e.g.,
“How can our marketing efforts affect awareness levels over
the next 12 months?”) point to actions required to achieve
the chosen objective.
Figure 4.8: The pyramid layout for an analysis plan

This line of investigation is rooted in the principle commonly

known as “The Five W's and How.” The Five W’s and How is
the idea that a complete report must provide answers to
questions that start with the interrogative words, who?
what? when? where? why? and how? The principle is
commonly applied in journalism, law enforcement
investigations, and research. The origin of the Five W’s and
How has been tied to Aristotle’s Nicomachean Ethics . [108]
For marketing analysis, analysts should provide answers to
what? why? and how? questions at a minimum to ensure a
complete story. Analysts can expand their analyses to report
on the other questions identified in the principle, as needed.
A Word on Anchoring Bias
The questions analysts ask are often influenced in the
direction of a relevant comparison value or “anchor” found
in their experience. In this way, anchoring bias (also known
as focalism) disrupts analysis by leading the analyst on
pursuits informed by their frame of reference but not the
data’s full picture. The anecdote to anchoring bias is
perspective. By broadening the aperture of the analysis, an
analyst can introduce a wider set of comparisons that will
challenge the original focal point. A more comprehensive set
of data, competing perspectives from outsiders, or new
comparisons will achieve this result.
In Summary: Planning for Your Analytics
Expedition
Identifying an advertiser’s Marketing Objective is critical for
successful analysis, but this process is frequently poorly
executed and often goes awry when complexity, breadth of
scale, and confusion between “objectives” and “results” are
introduced. By keeping things simple, an analyst can plan
for a successful analysis by choosing from six primary
Marketing Objectives when planning for analysis. Finding
which Marketing Objectives are relevant for a brand is
possible by using the CDJ and several key questions.
We saw that prioritizing objectives is an important exercise,
as the success of an analysis hinges on the objective’s
clarity and impact. Analysts should use a modified scientific
research approach, commonly known as the Five W’s and
How, to design analysis plans. Furthermore, analysts must
diligently guard against anchoring bias to ensure they are
asking the right questions as they set off on their analytics
journey.
In the next lesson, we’ll delve into the techniques analysts
can use to collect data for analysis.
 PART 4 | LESSON 3
Collecting Data, Data
Everywhere

Five things discussed in this lesson:

Tying collected data to key questions – and to

the identified Marketing Objective – is critical for
successful analysis

Data can come in many forms, including raw,

processed, unstructured, and structured

Analysts can use three techniques to collect

data: accessing facilitated downloads, using APIs,
and scraping the web

Ensuring data quality requires understanding

how bias enters the data and critical points of data
handling that frequently lead to errors

Tools help manage data, but each option has

pros and cons

With a nod to The Rime of the Ancient Mariner by Samuel

Taylor Coleridge [109] (“Water, water, every where, Nor any
drop to drink.”), marketers frequently bemoan the seas of
data that surround them and the frustrating reality that
insights are difficult to find: “Data, data everywhere, and not
a drop of insight.”
Thus far, we’ve seen the explosive growth of data in today’s
marketing landscape, the technical process of data
collection, and how brands can use that data to make
decisions. We’ll get to the development of insights through
analysis in the next lesson. For now, let’s investigate an
important question: How do analysts get their hands on
data? Data collection, which was once fraught with
challenges, has become surprisingly effortless.
Before we dive into the data collection process, let’s explore
the kinds of data analysts should collect.
Unstructured Versus Structured Data
Marketing analysis data comes in two forms: unstructured
and structured. Unstructured data doesn’t come in a
predefined, standardized format. It’s usually text-heavy and
might contain dates and numbers that don’t fit a uniform
description, as well as awkward (but increasingly important)
data, such as images, sounds, and video.
It’s estimated that unstructured data might account for 80
percent of all data in organizations. [110] Understanding
unstructured data and how to unlock it is very important.
Collecting and cleaning unstructured data and extracting
relevant information from it typically requires a tool (or
combination of tools).
What does unstructured data look like? A sampling of Twitter
activity collected using a tool like TweetReach gives a
perfect example. Basic text, links, videos, and still images
appear in tweets. Some tweets include hashtags that can be
an important part of the text string. Other times, a tweet’s
content can appear haphazard and disorganized. Although
Twitter confines tweets to a certain number of characters,
data length isn’t defined. That’s what makes it unstructured.
Figure 4.9: Unstructured data (and, to the author’s point, too
many #hashtags)
In contrast, structured data is neatly organized and
conforms to a clear and consistent data format. Each
column is well-defined. After seeing a data sample, an
analyst knows the exact format and layout they can expect
from the entire data table. Population statistics from the
U.S. Census Bureau are structured data.
Figure 4.10: Structured data tables from the U.S. Census Bureau

Although structured data comprises a relatively small portion of the

data available today, its efficient manner makes structured data tables
very important for analysts.
Collecting Data
Analysts access data in three ways. We’ll refer to these techniques as
accessing facilitated download sites, using application programming
interfaces (APIs), and scraping webpages.
Accessing Facilitated Download Sites Through facilitated
downloads, a data owner provides a place for others to visit
and collect data. These are tightly managed data releases.
Typically, a user interface – often called a graphical user
interface (GUI) – acts as an access point to this data. When
logins are required, access to the data can be monitored and
managed, allowing the data owner to determine the level of
access. In this technique, I’m including companies’ systems
that house proprietary data and allow analysts to access the
information. For example, when analysts collect first-party sales
data from corporate accounting systems, they do so through
facilitated downloads. Similarly, accessing DMP-managed
second- or third-party data is done through a facilitated DMP-
designed download process.
Another example of a facilitated download is the way consumers can
access U.S. Census Bureau data. As the principal agency of the U.S.
Federal Statistical System responsible for producing data about the
American people and economy, the U.S. Census Bureau built the
Explore Census Data website (data.census.gov ) to encourage access to
this data. Analysts will find an easy-to-use and simply designed GUI
that facilitates access to census data.
A data owner’s facilitated download site doesn’t always mean easy
access. The Internet Movie Database (IMDb.com ) has a facilitated
download source, but it’s difficult to locate. By navigating to
datasets.imdbws.com analysts will find subsets of IMDb data that are
available for customers’ personal and noncommercial use in a tab-
separated-value (TSV) format. The site lacks the clarity and elegance of
other facilitated download sites, however.
We live in an era of such plentiful data that an analyst needs to do little
more than sign up for a newsletter to earn tremendous amounts of
information. For example, Kaggle (kaggle.com ) boasts more than
19,000 public datasets to help analysts “conquer any analysis in no
time.” [111] Signing up for the Kaggle e-newsletter will deliver weekly,
interesting, and new datasets that anyone can download from the
Kaggle website. The launch of Google Dataset Search
(datasetsearch.research.google.com ) places almost 25 million datasets
at the analyst’s fingertips.
Accessing data from facilitated download sites should be the primary
method for collecting data today.
Using APIs
A second approach for accessing data is using APIs. APIs let computers
connect to other computers to complete machine-to-machine data
transfers. An API brings Google Maps data to restaurant websites,
weather data to news sites, and real-time financial market information
to investment websites. Analysts must understand scripting language
and how to collect access tokens API owners grant to build API
connections. While accessing data directly via APIs can be valuable, the
time required to build those connections often outweighs the benefits.
A number of data analysis tools have simplified API access for the
analyst. Many tools described in Part 3 of this book, including
Hootsuite, TweetReach, and Searchvolume.io, collect second-party data
through API connections to social media networks and other data
sources. These tools feature easy-to-use GUIs where analysts input
criteria for the data they want to collect (e.g., keywords, date ranges,
etc.). The tool builds the API query based on the analyst’s inputs,
establishes the API connection and handles the necessary
permissions/token requirements.
In this way, analysis tools simplify data collected through APIs.
Web Scraping A third way analysts can access data is through
web scraping, also known as “screen scraping,” “web data
extraction,” and “web harvesting.” In the early days of the web,
black hat analysts found and collected unique data treasures
using web scraping. Open-source datasets and facilitated
download sites have largely eliminated the need for web
scraping, however.
In my opinion, analysts are rarely required to build web scrapers in R or
Python, or use tools like Parsehub (parsehub.com ) to scrape data from
websites. First, the time it takes to construct the code or configure the
web scraping tool can be prohibitive. Second, so much data is now
available from data owners, who have created data-sharing processes,
that the need for scraping has been greatly reduced. Finally, and
perhaps most importantly, if data isn’t available through a facilitated
download site or an approved API, the owner likely doesn’t want to
provide access to it. In our current environment of heightened data
privacy concerns, it’s hard to think of reasons that justify the
acquisition of data not intended to be collected.
For these reasons, analysts should think twice before heading down the
web scraping path.
Ensuring Data Quality
Analysts must collect data in a way that ensures its quality. An analysis
is only as good as the data that analysts collect. Errors are possible at
several points in the data collection process. Understanding these
points of “data danger” can help analysts avoid common errors and
preserve data quality.
Questionnaire Bias Questionnaire bias can creep in through the
questions consumers answer, typically in surveys.
Questionnaire bias leads respondents to a conclusion or
influences their responses.
For example, a survey that asks, “What words would you use to
describe your love of (brand X)?” implies consumers’ experiences with
the brand have been positive. Analysts can eliminate questionnaire
bias by designing surveys that feature only neutral, unbiased
questions.
Sampling Bias Sampling bias, also known as “selection bias,” is
another way that data can be tainted. When collecting data
from a sample of a population, insights from that data are
applicable to the larger population only if the sample
represents the whole. For example, soliciting opinions on a
public official’s performance from only high net worth citizens
wouldn’t provide insight into the views of the entire constituent
population. When samples don’t represent the whole, sampling
bias prevents pattern extrapolation for the entire population.
Analysts can reduce the effects of sampling bias by ensuring
that surveyed consumers are drawn from all elements of the
population, including obvious attributes, such as gender, age,
and income, and those more difficult to discern, such as
interests, attitudes, and preferences.
Data-Handling Errors When analysts handle data, they can
introduce errors that contaminate the collected information.
Understanding these critical points can help analysts avoid
data contamination: ➔ Data integration errors – Joining
tables can introduce errors. Data accidentally copied, deleted,
or incorrectly matched are examples of integration errors that
contaminate datasets and have limited value. Handling data
carefully is the best way for analysts to minimize errors.
➔ Data input errors – Errors typically occur when analysts collect
data manually. They can transpose numerals, misread them, and enter
them incorrectly. Enacting data quality control methods, such as
normalizing data (i.e., allowing data inputters to select from a
predefined set of inputs), peer reviews, and other techniques can help
reduce input errors.
➔ Calculation errors – When calculations are used to
transform data, errors can be common. When analysts use
calculations to transform data, they can make errors. Peer
reviews and testing calculations on small portions of the
data before rolling out to the entire dataset can help reduce
such errors.
A Word on Data Tidying
Even if they were to eliminate all errors in their data,
analysts must ensure the data is well-organized before they
begin their analyses. Ensuring “tidy data” is a standard way
to structure a dataset that makes it easy for an analyst or a
computer to extract needed variables. Hadley Wickham, the
chief scientist at RStudio, has defined Tidy Data as datasets
that are arranged such that each variable is a column, each
observation (or case) is a row and each value has its own
cell. [112] Third-party data and data collected from public
sources often lack the structure and organization to meet
Wickham’s definition.
Tools, such as Trifacta Wrangler (trifacta.com/start-wrangling
) and Open Refine (openrefine.org ), can help analysts work
with messy data by cleaning it, modifying it from one format
to another, and transforming it into proper Tidy Data.

A Word on Managing Collected Data

As you move collected data into a management system,
you’ll have many tools from which to choose. Using Excel or
Google Sheets to manage data that’s lightweight or limited
in scope and scale is acceptable. I’ve conducted many
analyses using only Excel. But if a dataset requires greater
statistical capabilities, you can look to a long and
distinguished set of viable statistical packages, including
SPSS, Minitab, and Stata. From software as a service to
statistical packages for the social sciences, a tool is
available that will provide the capabilities you need to
manage collected data and conduct your analyses.
Balance computational power with ease of use until you find
a tool that meets your data and efficiency needs.
In Summary: Collecting Data, Data
Everywhere
Linking a singular objective to data is critical for the
collection to be conducted efficiently, regardless of whether
those data are unstructured or structured. Analysts can use
three techniques to collect data – accessing facilitated
downloads, using APIs, and web scraping – although the
time investment, privacy trappings, and other headaches
associated with web scraping should dissuade analysts from
pursuing that activity. Regardless of how analysts collect
data, ensuring data quality means understanding how bias
enters the data and critical points of data handling that
frequently lead to errors. Once analysts collect data, they
can manage it through a wide variety of tools, each with its
pros and cons.
In the next lesson, we’ll explore approaches to finding
insights in data.
PART 4 | LESSON 4
 Analyzing for Insights

Five things discussed in this lesson:

Analysts who seek to reduce large amounts of

data into valuable insights will be more successful
than those who mine data for rare and undiscovered
jewels

Analysts can choose from five analysis

categories, each with its own degree of difficulty and
depth

Context can be the most important element an

analyst provides

Finding patterns in data requires analysts to

create views into the data, often from multiple angles

Even if an analyst’s data is error-free,

interpretation bias can skew results

We’ve reached the point where we can begin analyzing

data. The proper approach to analysis is more like making
diamonds – compressing all the things we learn into tight,
consumable pieces of insight – than hunting for gold.
Developing “diamond insights” forces analysts to delete
extraneous facts that don’t contribute to the story that
emerges from the data. Such insights will be on display as
we attempt to synthesize findings into a concise 60-second
story that expresses the true heart of the matter and
becomes the yardstick by which we’ll measure the value of
our analysis. As Fassnacht said, “If there is not a 60-second
story, then there is no story at all.” [113]

The Five Categories of Marketing Data

Analysis
Five primary categories of marketing data analysis reveal
myriad insights. Each method features a varying level of
sophistication and returns a corresponding depth of insight.
Let's take a look at them.
Descriptive Analysis The first category of marketing
data analysis is descriptive analysis. This is typically
the initial kind of analysis an analyst performs. The
objective of descriptive analysis is to create a
summary of the data to yield useful information and
prepare the data for further analysis. For example,
summing visitors by month from a dataset of daily
visits is a way to gain a descriptive understanding of
total monthly website visits. Calculating a mean,
median, mode, count, and any other standard
statistical values also fits this category of analysis.
Descriptive analysis will yield clear answers to simple
questions the analyst has asked but won’t provide
much more detail. Monthly website visitors can
provide insight into traffic trends but answering more
nuanced questions requires more sophisticated
analysis techniques.
Inferential Analysis Inferential analysis is a second
category of data investigation. This approach
features the use of a small set of data to infer
something about a larger set. Inferential analysis is
commonly the goal of analysis conducted on survey
data that has been modeled to account for sampling
bias (discussed in the previous lesson).
Exploratory Analysis The third category of analysis is
exploratory analysis. Typically, visual-based,
exploratory analysis seeks to discover connections
and patterns in data. This category might not always
answer questions, but it reveals interesting
connections that lead to deeper investigation and
understanding. In this way, exploratory analysis can
be a good early step on a broad dataset that points
analysts to other techniques.
Causal Analysis
A fourth category of analysis is causal analysis. Causal
analysis seeks to determine how the movement of one
variable registers on other variables. Typically regarded as
the gold standard for data analysis, causal analysis uses
correlation and regression techniques to explore the
relationship between attributes in a dataset.
Predictive Analysis
Predictive analysis is the final category of marketing data
analysis. Closely linked to causal analysis, predictive
analysis uses models to predict the future value of data
attributes when market forces move some related data.
Successful predictive analysis depends on the analyst
having the right data and the right data quality. If the model
includes data errors, the predictive value will be diminished.
Predictive analysis is the most sophisticated method of
modeling and analysis.
Media optimization modeling (e.g., MMMs), attribution
modeling (e.g., MTAs) and other kinds of response modeling,
such as consumer response modeling, are examples of this
analysis category.

A Word on the Importance of Context

While conducting analyses, it’s important to consider
context. A number rarely means much on its own. Unlocking
the meaning of a number requires a baseline – a
comparative number so that the audience can evaluate the
figure’s value. Internal or external benchmarks, goals, and
comparisons to previous performance add context and
insights to data.
Finding Patterns in Data
Often, the best way to find patterns in data is to visualize
that data. Visualizing data expresses stories that are
unrecognizable when the data is in tabular form. We look for
five data patterns.
Figure 4.11: Categories of patterns found in data

Change will present as either gradual change or sudden

shifts. Clustering represents a collection of data points that
are similar to one and yet different from others. Visual
techniques that demonstrate relativity allow analysts to
identify how two different data points relate to each other.
Ranking determines what’s at the top of the scale, what’s at
the scale’s lowest point, and everything in between. Finally,
how one set of data influences or affects another is revealed
through correlation. Each of these patterns indicate the
presence of powerful insights that can aid the analyst’s
story.
Certain visual techniques will reveal each pattern and others
will obscure them. For example, a line graph will show
change over time, but not ranking. A pie chart might be
great at depicting relativity but won’t reveal change. In
other words, it’s important to identify the appropriate visual
to answer each specific question and ensure they’re in
alignment. Figure 4.12 can help guide the analyst to the
right visual technique.
It’s important to identify the techniques analysts use to
answer questions and ensure they’re in alignment. Figure
4.12 can help guide the analyst to the right visual
technique.
Figure 4.12: Matching pursuit of patterns to visual forms

Analysts can use a variety of tools to apply each category of

visual patterns to the data they’re analyzing: ➔ Change –
Analysts can produce common visual forms, such as line
charts and bar charts, using nearly any data analysis or
visualization tool they choose – from Excel spreadsheets to
more advanced data visualization tools like Tableau.
➔ Clustering – While Excel, Google Sheets and other tools
have cluster analysis capabilities, the process is unwieldy.
Statistical software, including SPSS, features more elegant
approaches to clustering data. In addition, analysts can
execute clustering analysis with R using its built-in K-means
command or conduct more robust clustering analyses using
fpk , pvclust , and other packages.
➔ Relativity – Any data analysis or visualization tool will
create pie charts and other common visual forms. Analysts
can use a tool like Wordle to conduct text relevance
analyses to visually quantify relativity in text.
➔ Ranking – Analysts can use any tool to create bar charts
and other simple ranking techniques. More sophisticated
visual forms, such as spider charts (also known as radar
charts) or Sankey diagrams, require the use of more
sophisticated tools such as Tableau, R, or D3.
➔ Correlation – While analysts can produce simple
correlation techniques using spreadsheet tools, they can
perform deeper correlation analyses using SPSS and other
statistical software or more robust analysis tools like R.
Countering Bias in Analysis
While biases like questionnaire bias and sampling bias can
lead to errors that degrade the quality of collected data,
biases during analysis can be equally crippling. We can
generally classify biases that creep into analysis under the
heading of “interpretation bias,” but analysts should know
about two common varieties: accessibility bias and
confirmation bias.
Accessibility bias People tend to make assumptions
based on limited information and fail to seek
information not already known. Analysts can protect
against this bias by ensuring they analyze the data
they need – not simply the data they have. The
simplest way for analysts to do that is to look
objectively at the question the analysis is answering
and compare the data included in the analysis with
the data they would collect in a perfect world. If the
current dataset matches the ideal data (or if
dependable proxies have been found for the desired
data) analysts can move content forward, knowing
they’ve minimized accessibility bias.
An example of accessibility bias can be seen in a global
research company’s recent analysis. The company
published an article in late 2019 that purported to validate
the value of its new MTA model with a headline that
boasted, “5 Game Changing Facts That Prove The Value Of
Multi-Touch Attribution.” As discussed in Part 3 of this book,
multitouch attribution models are under extreme stress from
data privacy and cookie deletion policies that limit the
consumer behavior data fueling their algorithms. These
limits threaten the ability of MTAs to accurately attribute
value to media touch points by forcing a heavy reliance on
probabilistic models that are less precise than models that
use more robust data. The proof points this company offered
to validate its MTA, however, were five simple outputs from
109 datasets the company had analyzed using its MTA.
Analysts at the company ignored the real question at play –
do MTA models accurately attribute value? – and focused on
an analysis of the data on hand. While that data may have
resulted in interesting insights, it failed to produce the
analysis that was really needed.
Confirmation bias Approaching an analysis with a
preconceived idea of what the data will say
introduces the risk of confirmation bias. Such
prejudice can lead an analyst to draw a conclusion
after finding supporting evidence for that conclusion,
without reviewing the entire dataset. The analyst will
arrive at the answer they seek, but not necessarily
the right answer. Analysts can counter confirmation
bias by challenging their assumptions and allowing
stories to rise from the data organically.
A discussion I had with the chief marketing officer (CMO) of
a multinational quick-serve restaurant company illustrates
the challenge of confirmation bias. During a three-hour
meeting, I presented an analysis that demonstrated the
importance of digital media to this brand and substantiated
the need to divert money from television advertising to
online video. Hard-hitting facts related to demographic
shifts, consumer behavior trends, and the ability to achieve
more measurable levels of brand engagement build a
strong, data-based case for how the brand could achieve
greater results with a media plan that increased its online
video presence. The executive dismissed my analysis with a
simple wave of his hand, saying, “We do our own analysis
on our media spend. Each month, I look at the curve of our
television ad spend over time and the curve of our sales
over time and the two are almost perfectly correlated. TV
drives our sales.”
What followed was a conversation about correlation and the
difference between it and causality. But we also talked about
confirmation bias and how the analysis the brand conducted
sought to confirm the CMO’s narrative rather than reveal
the best results for the brand.
While this isn’t a comprehensive list of biases that can affect
analysis, it does feature the most significant challenges
posed at this stage of the analytics journey. The analyst can
guard against each of the biases listed above by being able
to spot them and dutifully working to prevent them from
influencing the analysis.

In Summary: Analyzing for Insights

 PART 5: Storytelling with
Data
Lesson 1 Pictures You See with Your Brain Lesson 2
Evaluation Framework for the Visual Form Lesson
3 Sophisticated Use of Contrast Lesson 4
Ensuring Clear Meaning Lesson 5 Refined
Execution through Visual Polish Lesson 6 On Your
Feet and Getting Your Story Across In Part 5, we dive
headlong into the most important aspect of digital
marketing analytics: transforming the data the analyst
compiled into a comprehensive, coherent, and meaningful
report.
I outline the key characteristics of good visuals and the
minutiae of chart design and provide a five-step process for
analysts to follow when they’re on their feet and presenting
to an audience. The goal is to equip analysts with the tools
they need to tell a compelling and memorable story that
“cuts through the noise” of the overwhelming amount of
information audiences experience every day.
 PART 5 | LESSON 1
Pictures You See with Your
Brain

Five things discussed in this lesson:

People struggle to retain information, presenting

a big challenge to analysts who want audiences to
remember what they say

Several studies illustrate how easy it is for

people to become overwhelmed by information

Analysts must limit the amount of brain activity

their stories require

Understanding how audiences process visual

information can help analysts design better stories

Using preattentive attributes is an effective way

for visuals to “cut through the noise”

Studies show that people forget 80 percent of what they

learn within 24 hours. [114] As analysts who’ve worked so
hard to construct stories by working through tons of data to
craft perfect diamond insights, we want our audiences to
remember what we say. Often, audiences forget information
presented because it slips through gaps in attention or the
information isn’t introduced in a memorable way. So, how
does an analyst – as a Data Designer eager to present their
story in an effective visual form – do that?
To ensure audiences comprehend and retain our messages,
we must understand how they perceive images visually.

Visual Perception and the Door Study

Our eyes are fantastic tools but play a minor part in our
system of visual perception. The eyes feed images to the
brain and it’s the brain’s job to perceive the object being
viewed. In effect, it’s our brains that are seeing. By
understanding this, we can better understand how our
audiences will experience the data stories we present. When
we create images to communicate messages visually, we
should think in terms of how those visuals will register in
their minds, which can be a complicated and confusing
place.
The famous “Door Study” from Daniel Simons and Daniel
Levin illustrates the complexity of our minds and how the
brain’s information processing center, the prefrontal cortex,
can be easily overwhelmed. In the experiment, one
researcher approaches a subject in a park, hands the person
a map, and asks for directions to a nearby place. While the
subject is engaged in the map, a pair of researchers
carrying a broad door walk between the first researcher and
the subject. At that moment, the first researcher ducks
behind the door and is replaced by a second researcher. The
second researcher, a completely different person in different
dress and with a different appearance, continues the
conversation that the first researcher started with the
subject. In most cases, the study found, the subject didn’t
notice they were talking to a different person. [115]
Figure 5.1: The famous “Door Study” from Simons and Levin
in action

Subjects in this experiment didn’t recognize they were

talking to a new person because their prefrontal cortexes
were overwhelmed. Processing a map, the location the
researcher wants to reach, and the path needed to go from
where they are now to where they want to be, require a
great deal of highly involved thinking. The part of our brains
responsible for this level of thought is the prefrontal cortex.
When the prefrontal cortex is engaged in such deep activity,
obvious things – like the fact that a different person is now
standing in front of the Door Study’s subject – can slip
through unrecognized.
What this means for us as Data Designers is clear: If we
create visuals that require highly involved thinking from our
audiences, (i.e., engage their prefrontal cortex), we
introduce the risk that they’ll miss even the most obvious
elements we hope to communicate. Preattentive attributes
are tools we can use to communicate messages without
triggering the prefrontal cortex.
Preattentive Attributes
Preattentive attributes are elements of an image processed
in spatial memory without conscious action. In essence,
preattentive attributes introduce contrast to a visual that
triggers unconscious reactions in our brains. It takes less
than 500 milliseconds for the eye and brain to process a
preattentive attribute of any image. Analysts can leverage
these attributes to make visuals easier to understand while
ensuring their audiences’ prefrontal cortexes are resting
quietly.
Preattentive attributes include size, color, shape, and other
design elements that introduce contrast (what I call
“contrived contrast”). Below are two sets of figures, each
featuring the same number of nines. The set on the right
uses the preattentive attribute of color to set the nines
apart from the other figures, while the set on the left
presents the numbers without contrast. Which of the two
sets makes finding the nines easier?
Figure 5.2: The preattentive attribute of color (hue) at work

Figure 5.2 demonstrates well how color can be used to

introduce contrast, but analysts have many other design
options. Orientation, size, enclosure, width, and intensity are
all ways we can get messages in visuals through our
audiences’ visual perception systems and into their brains.

A Word on the Effect of Neuroscience

British molecular biologist Francis Crick posited that all
actions, emotions and beliefs are the product of physical
activity within the brain that’s consistent between all
people. Marketers can put Crick’s assertion into practice by
measuring a person’s raw initial reactions to something in
the form of electrical activity in their brain through
electroencephalograms (EEGs), which record the brain’s
electrical activity, or functional magnetic resonance imaging
(fMRI), which measures brain activity by detecting changes
associated with blood flow. In this way, data collected from a
consumer’s neurobiology reveal the nonconscious emotions
and responses that affect decision-making. These data allow
marketers to determine how a person really feels about
stimuli like advertisements, packaging, in-store experiences,
or offers devoid of things like social pressure or bias.
Neuroscience has other ways to remove marketers’
guesswork. Eye-gaze detection, which reveals the parts of
ads consumers are drawn to and those they ignore, can
improve ads’ effectiveness. By compiling data on where
people’s eyes are drawn, marketers can highlight attention-
grabbing elements of their creative, eliminate distractions,
and determine where to place critical information related to
products or offers.
Neurobiology can provide clear and consistent insight into
human behavior, but today, its technological limitations are
throttling its applications. Experiments such as Nielsen’s
Consumer Neuroscience studies use medical-grade
equipment and best-in-class technologies to collect data
through multiple EEG sensors that measure memory and
attention (at a rate of 500 times per second), facial coding
to measure expressed emotions, and eye tracking to
measure visual focus on content. The extensive amount of
machinery needed to collect these data, however, relegates
the application of this equipment to the laboratory setting.
Moran Cerf, a leading neuroscientist, believes that for the
practice to reach its full potential, the recording technology
must become much less intrusive so that a person can wear
it constantly “from boardroom to bedroom.” In 2018,
researchers affiliated with Ulsan National Institute of
Science and Technology in Ulsan, South Korea, introduced a
biosensing contact lens capable of detecting glucose levels
in patients with diabetes. [116] According to the research
team, this innovative “smart lens” featured built-in pliable,
transparent electronics that monitor glucose levels from
tears in the eye.
Although the device isn’t commercially available, less
inconspicuous wearable devices have been successfully
launched to monitor glucose, heart health
(electrocardiography), muscle and nerve control
(electromyography), and circulation
(photoplethysmography). As such devices evolve to become
less obtrusive, their practical applications in marketing
become more likely. As discussed previously in this book,
however, how marketers navigate consumers’ privacy
concerns successfully and prove real value for those who
would consent to wearing such devices for marketing
purposes is unclear.

In Summary: Pictures You See with Your

Brain
People struggle to grasp and retain information, which
presents a big challenge to analysts who want audiences to
remember what they have to say. Several studies illustrate
how easy it is for people to become overwhelmed by
information. As a result, analysts must try to limit the
amount of brain activity their stories require. Understanding
how audiences process visual information can help analysts
design better stories. Using preattentive attributes is an
effective way for visuals to cut through the noise. To this
end, it’s no wonder that modern marketing has found
answers to these questions through the use of
neuroscience.
In the next lesson, we’ll explore frameworks that answer the
question, “How should we define good data visualization?”
 PART 5 | LESSON 2
Evaluation Framework for the
Visual Form

Five things discussed in this lesson:

Data visualization is an intricate practice where

one misstep can have catastrophic consequences for
data stories

Evaluating the effectiveness of data

visualizations can be done through a framework that
accounts for each element of the dataviz’s complex
system

For dataviz to be successful, analysts must

ensure the quality of the data collected, define the
objective, design the story, and choose a visual form

Defining good visual form requires an even

deeper analysis and evaluation framework

Good visual form features a sophisticated use of

contrast, clear meaning, and refined execution

Designing effective data visuals is a complicated, difficult

task. As we’ve seen, a long and arduous journey is required
to reach the point where we’re ready to convey our findings
through images. But once we reach this last step in our
journey, the work – in many ways – has just begun. It can be
a frustrating and beguiling task for even the most
accomplished among us. British data journalist, information
designer, and author David McCandless captured this idea
perfectly:

“Visualisation is hard. I’ve written books, created

software, directed films in my career, but visualisation is
by far the most challenging discipline I’ve ever engaged
with. It’s something about the precision needed at every
level, I think. Concept, data, story, design, style – all are
precision arts. In visualisation, they’re stacked one on top
of the other. If one sags or slips, the entire edifice can
collapse.” [117]

Where does an analyst begin when facing a challenge that

bedevils even world-renowned dataviz professionals? Let’s
start by understanding what makes for “good” data
visualization.
Evaluating the Effectiveness of Dataviz
In a brief post to his Information Is Beautiful blog,
McCandless offered an idea of what makes for good data
visualizations through a simple graphic (or, as McCandless
put it, he was simply “thinking aloud in visuals”). [118] This
framework can be a useful point of orientation.
Figure 5.3: McCandless’ take on “What Makes a Good
Visualization?”

The framework expresses the central idea that constructing

visuals is tremendously challenging. Every element must
“work” on its own and, more importantly, mesh together in
a visual form that an audience understands. By including
elements beyond just the “pretty picture” (visual form) to
include all the elements of the creative process, the
framework puts a fine point on the importance that the
analyst bring all three functional roles they play together –
Techie (information), Data Strategist (story and goal), and
Data Designer (visual form).
The framework also provides a clear answer to the question,
What makes a good data visualization? or, What makes a
visualization good? The answer can be elusive. The visual
nature of graphs and charts often allows us to know a
visualization is “good” without being able to articulate what
gives us that feeling. McCandless’ visual gives us a
language for that evaluation by saying the four major
components of any visual – information, story, goal, and
visual form – must be present and effective.

Connecting Data Visual Elements to the

Analysis Journey
I was instinctively drawn to McCandless’ visual framework
for “What Makes a Good Visualization?” because I instantly
recognized how it tied to the Marketing Analytics Process
(MAP) that I had been using (and teaching) for years. Each
element McCandless details is the direct output of a MAP
step.
Figure 5.4: The elements of effective dataviz align perfectly
to the MAP

Let’s take a deeper look at each element of McCandless’

framework and how it aligns with a MAP step.
➔ Goal – This is the functional purpose of the data we
collected and the objective we’re working toward, focusing
our analysis journey. For all intents and purposes,
McCandless’ “goal” is the objective we worked so hard to
identify and clarify during the MAP Plan step.
➔ Information – Information is the data we collected; data
we hope is accurate, deep, and robust. It was the primary
focus of the MAP Collect step. The better that data is, the
better our data visualization will be. Information expressed
through data is a clear, important element of a successful
visual.
➔ Story – Knitting objective and data together is the story.
The story is what takes data and moves it toward our
objective. It’s a narrative we use to guide our audience and
is a collection of the insights we mined from our data. The
story is the outcome of the MAP Analyze step.
➔ Visual Form – The visual form is the image we construct
to express our story to our audience, which uses the data
we collected to drive toward our objective. The visual form
is the ultimate output of the MAP Report step.
We’ve discussed how to design a clear objective, what
makes good data, and how analysis techniques can find
patterns in data to produce an effective story. But as we
enter the MAP Report step, we have yet to define the
required elements of visual form. The McCandless
framework provides little insight. To be sure, we know the
visual form is important – even critical – to our work as
analysts. Yet, how do we know when we’ve created a good
visual?

Understanding the Components of Visual

Form
In my experience, effective visual form does three things
well. It expresses clear meaning by highlighting its
message, it demonstrates a sophisticated use of contrast to
direct the audience’s attention, and it limits distractions
through refined execution.
Clear Meaning Through an effective use of common
elements, including titles, subtitles, and other visual
guides, good visual form clearly communicates the
insight we intend to convey. Although these elements
are familiar (chances are that the first chart you
constructed as an analyst included a title), their use
can be deceptively elaborate. In Lesson 3, we’ll
examine the design elements that help convey clear
meaning in charts.
Sophisticated Use of Contrast By leveraging
preattentive attributes to create contrast, good visual
form separates important data from the rest through
visual context. Contrast takes many forms, ranging
from obvious to nuanced. In Lesson 4, we’ll
investigate each form of contrast and how analysts
can use the technique to create immediate,
instinctive connections with audiences.
Refined Execution Finally, and perhaps most
importantly, the third essential element of good
visual form is refined execution. It’s a deep attention
to detail: the choice of font we make, the way we
apply color, how we allocate space to the page. Each
design decision might appear to have a subtle effect
on our visual, yet these decisions can mean the
difference between an image that conveys its
message effectively and one that doesn’t. In Lesson
5, we’ll delve deeply into the guidelines for refined
execution.
Combined, these three elements make for good visual form.
We’ll use them to provide detailed insight into McCandless’
concept of visual form. As Data Designers, these are the
elements we can affect, control, and improve.

In Summary: Evaluation Framework for the

Visual Form
Data visualization is an intricate practice where one misstep
can have catastrophic consequences for a data story.
Evaluating the effectiveness of data visualizations is
possible through a framework that accounts for each
element of the dataviz’s complex system. For dataviz to be
successful, analysts must ensure the quality of the data
they collect, define the objective, design the story, and
choose the visual form. Defining good visual form requires
an even deeper analysis and evaluation framework. Good
visual form features a sophisticated use of contrast, clear
meaning, and refined execution In the next lesson, we’ll
delve into the first of the three elements of good visual
form: sophisticated use of contrast.
PART 5 | LESSON 3
 Sophisticated Use of Contrast

Five things discussed in this lesson:

Using contrast in visuals is critical for analysts to

communicate insights quickly and effectively

Size, color, shape, and “contrived” (designed)

are methods for contrast that analysts can use when
creating visuals

Using contrast with numbers can be an effective

way for analysts to create effective visuals and to
improve audience understanding

Several techniques from FCB’s John Kenny can

bring contrast and understanding to numbers in an
analyst’s data story

Lost art of sketching visuals with pen and paper

can help analysts become more efficient in the way
they create dataviz

Contrast is an important technique analysts can use to

connect with their audiences instantly. Importantly, contrast
leverages preattentive attributes to trigger reactions in the
brain while leaving the prefrontal cortex quiet and at peace.
This allows analysts to deliver information without evoking
much thought from audiences, thereby keeping their
attention focused. Several types of contrast improve visuals
and communicate insights efficiently.
Size Contrast
Introducing different-size objects on a page captures
attention. The more striking and apparent the size
difference, the more attention the objects will attract. The
visual displayed in Figure 1.8 of this book, which offers a
comparison of daily data-creating human activity by
visualizing the amount of data in a collection of different-
size boxes, is a good example of size contrast.
The variety communicates the volume of data each activity
produces without requiring the audience to digest figures.
The graphic below in Figure 5.7 is from David McCandless
and captures the shocking contrast between the time
American adults spend watching television each year and
the amount of time required to create Wikipedia (data from
Clay Shirky).
Figure 5.7: “Goggle Boxes” an example of size contrast [119]
Color Contrast
Using contrasting colors in an image can be an effective
way to attract attention. Using objects in many colors in a
chart will convey that each object is unique. Color also helps
images “pop” on a page. When analysts select muted colors
for chart backgrounds – a gray shade works well – they can
draw attention to important elements by designing them in
a vibrant color. Different levels of color saturation can
achieve the same result. The example in Figure 5.8
demonstrates how a combination of muted and vibrant
colors can produce color contrast to “lift” elements off the
page.
Figure 5.8: Muted and vibrant colors draw attention to a
chart element
Shape Contrast
As they do with size and color, audiences will detect
different shapes in images. Differences in shapes
communicate the uniqueness of each element instantly.
Icons are a particularly effective form of shape contrast as
audiences understand that different icons represent unique
elements.
Figure 5.9: Icons introduce shape contrast for various data
categories
Contrived Contrast
A final category of contrast is what I call contrived contrast.
Contrived contrast is the use of boxes, callouts, annotations
and other preattentive attributes to distinguish items in a
visual. These are purposeful, planned introductions of
contrast that attract attention. Use this approach with
techniques, including size, color, or shape contrast, or when
those more organic techniques aren’t an option.
The series of LUMAscape charts from the investment bank,
Luma Partners, is a good example of contrast techniques,
but the use of contrived contrast is the most effective
aspect. The LUMAscapes organize the famously complicated
world of advertising technology (“ad tech”) by grouping
similar companies on one page. The size of the groupings
represents the relative number of companies in each area of
the ad tech industry. Categories of groupings are set apart
from one another using different colored labels. But the
introduction of contrived contrast in the form of enclosures
placed around each grouping (as well as boxes placed
around companies, indicating they were recently acquired or
shuttered) is the contrast technique that’s the most
attention-grabbing.
Figure 5.10: The LUMAscape demonstrates contrived
contrast well

Bringing Contrast to Numbers through

Context
Contrast also punctuates figures and facts in presentations.
The difference in technique here is that we won’t use visual
contrast to draw our audience’s attention. Rather, we’ll
introduce context to the figure or fact to lift it from the
surrounding chatter or release its meaning.
John Kenny, the head of planning at FCB Chicago, has
developed a number of effective techniques, collected in
what he calls the “Numerical Comparisons Tool,” to bring
greater context to numbers using principles of behavioral
economics. To demonstrate Kenny’s techniques, we’ll look at
ways to better communicate a large and difficult-to-evaluate
figure: The annual consumption of 7,117,500,000 barrels of
crude oil in the United States.
This figure is purely fact. More than 7 billion barrels of crude
oil (or anything, for that matter) seems like a great amount.
But in actuality, we don’t know the extent of this figure’s
value. Millions and billions are strangely familiar, yet
foreign, concepts to most people. A quick perusal of any
news site will present figures of these sizes in any number
of headlines. Yet, throwing out “billions” alone will go right
over the audience’s collective head. Sure, the figure seems
big, but we don’t have a point of comparison. We lack the
context needed to anchor an evaluation. Kenny’s five
techniques provide the context needed to frame numbers in
a way that makes them understandable.
Let’s look at Kenny’s techniques and how they can provide
context for the figure discussed above.
Translating to Intuitively Understandable Units The
first of Kenny’s techniques is to put the number in a
unit that people understand. In this case, we can take
that idea of 7.1 billion barrels of crude oil and
rephrase it in this way: “Americans use enough oil
each day to make 36 billion plastic water bottles.”
We’re still using the unit of billions, so we haven’t
compromised the large scale of American’s use of
crude oil. But we’ve tied this figure to an item that’s
relatable. Everyday objects, like water bottles, are
effective units. No doubt as you read this book, you
probably have a water bottle near you (but, most
likely, not a barrel of crude oil anywhere to be seen).
To imagine 36 billion plastic water bottles puts the
figure into immediate context and is more
understandable.
Using Familiar Comparisons (Like Distance) Another
technique is to use a familiar comparison like
distance. In this way, we can translate 7.1 billion:
“Each day, Americans use enough oil for 39 trips to
the sun and back.” We see the sun every day (at
least every good day). We might not understand how
far away it is in miles, but we clearly understand the
sun is a long distance from Earth. To think that
Americans use enough oil to make 39 round trips to
the sun must require an enormous amount of oil. This
rephrasing puts 7.1 billion in a better context.
Using Familiar Comparisons (Like Time) Another type
of familiar comparison can be time. Using this
technique, we can rephrase 7.1 billion: “In just two
minutes, Americans use 1 million gallons of oil.” As
we’ve said, people can’t relate to millions and
billions, but they know 1 billion (or 1 million) of
anything is a significant amount. People also know
that two minutes is a relatively short time frame. To
say that we use 1 million of anything in two minutes
makes the expression of size much easier to
comprehend.
Making It Personal Making the figure personal is
another Kenny technique. The 7.1 billion example
made personal would play out in this way: “In a year,
a typical American family uses 70 barrels of oil.” This
technique presents the number in a meaningful way.
While people might not know what 70 barrels of oil
looks like, this technique brings the lofty figure of
total American consumption to a personal level that
is easier to understand.
Finding the Moral Dimension Finally, a high-risk, high-
reward technique from Kenny is finding a moral
dimension to bring context to a figure. In the case of
7.1 billion, we can cast it in a context that elicits an
emotional response: “In less than four decades, the
world’s finite oil supply will be gone forever.”
Phrasing the number like this will undoubtedly make
it more understandable for our audience. It can also
polarize people. Before using this approach, an
analyst must have some understanding of how the
audience feels about the subject to minimize the risk
of offending people or turning them off.
The process of identifying a goal, collecting data, and
creating a story requires facts and figures. Using Kenny’s
techniques will help ensure our numbers are relatable,
meaningful, and understandable.

A Word on the Importance of Sketching

Visuals
Sketching is, unfortunately, a bit of a lost art. Too often, we
have the attitude that we must rush to our computer to
generate graphics. If we took a moment with pen and paper
to sketch our early ideas, however, we could design with
much more freedom. We’d no longer be limited by our
understanding of the application or its design restrictions.
Through sketching, we’re able to explore contrast in a space
bound only by the edges of our creativity. The great data
visualization artists sketch before placing fingers on
keyboards. Sketch your ideas first, then bring them to life
using dataviz tools. It’s a good practice that will make you
more efficient when designing visuals and telling stories
with data.
Figure 5.11: My dataviz sketchbook
In Summary: Sophisticated Use of Contrast
Bringing contrast to visuals is critical for analysts to
communicate insights quickly and effectively. Size, color,
shape, and “contrived” (designed) are contrast methods
that analysts can use when creating visuals. Adding contrast
to numbers can be an effective way for analysts to ensure
people understand charts. Several techniques from John
Kenny of FCB can bring contrast and understanding to
numbers in data stories. The lost art of sketching visuals
with pen and paper can help analysts create dataviz
efficiently and can reveal new approaches to contrast that
analysts wouldn’t have thought of while sitting at their
computers.
In the next lesson, we’ll delve into the second of the three
elements of good visual form: how to convey clear meaning
through chart elements.
PART 5 | LESSON 4
 Ensuring Clear Meaning

Five things discussed in this lesson:

Ensuring clear meaning in our messages hinges

on other elements of our visuals being successful

Using titles and subtitles properly is critical

when designing charts

Highlighting messages visually is the most

effective way to convey stories

Using visual cues, applying annotations, and

labeling items directly are simple techniques that
help to ensure clarity and reduce confusion

Eliciting an emotional response from an

audience through a story is a high-risk, high-reward
approach for ensuring memorable messaging

Most of the building blocks for creating clear messages

come early in the data analysis journey. The process of
identifying a concise Marketing Objective, collecting data,
and finding interesting patterns in data enrich our content.
When completing these steps correctly, we have an
excellent opportunity to present inherently clear messages.
Design techniques, however, can enhance our messages by
enriching visuals. Using titles, subtitles, and other visual
guides effectively ensures our designs present insights
clearly. These elements are common, but their use can be
deceptively elaborate. Let’s look at how to apply them
properly.
Properly Titling Your Visual
The most important thing analysts can do is to write
thoughtful, purposeful headlines and subtitles for their
charts. Headlines and subtitles are included on nearly every
dataviz, but Data Designers typically take them for granted.
Despite this indifference, these are important for audiences.
They’re typically among the first things audiences will read
on a chart. They offer the savvy Data Designer a golden
opportunity to start the audience experience positively.
Headlines
Chart headlines should be in plain English and answer,
“What am I looking at here?” For example, a headline such
as “Product X sales over the past five years” introduces a
chart clearly. It orients the audience to the data in the
image, importantly sparing them the prefrontal cortex-
demanding task of figuring that out for themselves.
In contrast, a chart with the headline, “How have we done
recently?” has the opposite effect. While written in plain
language, this headline triggers more questions than it
answers. How is performance defined? What does ‘recently’
mean? Should we focus on data from all the years included
in the chart, or just those more recent? We lose audience
attention when headlines trigger questions.
Place the headline above and left-aligned with the dataviz.
Position the headline horizontally and make it bold in a
larger font than the graphic so it stands out against other
chart elements (e.g., subtitle and labels).
Basic Guidelines Pro Tips

● Use clear, concise ● Avoid using acronyms or

language in the headline abbreviations in your
● Explain plainly and headline
unambiguously what the ● Avoid using clever
chart presents headlines, regardless of how
● Place the headline above comfortable you are with the
the dataviz, aligned to the data (you can bring
left side of the chart personality to your talk
● Print the headline through your presentation
horizontally, make it bold, style)
and a few font sizes larger ● Flip rapidly through the
than other chart elements pages of your presentation,
reading only the chart
headlines. Did you include
all the topics you intended to
cover? Are they in the proper
order?
Subtitles
Every chart should contain a subtitle printed directly below the
headline and above the dataviz. The subtitle is a succinct
description of the insight an analyst wants the audience to
take away from the visual. Through the subtitle, the analyst is
effectively telling the audience, “This is what you should think
once you have looked at the data in my chart.” Like the title,
the subtitle should be in plain, unambiguous language and
shouldn’t contain acronyms or abbreviations.
A subtitle such as “Last year’s sales make up for losses
suffered in the previous four years” makes it immediately
clear what the chart conveys. Conversely, a subtitle that
proclaims, “We are doing great!” is simple but doesn’t connect
to the data in the chart. Such a statement is “intellectually
blank,” lacking any insight into the chart’s underlying
message. Like a poorly designed title, this latter subtitle will
leave the audience with more questions than answers.
Proper use of subtitles is important for a number of reasons. It
allows analysts to control a chart’s message. By printing the
insight on the page, we’re introducing the truth we believe the
visual holds. Whether the audience readily accepts that truth
or is skeptical of its veracity, we’ve made the statement plain
and clear. In doing so, an analyst can substantiate that insight
by pointing to corroborating patterns in the data. In addition,
the meaning of the chart is clear to those who weren’t at the
presentation or didn’t hear the narrative that accompanied the
visual when the presentation lands on their desk or in their
inbox.
The subtitle shouldn’t be in a bold font so that it stands apart
from the headline. It should be a few font sizes smaller than
the headline.
Basic Guidelines Pro Tips

● Include a subtitle on every ● Avoid using acronyms or

chart you make abbreviations in your subtitle
● Write your subtitle in plain ● Flip rapidly through the
language that concisely pages of your presentation
conveys the insight the reading only the chart
audience should take from subtitles. Did you include all
your chart of your insights? Are they in
● Avoid intellectually blank the proper order?
statements in your subtitles
● Place the subtitle directly
below the headline, above
the dataviz, and aligned to
the left side of the headline
● Print the subtitle
horizontally in a “normal”
(i.e., not bold) font a few
sizes smaller than the
headline
Highlighting Messages Visually
After properly headlining and subtitling their charts, analysts
can introduce other design elements that will guide audiences
to data that supports the charts’ professed insights.
Using Visual Cues
Visual cues – arrows, boxes, and shaded areas – can be
extremely effective additions to visuals. Think of these visual
cues as tools that direct audiences to important figures, data
points, and patterns. As we learned about visual perception in
Lesson 1, these cues stand out to audiences as preattentive
attributes. Their brains can’t help but notice anything we point
out with a tastefully designed arrow (or encase in a box, or
place in an area shaded light gray, etc.).
Figure 5.5: Visual cues will direct your audience’s attention to
the important areas of your chart

To avoid overloading the visual with “look here!” cues, a good

practice is to use no more than one cue on a chart. In this
way, we guide our audience to our important data while
ensuring that it remains critical. When everything on our chart
is called out as important, nothing will be significant.
Remember that change – not sameness – catches our
audience’s attention.
Basic Guidelines Pro Tips

● Become familiar with ● Once you find a visual cue

multiple types of visual cues that works effectively, reuse
● Gauge the need for visual it in other charts that benefit
cues by presenting your data from similar attention-
with and without them. If the grabbing tools
effect is positive, keep them
(and remove them if they’re
unnecessary)
● Apply visual cues
frequently, yet rarely exceed
more than one cue per chart
Applying Annotations
Like visual cues, think of annotations as tools that direct
attention to important areas of a chart. Unlike image-based
visual cues, however, annotations are small bodies of text that
inform the interpretation of figures, data points, and patterns.
Write annotations in clear, plain language. They shouldn’t
appear as labels. Instead, express them through short
sentences or phrases. As annotations should be tied to specific
areas of the chart data, use a thin line to connect the
annotation to the data it explains, whether that data is
contained in a single point or a range of points.
Figure 5.6: Annotations on a chart can identify data that
supports the subtitle’s insight [120]

When chart data is dense or insights are subtle, annotations

can make clear how annotated data supports the insight
expressed in the chart’s subtitle. This is particularly important
for ensuring consistency or interpretations when people see
the chart outside the presentation. In the same way the
subtitle acts as a written record of the chart’s insight,
annotations serve as documentation of the insight’s
substantiation.
Basic Guidelines Pro Tips

● Separate annotations from ● Use annotations when you

labels by writing them in know a presentation will be
short sentences or phrases shared outside the viewing
● Use thin lines to connect audience
each annotation to the data
point or range of data it
explains
Labeling Items Directly
Ensuring audiences focus on chart data means eliminating
unnecessary visual elements that draw the eye away. One of
the most distracting visual elements, if not the most
distracting visual element, is a familiar item Data Designers
use extensively: the chart legend. Typically, the legend sits to
the upper right of the chart and contains the key to connecting
colors, lines, and other visual components of the chart to their
names.
While the explanations in the legend seem intuitively helpful,
the placement – separate from the visual – creates havoc for
an audience. A legend forces people to dart their eyes from
each chart element to the legend and back again to
understand what the data represents. The mental action of
tying a color and label in the legend to its corresponding data
in the chart is a task that only the prefrontal cortex can
accomplish. As we’ve learned, engaging audiences’ prefrontal
cortexes forfeits their attention.
Use direct labeling rather than including a legend on a chart.
Place labels at the end of lines, on pie chart slices, or at the
base of bars. One- or two-word labels will keep the chart
clutter-free. Use color to separate important labels from less
critical support data. A label written in an eye-catching red will
stand out against gray labels. Coordinating label colors to
match the color of the data they describe optimizes
preattentive attributes and helps audiences connect labels
with data.
By labeling chart elements, we hold audiences’ attention,
allowing us to deliver clear, distraction-free messages.
Basic Guidelines Pro Tips

● Use labels instead of ● Coordinate label colors to

legends on your charts match the color of the data
● Place labels at the end of they describe
lines, on pie chart slices, or
at the base of bars
● Write one- or two-word
labels to minimize clutter in
the chart
● Use colored labels to
distinguish important data
from support data
A Word on Leveraging Emotions
Making emotional connections with audiences through the
data we describe, the insights we reveal, and the style in
which we present can help ensure our messages are
memorable. Consistently eliciting the same emotion
throughout chart presentations will help ensure audiences
understand our messages.
Reciting facts won’t secure emotional connections with
audiences. Plan the emotions you want to elicit with each
chart in your presentation and rehearse drawing those
emotions from your audience. One caveat: The emotion must
be authentic. Eliciting emotions from an audience is a powerful
venture but can alienate people quickly if they don’t agree
with the emotional response you’re trying to manufacture (or,
perhaps worse, react with a different emotion). Use this high-
risk, high-reward approach when you’re comfortable with your
material and how an audience will react to your presentation.
In Summary: Ensuring Clear Meaning
Being singular in message is one way analysts can ensure
clear meaning. Proper use of titles and subtitles in charts is
the most important thing analysts can do when designing
charts. Highlighting messages visually is the most effective
way to convey stories as using visual cues, applying
annotations, and labeling items directly are simple
techniques analysts can use to ensure clarity and reduce
confusion. Eliciting an emotional response from an audience
through stories is a high-risk, high-reward approach for
ensuring memorable messaging.
In the next lesson, we’ll look at the final element of good
visual design: refined execution.
PART 5 | LESSON 5
 Refined Execution through
Visual Polish

Five things discussed in this lesson:

Refined chart design requires particular

attention to the details of fonts, labels, lines, and
other elements

Color choice can sharpen an analyst’s story, but

can cause distractions if not applied properly

Several well-defined, tried-and-true harmonic

color themes are available for analysts

Analysts shouldn’t feel beholden to a single data

visualization tool when building charts and rather use
multiple tools to ensure high-quality designs

Several tests can help analysts evolve and

improve their data visualizations

If contrast calls attention to items and simple chart

elements help convey clear meaning, the role of visual
polish is to ensure items that support critical data sit quietly
in the background. Refined execution can be best thought of
as a collection of rules and recommendations that eliminate
distractions from charts. These design guidelines cover
everything from color choice to font usage to specific
elements of charts (e.g., labels, lines, etc.). By incorporating
refined execution In their practices, analysts can ensure
their charts are distraction-free.
Let’s explore the elements of refined execution.
Color Theory Application in Charts
Color is often the first bit of data our eyes perceive in
visualizations. For this reason, it’s critical to examine color’s
influence on audience perception, as well as its influence on
other colors. Analysts must create harmonious visuals by
being thoughtful about color choices. To do so, it’s first
important to understand a few key concepts about color.
Hues
Think of hue as the dominant color of an item. The hue of a
cloudless sky at mid-day, for example, is blue. Hue is
technically defined as “The degree to which a stimulus can
be described as similar to or different from stimuli that are
described as red, green, blue, and yellow.” [121] Different
hues can be harmonic (e.g., blue and green) and contrasting
(e.g., black and white).
Saturation and Value Color saturation and color value
refer to the intensity of color in an image. In technical
terms, saturation and value are the bandwidth of
light from a source. By increasing (making darker)
and decreasing (making lighter) the saturation or
value of a hue, analysts can create a variety of color
shades that work well together. The difference
between color saturation and color value is subtle.
Color saturation varies by adding shades of gray
(ranging from white to black) to the hue to affect its
vibrancy. Value, on the other hand, affects hue by
eliminating color from the hue (i.e., making the hue
more or less transparent). Color saturation and color
value are important and effective tools for Data
Designers.
The Color Wheel The color wheel represents the
spectrum of primary, secondary, and tertiary colors
(more on those later) by wrapping them onto a circle
in a logically arranged sequence. Developed by Sir
Isaac Newton in 1666, the color wheel allows for the
selection of colors in ways that ensure harmony and
eye-pleasing themes.
Figure 5.12: Newton’s original design for the color wheel

Primary Colors Red, blue, and yellow are the three

primary colors on the color wheel. All other colors are
derived from these three hues. The primary colors
can’t be formed through a combination of any other
colors.
Secondary and Tertiary Colors Green, orange, and
purple – secondary colors – are formed by mixing
primary colors. Colors formed by mixing a primary
color and a secondary color are the tertiary colors
included on the color wheel: blue-green, red-violet,
yellow-orange, etc.
Distinguishing Colors in Tools Color can be specified
in different ways. Each data visualization tool allows
analysts to identify, select and modify colors using
one (or more) of three common models.
Figure 5.13: Common color models most dataviz tools use

➔ RGB color model is an additive color model in which red,

green, and blue are combined to create a broad array of
colors. The name comes from the initials of the three
additive primary colors: red, green, and blue.
➔ CMYK color model (also known as “process color” or
“four color”) is a subtractive color model, used in color
printing, and is also used to describe the printing process.
CMYK refers to the four inks used in color printing: cyan,
magenta, yellow, and key (black).
➔ Hex code describes the composition of a certain color in a
specific color space, usually RGB. The first value pair refers
to red, the second to green, and the third to blue, with
decimal values ranging from 0 to 255, or in hexadecimal 0
to FF (#RRGGBB).
Harmonic Color Themes
Two or three colors – plus, the measured use of black, a
neutral gray, and white – provide an ample (and
manageable) set of hues for any presentation. To ensure
harmony in color choice, analysts should choose from
several established themes. These themes should include
colors that work together to enhance visuals, rather than
create distractions.
Figure 5.14: The examples of harmonic color themes

Monochromatic Color Theme Monochromatic color

themes use one hue to minimize color distraction on
the page. Through variations in saturation and value,
the selected hue can be expanded to fulfill the color
needs of any presentation or visual. Choose one color
that has particular relevance and round out the color
palette with variances of saturation and value.
Analogous Color Theme Analogous color themes use
Newton’s logical order of color properties to create
eye-pleasing combinations. Select colors that are
next to one another on the color wheel to create color
harmony. Three analogous colors will typically
provide all the variations a Data Designer will need,
but four colors can be used to expand the theme
further.
Complementary Color Theme Complementary color
themes use Newton’s insights into color properties to
create a palette of colors that work well together.
Choosing colors that are directly across from one
another on the color wheel will produce harmonious
pairs of hues to form the basis of the theme.
Variations in saturation and value can round out the
palette.
Split Complementary Color Theme Using colors on
either side of complementary colors on the color
wheel is another technique for creating harmonious
color palettes. These split complementary color
themes are effective at matching a set of base colors
for a presentation to a contrasting hue that can be
used to “pop” elements of a chart to grab attention.
Triadic Color Theme Triadic color themes are created
by picking three colors from the color wheel that are
perfectly spaced from one another. The natural
properties of these colors produce vibrant contrast
that works well in a palette, sometimes surprisingly.
Additional Color Considerations
According to the nonprofit organization, Colour Blind
Awareness, colorblindness (i.e., color vision deficiency or
CVD) affects approximately 1 in 12 men (8 percent) and 1 in
200 women (0.5 percent) in the world. [122] This means more
than 300 million colorblind people (about 4.3 percent of the
entire population) are alive today. It’s likely that someone in
your audience is colorblind and struggles to distinguish
different hues (e.g., red, yellow, blue, and green). Limit your
color palette to two or three hues and use different degrees
of saturation, as well as black, gray, and white to round out
the palette. These techniques allow analysts to introduce
contrast in ways that won’t exclude colorblind people in the
audience.
Elements of Refined Design
Refined execution requires attention to detail. Every
element we place on our charts will affect our audience’s
ability to understand the visual. Adopting a set of guidelines
that ensures the proper application of these details will
result in better charts.
Fonts
Font choice affects the legibility of graphs. Fonts can help
establish a tone and make it easier for audiences to
understand chart elements. Comic Sans, handwriting, and
other “fun” fonts don’t establish an appropriate tone for a
business presentation. It’s more important to be seen as
credible than it is to express personality. The serious,
practical font choices of well-established and respectable
organizations, including The New York Times and The Wall
Street Journal, are appropriate benchmarks. You’ll never find
a headline in these publications in Groening.
Unadorned fonts, including Helvetica and Arial, improve
legibility. In addition, being consistent in font treatments
(e.g., using the same font for every chart headline in the
presentation) helps audiences to understand your content.
 Basic Guidelines Pro Tips

● Choose serious, ● Adopt a presentation style

businesslike fonts rather that mirrors The New York
than “fun” fonts ● Limit the Times and The Wall Street
number of fonts in your Journal rather than USA
presentation to two, possibly Today
three ● Apply fonts
consistently for each chart
element throughout your
presentation (e.g., headlines
are always the same font) ●
Choose unadorned fonts like
Helvetica and Arial

Labels As discussed in Lesson 3, labels ensure clear

meaning in charts, and analysts should use them
instead of keys and legends. Another best practice will
keep a chart clutter-free: Never rotate text in a
presentation. All text – particularly axis labels – should
be horizontal. Text that’s rotated vertically is difficult to
read and a distraction.
Basic Guidelines Pro Tips

● Label chart elements ● Color your labels in the

clearly same hue as the data it
● Eliminate keys from your describes
presentations
● Present all text
horizontally (i.e., never turn
text vertically or on an
angle)
Lines
Whether they’re straight or in an arc, lines should be
unadorned and distraction-free. Avoid shadows, 3D, dashes,
and dotted lines. In addition, don’t use data markers in lines,
as they call attention to specific data points to the detriment
of a line’s trend.
Basic Guidelines Pro Tips

● Never add shadows or 3D ● Color lines conveying

attributes to lines important data in vibrant
● Present data with only hues and all other lines in
solid lines (e.g., do not use visually mute gray hues
dashed lines in charts) ● Use a maximum of three
● Never add data marker to four lines and keep the
symbols to lines style simple
Other Chart Elements
It’s also important for analysts to make conscious choices
about other elements of their charts. Any element on a page
can improve or detract from a visual’s ability to convey its
intended message. Images, icons, and placement of the visual
can affect audience perception.
Ensure that images convey authentic emotions. Avoid
including stock photos of people in presentations, as these
images appear spurious at best and ridiculous at worst.
Prohibit the use of Clip Art as these comically designed images
reduce an audience’s respect for a presentation. Use icons
rather than Clip Art. Icons should be simple as ornate versions
lose detail in small sizes.
Space on a page affects an audience’s visual perception.
Analysts should place elements consistently from page to
page in their presentations. In other words, analysts should
use the “position” coordinates in their design tools to ensure
that charts are always in the same spot. Analysts can further
improve design consistency by ensuring charts, icons, and
images are the same size (i.e., height and width) throughout
their presentations.
Basic Guidelines Pro Tips

● Avoid using stock images ● Place objects (e.g., charts)

of people in the same spot on your
● Never, ever, EVER use Clip presentation pages to create
Art in presentations ● When harmony ● Ensure charts,
using icons, opt for simple icons, and images are a
images (a lifelike car icon consistent size (i.e., height
doesn’t look like a car when and weight) throughout your
shrunken) presentation
Testing to Improve the Visual
Graphics are living, evolving depictions of insights. Every new
bit of information and every audience reaction can help inform
and improve chart design. To this end, analysts should perform
three important tests on their graphics periodically to ensure
each iteration improves clarity.
The Spartan Test
To conduct the Spartan Test, examine each element of the
graphic to ensure it makes a positive contribution to the visual
as a whole. Delete any unnecessary elements methodically. If
you need to restore an element you delete, “Ctrl+Z” will undo
the delete action and return the element to its rightful place.
Keep in mind that every graphic should include the fewest
elements possible. The Spartan Test is a great way to see that
less is more when creating data visualization.
The ‘Peek Test’
To conduct the “Peek Test,” set the graphic aside for a few
moments. If the graphic is printed on a piece of paper, turn
the paper over so the graphic is face down. If the graphic is on
a computer screen, switch to a different application or walk
away from the computer. You can then return a short time
later and look at the graphic. Where do your eyes go? If you’re
drawn to an area you think is important, you can rest assured
your audience’s eyes will go there, too. If, however, your eyes
pick up some other aspect of the graphic, you should revise it.
A recalibration of preattentive attributes could change the
graphic’s focus. The Peek Test helps ensure the audience will
see what you want them to see first.
The ‘Colleague Test’
The most important test is what I call the “Colleague Test.” In
this test, show the graphic to a colleague who hasn’t had any
association with the project. Make sure your colleague knows
very little about the data you collected, your objective, and
the story you’re trying to tell. The less context your colleague
has, the more value the test will provide. Show the
visualization to your colleague but, ideally, don’t say anything
about it during the test. Simply ask your colleague for the
meaning they take away from the visual. If your colleague
understands the design, certainly people who are familiar with
the project and stakeholders will also likely comprehend it. The
Colleague Test will tell you if your visual is presentation-ready
or requires more work.

A Word on Combining Tools to Refine Visuals

A key to producing refined visuals is the design flexibility of
the tools we use. When we attempt to design a graphic in a
single tool, we’re limited to the capabilities of that tool. As
Data Designers, we should feel capable of using any number
of tools to create the best visuals we can. After all, we want
our designs to be based on our creativity rather than loyalty to
a tool. You can export basic visuals from desktop applications
such as Excel; analysis tools such as R; and online tools,
including iCharts, into PowerPoint, Keynote, and Google Slides.
After importing visuals, you can fine-tune headlines, subtitles,
labels, annotations, and other design elements. To further
enhance visuals, consider using sophisticated design tools like
Adobe Photoshop or Illustrator, which allow you to change
colors, crop images, and resize visuals.

In Summary: Refined Execution through

Visual Polish
Refined chart design requires particular attention to the details
surrounding fonts, labels, lines, and other elements. Colors
can sharpen an analyst’s story but can cause distractions if
not applied properly. Analysts can choose from several well-
defined, tried-and-true harmonic color themes. Other color
considerations include how colorblind audiences will
experience an analyst’s color choices. Analysts shouldn’t feel
beholden to a single data visualization tool when building
charts. Instead, they should use a combination of tools to
ensure high-quality designs. Several tests can help analysts
improve their data visualizations.

In the next lesson, we’ll explore the final step in our

communication journey: presenting a data story to an
audience.
 PART 5 | LESSON 6
On Your Feet and Getting Your
Story Across

Five things discussed in this lesson:

How analysts present their data stories can

affect an audience’s understanding as much as the
content can

David McCandless, the data visualization artist,

has an effective approach to data storytelling earned
from years of experience

McCandless’ approach to data storytelling can

be defined as five distinct steps that I call “The
McCandless Method”

Analysts can apply other elements of

presentation style using simple guidelines

Like every element of marketing analytics,

presenting data visualizations is a skill that analysts
develop through good habits

The final step in the communication journey is for analysts

to stand on their feet and deliver their data stories to
audiences. The activities surrounding objective setting, data
collection, analysis, and story design led to this point, and
its importance cannot be understated. How analysts present
their data stories can have as much effect on audience
understanding as the content.
Luckily, a simple process helps analysts present data stories
effectively. Performing each step in the proper sequence
allows analysts to communicate insights clearly, focusing
audience attention on their work.

‘The McCandless Method’ of Data

Presentation
The proper approach to presenting graphics can be a
learned behavior. Studying dataviz professionals carefully
reveals a consistent set of steps analysts can use to lead
audiences through data stories. I’ve developed a process I
call The McCandless Method. Named after David
McCandless, this five-step process is based on the
successful way he has presented data visualization.
Figure 5.15: The five steps in The McCandless Method of
data presentation

Step 1: Introduce the Graphic by Name The first step

in The McCandless Method is to introduce the graphic
by name, allowing audience members to focus their
attention on the graphic rather than you. The name
should be clear, unadorned, and obvious. Ideally, the
name should be the chart headline (see Lesson 4 for
more direction on using chart headlines). Often,
analysts will give their charts clever names to make
them memorable, but the names don’t set up the
stories. Seasoned analysts can use this technique,
but novices should stick to easy-to-comprehend chart
names.
Step 2: Answer the Obvious Questions The second
step in The McCandless Method is to answer the
obvious questions that will jump to audience
members’ minds before they can ask those
questions. By doing so, analysts can minimize
possible confusion and ensure audiences pay
attention to analysts’ stories rather than try to
decode chart elements.
As analysts share their presentations, they can identify the
questions people will likely have. The more experience
analysts have in presenting charts (formally or informally),
the more adept they’ll become at identifying elements that
require explanation. Obvious questions analysts should
address immediately include what contrast – color, size,
shape, or contrived – means in the graphic, what data is
presented (i.e., the axis labels of the chart), and where the
data came from.
Step 3: Give Away the Insight The third step in The
McCandless Method is to state the insight from the
graphic. An analyst should reveal their story by
telling audience members what they’ll see (or feel)
before they substantiate the insight. The audience
may be curious, even disbelieving, at this point. By
sharing the insight before diving into supportive
facts, the analyst will benefit in two ways: (1) They
can reveal an insight before the audience becomes
distracted with facts and figures, and (2) Sharing the
insight upfront paves the way for the analyst to
deliver the supporting facts the audience seeks. The
insight the analyst should reveal is the chart’s
subtitle (see Lesson 4 for tips on chart subtitles).
For years, I had this step out of order. I presented my
evidence patiently and dutifully before revealing (with great
aplomb) the insight of my analysis. I believed that by
building tension and methodically constructing my story, I
was guiding my audience down the primrose path of my
logic to its dramatic conclusion.
Eventually, I learned it didn’t matter how airtight my story
was or how memorable my facts were. The longer it took me
to reveal the insight, the less likely it was that I had my
audience’s attention. Sometimes, the mounting body of
evidence overwhelmed people, leaving them unable to
concentrate in the critical moment of my big reveal. Other
times, someone disagreed with a fact I delivered, causing
that person to shut down and disregard what I said next.
When I reversed the order of my story from “evidence →
insight” to “insight → evidence” my presentations became
more effective than ever.
Step 4: Provide Examples Once the analyst
establishes the conclusion they want the audience to
reach, they can provide the substantiation the
audience will crave by providing examples. The order
in which the analyst delivers their evidence is
important and should be done in a way that
facilitates agreement from the audience. The analyst
can do this by moving hard-hitting facts to the
beginning of their narrative when facing a skeptical
audience. This will allow the presentation to start off
on a high note and earn momentum for the evidence
that follows. If the audience is friendly, the analyst
has the latitude to present evidence in a way that
ends with a dramatic, “no doubt about it” fact that is
as entertaining as it is illuminating.
Step 5: Close
Finally, it’s important to transition to the next point in the
presentation. Foreshadowing the next point (e.g., “on the
next slide you’ll see something even more amazing”),
calling out a fact that links to something on the next slide
(e.g., “I’ve shown you that 90 percent of U.S. adults visit
YouTube each month – that will be important context for our
next slide”), or telling audience members why the graphic
matters to them (e.g., “This is important because it
represents a new opportunity for your company to increase
household penetration”) are all effective ways to transition
to the next point while bringing your audience along with
you. Close the presentation of your visual in the way that
fits your needs and continues the momentum to the next
topic of discussion.
The McCandless Method in Practice
McCandless has demonstrated this process repeatedly, but
one example of his that stands out is from a TED Talks
program, David McCandless: The beauty of data
visualization. You can easily find video of this presentation
on YouTube. Here is the narrative of McCandless’
presentation:

This is the ‘Billion-Dollar-O-Gram’ and this image arose

out of frustration I had with the reporting of billion dollar
amounts in the press. That is, they're meaningless
without context. ‘$500 billion for this pipeline, $20 billion
for this war’ doesn't make any sense. So the only way to
understand is visually and relatively.
So I scraped a load of reported figures from various news
outlets and then scaled the boxes according to those
amounts. And the colors here represent the motivation
behind the money: purple is fighting and red is giving
money away and green is profiteering.
And what you can see straight away is you start to have
a different relationship to the numbers. You can literally
see them. But more importantly you start to see patterns
and connections between numbers that would otherwise
be scattered across multiple news reports.
Let me point out some I really like: OPEC's revenues
green box here, $780 billion a year. And this little pixel in
the corner (representing) $3 billion? That's their climate
change fund.
Americans are incredibly generous people. Over $300
billion-a-year donated to charity every year. Compared
with the amount of foreign aid given by the top 17
industrialized nations at $120 billion.
And then of course the Iraq war predicted to cost just $60
 billion back in 2003 and then mushroomed slightly after
Afghanistan and Iraq now to $3,000 billion.
So now it's great because now we have this texture and
we can add numbers to as well. So we say, ‘well a new
figure comes out and, let’s see, African debt…how much
of this diagram might be taken up by the debt Africa
owes to the West?’ Let’s take a look: so there it is to $227
billion is what Africa owes.
And the recent financial crisis, how much of this diagram
might that figure take up? What did that cost the world?
Let’s take a look at that. Dooosh! I think is the
appropriate sound effect from that much money: $11,900
billion.
So by visualizing this information we’ve turned into a
landscape that you can explore with your eyes. Kind of
map, really. An ‘Information Map’. And when you’re lost in
information, an Information Map is kind of useful.

Each step in The McCandless Method detailed previously is

in this narrative.
Step 1: Introduce the Graphic by Name McCandless
begins by sharing the graphic’s name. Although the
Billion-Dollar-O-Gram gives little insight into what the
graphic represents, McCandless does a good job of
explaining the topic in his introduction. Again,
McCandless is a professional, published data
visualization artist. Until we reach that rank, we
should stick with less colorful names for our charts.
What is important is that the audience members can
understand what they’ll be hearing, and their
attention is on the graphic – not on McCandless.

This is the ‘Billion-Dollar-O-Gram’ and this image arose

 out of frustration I had with the reporting of billion dollar
amounts in the press. That is, they're meaningless
without context. ‘$500 billion for this pipeline, $20 billion
for this war’ doesn't make any sense. So the only way to
understand is visually and relatively.

Step 2: Answer the Obvious Questions Next,

McCandless explains the data’s source, what the
colors on the image mean, and what the different
boxes represent. Given our understanding of
preattentive attributes, the elements McCandless
calls out would jump off the screen at the audience.
Answering these questions immediately allows the
audience members to focus on what McCandless is
saying, rather than dividing their attention between
the data story and the unresolved questions they
have about the chart.

So I scraped a load of reported figures from various news

outlets and then scaled the boxes according to those
amounts. And the colors here represent the motivation
behind the money: purple is fighting and red is giving
money away and green is profiteering.

Step 3: Give Away the Insight McCandless then

reveals the insight from his image. He makes his
point concisely and leaves no room for interpretation.
McCandless has now established a statement of fact
that audience members want to have substantiated,
thereby setting himself up to deliver the resolution
they seek in the next few moments.
 And what you can see straight away is you start to have
a different relationship to the numbers. You can literally
see them. But more importantly you start to see patterns
and connections between numbers that would otherwise
be scattered across multiple news reports.

Step 4: Provide Examples Next, McCandless shares

examples that illustrate the insight he provided. As
discussed earlier, if McCandless had flipped the order
of these last two steps, he could have told a
masterful story, but the audience would have gotten
lost trying to determine where he was headed or,
perhaps worse, drawn a conclusion different from his.
This would have left audience members confused and
impatient when any fact McCandless presented didn’t
fit the story they’d constructed in their heads. By
following the “insight → evidence” approach,
McCandless dramatically increases the likelihood that
the audience will be receptive to his narrative.

Let me point out some I really like: OPEC's revenues

green box here, $780 billion a year. And this little pixel in
the corner (representing) $3 billion? That's their climate
change fund.
Americans are incredibly generous people. Over $300
billion-a-year donated to charity every year. Compared
with the amount of foreign aid given by the top 17
industrialized nations at $120 billion.
And then of course the Iraq war predicted to cost just $60
billion back in 2003 and then mushroomed slightly after
Afghanistan and Iraq now to $3,000 billion.
So now it's great because now we have this texture and
we can add numbers to as well. So we say, ‘well a new
 figure comes out and, let’s see, African debt…how much
of this diagram might be taken up by the debt Africa
owes to the West?’ Let’s take a look: so there it is to $227
billion is what Africa owes.
And the recent financial crisis, how much of this diagram
might that figure take up? What did that cost the world?
Let’s take a look at that. Dooosh! I think is the
appropriate sound effect from that much money: $11,900
billion.

Step 5: Close
Finally, McCandless takes this graphic and makes it much
more personal for the audience, explaining why it’s
important and ending his presentation.

So by visualizing this information we’ve turned into a

landscape that you can explore with your eyes. Kind of
map, really. An ‘Information Map’. And when you’re lost in
information, an Information Map is kind of useful.

These five steps will help ensure audiences understand

analysts’ graphics. It would be a shame to put so much
effort into researching, constructing, and polishing visuals,
only to lose people in those last few yards. Analysts must
communicate effectively by following The McCandless
Method for every graphic they present.
A Word on Presentation Style
The style in which we stand on our feet and present our
visuals – the last few yards of our communication journeys –
is very important. If we don’t present our data stories in a
way that conveys confidence and shifts attention from us to
the substance of our work, we run the risk of losing our
audience and wasting all the work we’ve done leading up to
that moment.
The McCandless Method will set you up for a successful
presentation. It will ensure you present data in the right
way. It’s important for you to build that method into your
practice. You should also build other critical habits into your
practice of presenting graphics. The basic guidelines
presented here will help ensure your presentation is
successful. Exceeding those guidelines to include the pro
tips detailed below will take your presentation style from
good to great.
Basic Guidelines

● Test the room’s equipment (e.g., overhead projector,

speakers, etc.) thoroughly and have an alternative to
your electronic presentation if the equipment fails
● Never apologize in your intro (“Sorry, I’m running
late!” or “My flight got in late last night, so I apologize for
my low level of energy”) because it sets a negative tone
for your presentation
● Plan an effective opening and powerful close to your
presentation, as the first and last thing audiences hear
stick with them the most
● Carefully craft transitions from one slide to the next to
ensure continuity in your story
● Be sensitive to your audience’s understanding of the
subject matter by avoiding acronyms and jargon
● Never expect a visual to speak for itself: Use The
McCandless Method for each chart
● Never assume the message is “too obvious” – state
your point and ensure it registers on your audience’s
faces
● Maintain eye contact with your audience by turning to
the audience when you speak – not the visual
 Basic Guidelines

● Before your presentation, meet and greet audience

members, establishing rapport with them (your audience
will give you the benefit of the doubt when they see
you’re likable)
● Don’t be afraid to pause the presentation, giving your
audience time to digest the visual or ask questions,
particularly when your message is complicated (being
understood is more important than being quick). “I know
this is complicated, so I’ll pause here and see if anyone
has any questions” signals it’s OK to “not get it”
● Avoid nonwords (e.g., “um,” “ah,” “you know”) and
upspeak (e.g., voice lilt) which do not project confidence

The Storyteller’s Dash: Dashboarding Data

Effectively
There are times analysts cannot be with their audience
when data is presented. Dashboards have become a go-to
solution for analysts who find themselves in this position. In
my opinion, however, dashboards are a necessary evil in the
world of data analytics.
Their ubiquity today makes dashboards a common and
frequently used tool for marketers. They can act as
interfaces to important data for analysts, marketing
decision-makers, partners, clients, and even customers.
They can serve an important purpose in the quest to
democratize data across stakeholder groups. They can
extend the reach of an analyst’s work by representing their
analysis without the analyst being physically present, on
their feet and getting their story across. The dashboards I’ve
seen, however, were poorly designed and sometimes
presented data irresponsibly by removing the analyst’s
voice from the analysis.
Designing dashboards shouldn’t be difficult. To this end,
follow five simple rules to create what I call the
“Storyteller’s Dash.” These rules help analysts realize the
dashboard’s power by conveying clear messages, unlocking
insights, and expressing patterns previously unseen.
Perhaps most importantly, they avoid creating what my
friend Avinash Kaushik aptly terms “Data Pukes” by limiting
the information presented and representing the analyst’s
depth of analysis and nuance.
But before discussing how we can create effective
dashboards, let’s get some basic definitions and details out
of the way.
First, a dashboard is a one-page visual interface that people
can view fully on a screen without scrolling. In other words,
they could print it on a single piece of paper. The interface
should be no larger than the resolution of the smallest
typical display. This means the dashboard interface is no
larger than 1,280 x 1,024 pixels if it’s designed for laptops /
desktops and 480 Å~ 800 pixels if designed for phones.
A dashboard doesn’t have multiple pages or tabs (if the
dashboard is presented in a tool like Microsoft Excel), nor
does it link to other dashboards. If multiple pages exist or if
additional information is required to complement or
supplement the data being presented, an analyst has
created a report – not a dashboard. When so much
information is presented that it requires several steps to
view it all, it’s nearly impossible for an audience to distill a
cogent understanding of what happened and what action
should be taken without the analyst there to guide them
through the story. And remember: You don’t travel with your
dashboard.
Finally, a dashboard is a living window into data and is never
set in stone. The real-time (or near realtime) nature of the
dashboard means that the picture it conveys is in a state of
constant change. Like the business world around them,
dashboards should evolve to reflect that change. This is the
power of the dashboard and what separates it from a static
report. Such freedom allows an analyst to reconfigure the
dashboard frequently by swapping out old metrics with new
ones that better define performance and success. It also
means the analyst has a responsibility to manage the
dashboard’s story proactively.
Now onto the five rules of the Storyteller’s Dash.
Tell a story
A dashboard should never be built as a simple portal to
data. That, as we have learned, is the role of the API
connection. Rather, a dashboard should be built with a
specific purpose in mind and a specific story to tell. That
story could be campaign performance, the overall health of
a business unit, or any other chronicle that’s so dynamic
and important that it demands being constantly monitored
and told. The dashboard’s story answers three questions:
(1) What’s our performance? (the data); (2) What does that
mean? (the insights); and (3) What should we do? (the
recommendations). When these elements are understood,
the analyst can limit what is presented to the data that tells
that story.
Know the audience
It’s critical for analysts to know who will use the dashboards
they create so they can calibrate the amount of data that
displays. If that person is a senior decision-maker, then the
dashboard should have a small set of purposeful metrics
that combine to tell the story you’d want the senior
decision-maker to hear. If the dashboard is designed for a
junior or internal audience, then the dashboard can
generally feature a bit more data, as these audiences
typically require more information in their roles.
Furthermore, an understanding of the audience allows for
the thoughtful management of data access. Those who
need the data can see it, while those who don’t can be
denied access. Most importantly, knowing the audience for
the dashboard leads the analyst to know what’s important
to them, what they need to know, and actions within their
scope. This information is vital for the dashboard to answer
the three questions that make up its story.
Practice “less is more”
An analyst should plan the real estate on the dashboard as
carefully as they would plan the attendee list for a meeting
with the dashboard’s main audience. Only let metrics onto
your dash if they have something important to say. Anything
that doesn’t contribute to the dashboard’s story should be
omitted as extraneous information that would distract or,
worse yet, guide your audience to the wrong conclusion. An
analyst should spend significant time trying to understand
exactly what critical few metrics drive the business. A
general rule of thumb Avinash Kaushik promotes is that a
dashboard should contain fewer than 10 metrics. I agree
wholeheartedly.
Build in context
No data on a dashboard should appear alone. A metric on a
dashboard should ever exist without context because it’s
how an analyst ensures the dashboard conveys insights
rather than creating questions. Context can be a view of the
data or a comparison to a target, a competitor, or previous
performance. This lets an analyst frame the data that makes
up the dashboard’s story. Keep in mind that all the rules of
sophisticated use of contrast, clear meaning, and refined
execution apply to dashboards. The one exception is the use
of headlines and subtitles. With static visuals, as we have
learned, these lines of text ensure the meaning of a visual is
apparent. The limited real estate dashboards provide means
that their use isn’t practical for every visualization of data.
Therefore, analysts must find another way to embed their
voices into dashboards. Hence, the final rule of the
Storyteller’s Dash.
Include the analyst’s voice
Never leave data interpretation to the audience. Remember
that the dashboard effectively sets the data free: You, the
analyst, don’t go with the dashboard. To this end, no
dashboard should exist without including a cogent set of
insights (in written words) that summarize its story by,
again, answering three questions: (1) What is our
performance? (the data); (2) What does that mean? (the
insights); and (3) What should we do? (the
recommendations). Without a record of the answers to these
questions, dashboards are missing the benefit of all the
analysis that went into creating them. A section for text
allows the intelligence from the analyst to bubble up to the
highest level. This may feel awkward as the dashboard is
being designed, as most (if not all) dashboarding tools offer
a plethora of dials and charts but no support for blocks of
text. But given the importance of the story, savvy analysts
will find a way to build text into their dashboards.
Here are some additional best practices of dashboarding to
keep in mind.
Basic Guidelines Pro Tips

● Include a visual indicator ● Find a champion for the

of the overall theme at the dashboard among your
top of the dashboard: a red stakeholders, who can
dot means ensure it gets into the hands
underperformance, a green of the right audiences (and
dot means on target, etc. out of the hands of the
● Put the names of the wrong ones)
analyst or analysts ● Establish a cadence for
responsible for the updates to the text and
dashboard on the dash, metrics presented on the
along with their contact dashboard (typically weekly,
information monthly, or quarterly)
● Apply all the rules of
contrast, clear meaning, and
refined execution in the
previous lessons to each
visual on the dashboard
● Ensure all text on the
dashboard is written in clear,
crisp language and avoids
jargon
A Word on Building Habits over Time
Building the habits discussed here into your practice will
give you the tools to expertly present your work; however,
these habits can be hard-earned. While some of the
presentation guidelines and tips described in this book are
obvious and (relatively) easy to adopt, others might be
foreign to you and force you to undo years of habits. The
process of bending those behaviors to fit the new
approaches introduced here can be tremendously
frustrating. And that’s OK.
Dona Wong sums up the need for patience in our practice:
“We don’t start out writing editorials. We start by learning
the alphabet.” [123] Adopting the presentation guidelines I’ve
discussed in this book will feel awkward at first. If you use
them regularly, they’ll eventually become automatic. Once
you reach that point, you will have truly arrived as a Data
Designer.

In Summary: On Your Feet and Getting

Your Story Across
How analysts present their data stories affects audience
understanding as much as the content does. David
McCandless, an expert data visualization artist, has an
effective approach to data storytelling earned from years of
experience. McCandless’ approach to data storytelling can
be defined as five steps that I call The McCandless Method.
Analysts can learn other presentation style elements by
following several simple guidelines. As with every element
of marketing analytics, presenting data visualizations is a
skill that analysts must build by adopting good habits.
CONCLUSION

D ata’s road from crude maps to gigabytes of

multidimensional information has been a long and
winding one. But it is far from over. If anything, the industry
finds itself at a critical crossroads that will determine its
future for decades to come.
Technology allows us to collect massive amounts of data
quickly and comprehensively, use advanced tools to
discover previously unrecognizable patterns in behavior,
and craft personalized marketing messages like never
before. Recent concerns about data privacy and the
introduction of regulations designed to address these
concerns, however, leaves the industry in a precarious spot.
What comes next for digital marketing analytics is unclear,
but if history is any guide, it will be fascinating.
GLOSSARY

Following are descriptions of words and phrases that pertain

to the concepts in this book.
#
#ferguson – Hashtag used to connect people and filter
news on social media during rioting that followed the
shooting of Michael Brown, an unarmed teenager, by a
police officer in Ferguson, Missouri, a St. Louis suburb.
60-Second Story – A concept Michael Fassnacht
introduced that expresses the true heart of the analyst’s
broader story concisely.
A
A/B Testing – A tactic used to test two versions of an item
(e.g., images, creative messages, etc.) by presenting those
items to consumers and tracking engagement rates to
determine which version is more effective.
Access Tokens – In computer systems, an access token
contains the security credentials for a login session and
identifies the user, the user’s groups, the user’s privileges,
and, in some cases, a particular application. Typically, one
may be asked to enter the access token rather than the
usual password.
Active Evaluation – Part of McKinsey & Co.’s CDJ
framework. Active Evaluation follows the Initial
Consideration Set and is when consumers evaluate what
products to buy, what brands to invest in, and where to
purchase them.
Adobe Analytics – A web analytics enterprise platform
formerly named Omniture. Adobe Systems acquired the
platform in 2009.
Adobe Audience Manager – An Adobe DMP that
integrates online and offline data from sources to deliver a
unified view of audiences.
Adobe Target – A rules-based testing and targeting tool
that can integrate with Adobe Analytics and create reports
for marketing offers, personalization, and UX testing.
Aided Awareness – The percent of consumers who
recognize an advertiser’s brand or product when they see or
hear either name.
Akobeng, Dr. A K – Pediatric gastroenterologist and
researcher who developed a hierarchy of evidence.
American FactFinder – A facilitated download source for
population, housing, economic and geographic information
operated by the U.S. Census Bureau.
Amplification Rate – The volume of “shares” for a piece of
content, or the rate at which those shares are collected (i.e.,
the number of consumers who shared the content/the total
number of consumers who saw the content).
Analogous Color Themes – Harmonic color themes that
use Newton’s logical order of color properties to create eye-
pleasing combinations by selecting colors that are next to
one another on the color wheel.
Analysis Gadgets – Data analysis and/or visualization tools
that offer a single, well-defined capability and feature low
levels of data flexibility.
Analyze Step – The third step in the Marketing Analytics
Process (MAP) during which the analyst analyzes data
they’ve collected to reveal patterns that help explore key
questions they’re investigating.
Annotations – An element of the concept of clear meaning
in good visual form, annotations are small bodies of text
that direct an audience’s attention to important areas of a
chart and provide a written interpretation of the data.
Answer the Obvious Questions – The second step in The
McCandless Method of presenting dataviz that provides
answers to the questions that will jump to audiences
members’ minds before they can ask those questions. By
doing so, analysts can minimize possible confusion and
ensure audiences pay attention to analysts’ stories rather
than try to decode chart elements.
Application Programming Interface (API) – A computer-
to-computer interface specific to an application or operating
system that allows third parties to extend functionality. Also,
a means by which analysts can collect data.
Applause Rate – The percent of users who see the content
that has the desired positive reaction. Such positive
reactions can come from a Like on Facebook, a favoriting of
a Tweet on Twitter, or saving an image to a Pinterest
account.
API – Application programming interface is a set of clearly
defined methods of communication between various
components. It allows a connection between two machines.
ART+SCIENCE – The blend of the art of data visualization
and the science of data analysis that produces the
successful approach to analytics.
ART+SCIENCE Mind – The mind of the data analyst who
successfully balances the three functional skills: Data
Strategist, Techie, and Data Designer.
AT&T – American telecommunications company that
purchased the first digital banner ad and placed it on
hotwire.com.
Audience Growth – Growth in the number of people
connected to the brand on the social media platform (as
“followers”).
B
Banner Ad – A form of internet advertising that embeds an
ad into a webpage.
Beck, Henry – English technical draftsman who created the
iconic London Underground Tube map in 1931.
Bing – American web search engine that Microsoft owns and
operates.
Boldness – The confidence and courage to act innovatively
and in a fastidious manner to produce work of the highest
quality. One of three characteristics of the successful
analyst’s mindset.
Booth, Charles – English shipowner, social researcher and
reformer, best known for his innovative philanthropic
studies on working-class life in London toward the end of the
19th century.
Bounce Rate – A website analytics metric that measures
the percentage of visitors who enter a website and leave
without viewing other pages.
Boxplot – A graphical method of depicting a group through
quartiles.
Brand Advocacy – Observable and measurable public
support for a brand, typically expressed through social
media platforms.
Brand Asset Valuator – Young & Rubicam developed
approach for valuing brand impact using attribute
assessments that are based on differentiation, relevance,
esteem, and knowledge.
Brand Equity – An approach to valuing brand impact that
combines effective marketing share, relative price, and
durability.
Brand Valuation – An approach to valuing brand impact
through models based on available financial data and
assumptions.
Brandwatch (formerly Crimson Hexagon) – A digital
consumer intelligence company headquartered in Brighton,
England. Brandwatch sells five products: Consumer
Research, Audiences, Vizia, Qriously, and BuzzSumo.
BrandZ Top 100 Most Valuable Brands Report – A
report highlighting the 100 brands with the highest brand
impact values based on an advertiser’s financial data,
market dynamics, and an assessment of the brand’s role in
income generation.
Brinker, Scott – Vice president Platform Ecosystem at
HubSpot and editor at chiefmartec.com who has authored
the marketing technology landscape supergraphic, the
“Martech 5000” since 2011.
Build Awareness – One of six primary Marketing
Objectives. By building awareness, the advertiser hopes to
increase recognition and recall of a product.
Business Objectives – Broad C-suite goals that a company
wants to achieve, stated in its own language, and are easily
measurable. These objectives are the result of the company
successfully achieving its Marketing Objectives.
BuzzSumo – A Brandwatch-owned content analysis tool
that provides insights into the types of content that
resonate with specific audience groups.
C
Campaign Objectives – Key performance indicators (KPIs)
through which the success of media campaigns is
measured.
Cart Abandonment Rate – The rate at which consumers
don’t complete the checkout process after placing an item
in an online shopping cart.
Causal Analysis – One of the five primary categories of
marketing data analysis. An approach that seeks to
determine how the movement of one variable registers on
other variables in the collected dataset or model.
Change – One of the primary categories of patterns found
in data. The trend or instance of observations becoming
different over time.
Chart Legend – Also known as a “key,” the chart legend
defines the elements of a chart and is placed on the page
near the chart. In this author’s opinion, using chart legends
are an unnecessary and distracting practice.
Clear Meaning – One of three critical elements of good
visual form. Clear meaning expresses the proper use of
common elements of a chart (e.g., headlines, subtitles,
annotations, and labels) to communicate insights
effortlessly.
Click-Through Rate – The percent of consumers who saw
an ad (search, display) and clicked on it.
Clickstream Analysis – Analyzing the way consumers
navigate websites through the use of tools such as Google
Analytics and Adobe Analytics.
Clip Art – Pre-made images used to illustrate concepts or
ideas and, in this author’s opinion, the hallmark of lazy and
subpar visualization.
Close – The fifth and final step in The McCandless Method of
presenting dataviz that transitions to the next point in a
presentation, continuing the momentum to the next topic of
discussion.
Clustering – One of the primary categories of patterns
found in data. The task of grouping a set of objects in such a
way that objects are in the same group.
Clutter – Untidy data/noise in the data that’s insignificant.
CMYK Color Model – Also known as process color or four
color, CMYK is a subtractive color model used in color
printing and is also used to describe the printing process.
CMYK refers to the four inks used in color printing: cyan,
magenta, yellow, and key (black).
Colleague Test – A test designed to improve a chart’s
quality by showing it to a colleague who has no association
to the analysis and determining if the chart is effective at
communicating its meaning. Passing the colleague test is
the bar analysts should set for the quality of their charts.
Collect Step – The second step in the Marketing Analytics
Process (MAP), during which the analyst collects data for
their analysis through the use of facilitated downloads,
application programming interfaces (APIs), and web
scraping techniques.
Color Contrast – Differences in color (i.e., hue) in a data
visualization that communicate distinct elements in the
chart instantly.
Color Saturation (or Value) – The intensity of color in an
image. In technical terms, saturation and value are the
bandwidth of light from a source. By increasing (darkening)
and decreasing (lightening) the saturation or value of a hue,
analysts can create a variety of color shades that work well
together.
Color Wheel – Sir Isaac Newton’s logically arranged
sequence of primary, secondary, and tertiary colors
wrapped onto a circle.
Completed Video Views (or Rate) – The number of times
a consumer watches a video in its entirety (or the percent of
consumers who watch a video in its entirety).
Complementary Color Themes – Harmonic color themes
that use Newton’s insights into color properties to create a
palette of colors that work well together by choosing colors
that are directly across from one another on the color wheel
to produce harmonious pairs of hues.
Consumer Response Modeling – Also known as
“response modeling,” this analysis technique improves
customer response rates by targeting those prospects most
likely to react to a particular treatment, campaign,
advertisement, media or promotion.
Control Group – The group in an experiment or study that
doesn’t receive treatment from researchers and is used as a
benchmark to measure how the other tested subjects
(treatment group) behave.
Consumer Decision Journey Framework – A journey
model designed by McKinsey & Co. that helps analysts
understand how consumers research and buy products.
Consumer Outcomes – The set of measurable
conversions, macroconversions and microconversions that
brands attempt to influence consumers to complete.
Consumer Sentiment – Also known as “sentiment,” the
tone of consumer mentions of a company, brand, or product
typically categorized as “positive,” “neutral,” and
“negative.”
Content Analysis – An approach to analysis that reveals
patterns in data that are otherwise unrecognizable.
Contrived Contrast – The use of boxes, callouts,
annotations, and other preattentive attributes to distinguish
items in a visual. These are purposeful, planned
introductions of contrast that draw the audience’s attention
and do not arise from inherent contrast found in size, color,
and shape contrast.
Conversion Rate – The percent of consumers who, after
clicking on an ad, take an action on the linked page. Tracked
actions could be making a purchase on the page,
downloading a file, or any number of actions the brand
deems to be positive.
Correlation Analysis – An analysis technique used to
calculate the statistical relationship between two variables.
Cost Per Action (CPA) – The cost invested in a search
campaign divided by the number of some tracked consumer
action. Those actions could be clicks (cost per click, or CPC),
or any other conversion actions the brand deems important.
Counterfactual – The expected behavior of the treatment
group if the intervention during the test period didn’t occur.
It’s calculated by applying the factor determined during the
pre-test period to the control group to project treatment
group behavior. During the test period, the difference
between the counterfactual and the treatment group’s
actual (observed) behavior is the incremental effect of the
intervention being tested.
Correlation – The process of seeing dependence of one
variable over another variable.
Curiosity – A strong desire to learn. One of three
characteristics of the successful analyst’s mindset.
D
D3 – A JavaScript library for producing dynamic, interactive
data visualizations in web browsers. It uses Scalable Vector
Graphics, HTML5, and Cascading Style Sheets standards.
Data-Driven – The kind of approach that uses data to
communicate the message.
Data Designer – One of three roles the analyst must play.
Someone who understands how to best express data stories
and the ability to use tools to visualize those stories.
Data Flexibility – An assessment of how much an analyst
can do with a specific tool. One of two measures (along with
ease of use) used to evaluate the fast-moving data analytics
and visualization tool market.
Data Management Platform (DMP) – A tool that collects
and organizes data, allowing brands to target specific
consumer audiences on ad networks and measure campaign
performance across segments and channels.
Data Strategist – One of three roles the analyst must play.
Someone who can bridge the data and marketing worlds.
Dataviz – Short form for data visualization and represents
all the colorful designs that are created to tell a story.
Declarative – Visual form of affirmative data
communication.
Deterministic Model – A model that features no unknowns
or randomness and thus will always produce the same
output from a given starting condition or initial state.
Descriptive Analysis – One of the five primary categories
of marketing data analysis. An approach that calculates
summary statistics that quantitatively describe or
summarize features from a dataset.
Diamond Insights – A concept that communicates the
need for analysts to expect insights to come from the
compression and distillation of large amounts of data
through robust analysis. It discourages analysts from
approaching insight development as a trivial process that
will reveal insights by sifting through a dataset, akin to a
pioneer searching for treasure (i.e., insights are better
thought of as diamonds, not gold).
Digital Cookies – Text files that sit in a browser’s cache
and allow a webpage to identify you.
Digital Fingerprinting – An approach to tracking
consumers by analyzing sets of information that can include
network protocols, operating systems, hardware devices,
and software among other things, thereby avoiding security
features and blockers.
Digital Marketing Maturity – The measurable degree to
which an organization invests in technology, tools, and
analysts to improve the effectiveness of online capabilities
for itself and its clients.
Direct Labels – A superior alternative to chart legends,
direct labels are placed directly on or next to elements in a
chart.
Door Study – An experiment introduced by American
researchers Daniel Simons and Daniel Levin to demonstrate
a concept they call “change blindness.” This is the
psychological phenomenon that occurs when an observer
overlooks a change in a visual stimulus.
E
Ease of Use – A personal assessment of how easy it is for
an analyst to use a specific tool. One of two measures
(along with data flexibility) used to evaluate the fast-moving
data analytics and visualization tool market.
Effective Reach (X+ Reach) – The number of people who
will see an ad the most effective number of times. The most
effective frequency (e.g., “1+ Reach” means everyone who
saw the ad at least once).
Engagement Rate – The percent of consumers who
engage in some behavior (e.g., sharing a video, clicking a
link in the video, etc.) Enterprise Platforms – Big,
powerful solutions that feature high data flexibility due to
the broad and deep sets of capability.
Enrich – To enhance and improve the quality of data.
Epoch – Periods marked by notable events or particular
characteristics.
Evergage – A cloud-based software that allows users to
collect, analyze, and respond to user behavior on their
websites and web applications in real time.
Evidence → Insight – The incorrect sequencing of a chart’s
takeaway and the data that substantiates that insight when
analysts explain dataviz. In this incorrect approach, analysts
present facts from the visual before revealing the insight
(see “Insight → Evidence” for the correct sequence).
Excel – A spreadsheet program that’s part of the Microsoft
Office suite.
Explanatory – An approach to data analysis where visuals
created from a dataset are used to search for and explore
its patterns.
Exploratory Analysis – One of the five primary categories
of marketing data analysis. An approach to analyzing
datasets to summarize their main characteristics, often with
visual methods.
F
Facebook – American online social media and social
networking service based in Menlo Park, California. Mark
Zuckerberg founded the service, which launched on
September 26, 2006.
Facilitated Downloads – A means of accessing data
through a structured process a data owner offers and
manages, typically (but not always) involving a graphical
user interface (GUI) that guides the data collector through
the required steps for access.
FAMGA – An acronym that represents five of the world’s
largest tech companies: Facebook, Apple, Microsoft, Google,
and Amazon.
Fassnacht, Michael – Former CEO and president of
advertising agency FCB Chicago, current CMO for the City of
Chicago, and an important thought leader in the
development of modern data analytics and marketing data
strategy.
Favorability – Also known as “preference,” the percent of
consumers who favor a product over a competitor’s product
or a set of competitive products.
First Moment of Truth (FMOT) – Part of the three-step
model of marketing that Proctor & Gamble developed to
describe the process of brand marketing. The time when
consumers find themselves at the shelf, are ready to
purchase, and have a number of options for the products
they want.
First-Party (1P) Data – Data an organization collects
through direct relationships with consumers.
Five W's and How, The – The idea that a complete report
must provide answers to questions starting with the
interrogative words, who? what? when? where? why? and
how? The principle is commonly applied in journalism, law
enforcement investigations, and research, and its origin has
been tied to Greek philosopher Aristotle.
Flooz.com – A now-defunct dot-com venture based in New
York City that went online in February 1999, and attempted
to establish a currency unique to internet merchants, similar
in concept to airline frequent-flier programs. The name
“flooz” was based upon the Arabic word for money.
Font Choice – The font or collection of fonts an analyst
chooses for their chart.
ForeSee Results – An online survey tool offered as part of
the ForeSee CX Suite that Verint acquired in 2019.
Fowler, Geoffrey A. – American journalist and The
Washington Post’s technology columnist, writing from San
Francisco.
Frequency – The number of times an ad was exposed to an
average person or household.
Full Episode Players (FEP) – Professionally produced, TV-
like content that can appear on any device type, across
apps and web browsers. The content is typically 30-60
minutes, with commercial breaks.
G
Gates, Bill – American business magnate, software
developer, investor, and philanthropist best known as the
co-founder of Microsoft Corporation.
General Data Protection Regulation (GDPR) – A
regulation in European Union (EU) law on data protection
and privacy in the EU and the European Economic Area.
Give Away the Insight – The third step in The McCandless
Method of presenting dataviz that reveals a chart’s insight.
Summarize this insight concisely in the chart’s subtitle.
Google – American multinational technology company that
specializes in internet-related services and products. It was
founded in September 1998.
Google Analytics – A web analytics service from Google
that tracks and reports website traffic, currently as a
platform inside the Google Marketing Platform brand.
Google launched the service in November 2005 after
acquiring Urchin.
Google Consumer Surveys – A Google-owned online
survey and market research tool.
Google Dataset Search – A Google search engine that
helps researchers locate online data that’s freely available
for use. It launched publicly on January 23, 2020.
Google Optimize – A free, Google-owned website
optimization tool that continually tests different
combinations of website content.
Google Sheets – A spreadsheet program included as part
of a free, web-based software office suite that Google offers.
Google Slides – A presentation program included as part of
a free, web-based software office suite that Google offers.
Google Trends – A Google website that analyzes the
popularity of top search queries in Google Search across
various regions and languages. The website uses graphs to
compare the search volume of different queries over time.
Gpairs – R programming language package used for
plotting data.
Graphic – Visual form of art that also involves charts (in this
context).
Graphical User Interface (GUI) – User interface that
allows users to interact with electronic devices through
graphical icons and visual indicators, such as secondary
notation, instead of text-based user interfaces.
Gross Rating Point (GRP) – Total of all rating points
during an advertising campaign.
Groupon – American global e-commerce marketplace that
connects subscribers with local merchants by offering
activities, travel, goods and services. Based in Chicago, the
company launched in November 2008.
H
Headline – A brief description that headlines what the data
shows.
Hex Code – A series of coded numbers and letters that
describes the composition of a certain color in a specific
color space, usually red-green-blue (RGB).
Histograms – A graphical representation of numerical data
based on the frequency.
Hollaback! – A photo blog and grassroots initiative to raise
awareness about street harassment and combat it by
posting photographs and narrative accounts of individuals’
encounters.
Hootsuite – A social media management platform created
in 2008 and headquartered in Vancouver. Hootsuite’s
interface is a dashboard, and supports social network
integrations for Twitter, Facebook, Instagram, LinkedIn,
Google+, and YouTube.
Hotwire.com – One of the earliest travel websites on the
internet, launching in 2000.
I
Ice Bucket Challenge – An activity involving the dumping
of a bucket of ice water over a person's head, either by
another person or self-administered. Video of the activity
was typically posted to social media websites to promote
awareness of the disease, amyotrophic lateral sclerosis
(ALS), and encouraged donations for research.
iCharts – A data visualization tool from iCharts, a company
headquartered in Sunnyvale, California, founded on
September 9, 2008.
Illustrator – A vector graphics editor Adobe developed and
markets for macOS.
Impression Share – The number of impressions a search
ad received divided by the number of impressions the ad
was eligible to receive.
Improving the Sales Process – One of six primary
Marketing Objectives. By improving the sales process, an
advertiser hopes to reduce the loss of potential consumers
by better facilitating the moment of purchase.
In-Demo Gross Rating Point (aka TRP) – Percentage of
the total audience that fits the intended in-demo target
profile (e.g., women aged 36-54).
Increasing Loyalty – One of six primary Marketing
Objectives. By increasing loyalty, an advertiser hopes to
increase the percent of consumers who repurchase the
advertiser’s product (i.e., repeat purchase rate) when a
need arises that the product can fulfill.
Incrementality Tests – On-demand experiments that
measure the incremental effect of a specific campaign or
tactic.
Influence Consideration – One of six primary Marketing
Objectives. By influencing consideration, an advertiser
hopes to increase the likelihood that a consumer will choose
the company’s product instead of a competitor’s product or
a substitute behavior (e.g., deciding to not buy anything).
Inferential Analysis – One of the five primary categories
of marketing data analysis. An approach that uses a sample
of data to infer something about a larger population.
Insight → Evidence – The correct sequencing of a chart’s
insight (first) and the data that substantiates that insight
(second) when analysts present dataviz (see “Evidence →
Insight” for the incorrect sequence).
Intellectually Blank – A statement included in a dataviz
subtitle that lacks insight into the chart’s underlying
message.
Internet Movie Database (IMDb) – An online database of
information related to films, television programs, home
videos, video games, and streaming content online. It
includes cast, production crew and personal biographies;
plot summaries; trivia; fan and critical reviews; and ratings.
The database launched in 1990.
Initial Consideration Set (ICS) – Part of McKinsey & Co.’s
CDJ framework. The set includes the relevant brands that
pop into consumers’ heads when they’re considering
products.
Intelligent Tracking Prevention (ITP) – A feature Apple
introduced in 2017 that reduces cross-site tracking in Safari
browsers by limiting cookies and other website data.
Introduce the Graphic by Name – The first step in The
McCandless Method of presenting dataviz wherein the
analyst presents the graphic’s name. The name should be
clear, unadorned, and obvious and be captured concisely in
the chart headline.
iPerceptions – An online survey and customer experience
management tool founded in 1999.
K
K-Means Clustering – A method of vector quantization
that’s popular for clustering data into like segments during
analysis.
Kaggle – A Google subsidiary that’s an online community of
data scientists and machine learning practitioners and is a
facilitated download source for data.
Kaushik, Avinash – Digital Marketing Evangelist at Google.
He’s a digital marketing analytics thought leader and the
author of Web Analytics: An Hour A Day and Web Analytics
2.0 .
Kenny, John – Head of planning at FCB Chicago. He
developed a number of effective techniques collected in the
“Numerical Comparisons Tool,” bringing greater context to
numbers using principles of behavioral economics.
Key Performance Indicator (KPI) – A type of
performance measurement. KPIs evaluate an organization’s
success or the success of a particular activity in which it
engages.
Key Questions – Specific questions or hypotheses that an
analyst investigates during an analysis. Answering these
questions helps the analyst guide a company toward its
Marketing Objective.
Keynote – A presentation software application developed
as part of Apple’s iWork productivity suite.
Klout – A now-defunct website and mobile app that used
social media analytics to rate its users according to online
social influence via the “Klout Score,” which was a
numerical value between 1 and 100.
Kozmo.com – A now-defunct dot-com venture that
promised free one-hour delivery of “videos, games, DVDs,
music, mags, books, food, basics & more” and Starbucks
coffee in several major cities in the United States.
Investment bankers Joseph Park and Yong Kang founded the
venture in March 1998 in New York City.
L
Label – A tag or a marker to any point on the chart.
Lee, Aileen – Cowboy Ventures founder who coined the
term “unicorn” to describe pre-IPO tech startups with a
market valuation of $1 billion or more.
Legibility – Clarity of what the data/charts say. The
graphical representation of the data should be clear enough
to read and understand.
Levin, Daniel – American researcher who, along with
partner Daniel Simons, introduced the Door Study to
demonstrate a concept they call “change blindness.” This is
the psychological phenomenon that occurs when an
observer doesn’t notice a change in a visual stimulus.
Lift Measurement – An approach to analysis that
measures how a campaign affects an identifiable and
measurable metric.
Linguistic Inquiry and Word Count (LIWC) – A
commercial content analysis tool that analyzes text.
LinkedIn – American business and employment-oriented
service that operates via websites and mobile apps. It
launched on May 5, 2003, and became a wholly owned
Microsoft subsidiary in December 2016.
Location Data – Information that a mobile device, like a
smartphone or tablet, provides about its current physical
position.
Logic Tree – A visual technique that divides possible
options into branches and makes a decision process visually
easy to interpret.
Look-Alike Analysis – A type of data analysis that applies
a model of a consumer’s attributes to a larger population
with the goal of identifying consumers who demonstrate
similar attributes.
Loyalty Loop – Basically a shortcut from a trigger to the
Moment of Purchase. When traveling along the Loyalty Loop,
a consumer experiences a trigger, and rather than going
through an Initial Consideration Set and Active Evaluation,
invests in a trusted brand immediately.
LUMAscape – A supergraphic Luma Partners created to
organize the famously complicated world of advertising
technology (“ad tech”) by grouping similar companies on
one page.
M
Macroconversion – The ultimate goal for a brand of a
consumer’s journey – most frequently a sale of goods or
services – and the culmination of a series of
microconversions.
Marbles, Jenna – Jenna Nicole Mourey, better known by
her pseudonym Jenna Marbles, is an American YouTube
personality, video blogger, comedian, and actress.
Mark I – The IBM Automatic Sequence Controlled Calculator
– called Mark I by Harvard University’s staff – was a room-
sized general-purpose computer built in 1944 and widely
regarded as the world’s first fully functional computer.
Market Share – The percentage of a market accounted for
by a specific product, brand, or company.
Marketing Effectiveness – The effectiveness of a
marketer’s go-to-market strategy for maximizing positive
results in the short- and long-term. It’s also related to return
on investment measures, such as return on advertising
spend (ROAS) and return on marketing investment (ROMI).
Marketing Mix Models (MMM) – A statistical analysis,
such as multivariate regressions on sales and marketing
time series data, that estimates the effect of various
marketing tactics on sales and forecasts the effect of future
tactics.
Marketing Objectives – Department goals that support
the Business Objective, are stated in a company’s language
and are measurable. Marketing Objectives are the result of
the company successfully achieving its media objectives.
Marketing Technology (Martech) – The term for the
software and tech tools marketers use to plan, execute, and
measure marketing campaigns.
Martech 5000 – The marketing technology landscape
supergraphic that Scott Brinker has created annually since
2011.
Match Rate – The percent of consumer data from a file
(e.g., a customer list) that’s matched to consumer data in
another file (e.g., cookie ID data).
McCandless Method – A five-step process for presenting
data stories to audiences that Kevin Hartman developed and
named after David McCandless. Hartman’s process is based
on the successful way McCandless has presented data
visualization.
McLean, Ross – Vice President - Mobile Qualitative at 20|20
Research and co-author of the “You + Big Idea” graphic,
along with Jamie Shuttleworth and Karl Turnbull.
Marketing Analytics Process (MAP) – The four-step,
recursive process that guides an analyst through the
journey of an analytics pursuit. It consists of the following
steps: plan, collect, analyze, and report.
Measurement Multiplicity – The concept of using a
combination of tools applied concurrently or in a planned
cadence to bring clarity to the brand’s performance and
market environment.
MECE – An abbreviation for Mutually Exclusive Collectively
Exhaustive. In this context, it means that facts the analyst
uses in support of an insight are mutually exclusive from
one another (i.e., no overlap) and yet, taken together,
represent the full breadth of the insight.
Media Objectives – Broad goals affected by various media
executions and tactics employed to present consumers with
the brand’s messages. Media objectives are reached when
campaigns are successful.
Microconversion – An important, measurable step a
consumer completes on the way to a macroconversion, such
as viewing a webpage, downloading a coupon, etc.
Minard, Charles Joseph – French civil engineer recognized
for his significant contribution in the field of information
graphics in civil engineering and statistics.
Minitab – A statistics package that Pennsylvania State
University researchers Barbara F. Ryan, Thomas A. Ryan, Jr.,
and Brian L. Joiner developed in 1972.
Moment of Purchase – Part of McKinsey & Co.’s CDJ
framework. This is the time when a consumer makes a
decision and buys a product.
Monochromatic Color Themes – A harmonic color theme
that uses one hue to minimize color distraction on the page.
Through variations in saturation and value, analysts can
expand the selected hue to fulfill their color needs for any
presentation or visual.
Montulli, Lou – A programmer who is well-known for his
work in producing web browsers. Montulli is credited with
inventing digital cookies in 1994.
Moz – A Seattle-based software as a service company that
sells inbound marketing and marketing analytics software
subscriptions.
Multitouch Attribution Model (MTA) – A method of
marketing measurement that evaluates the effect of each
touchpoint in driving a conversion through a sophisticated
model, thereby estimating the value of that specific
touchpoint.
MySQL – A free, open-source relational database
management system. Its name is a combination of “My,”
the name of co-founder Michael Widenius’ daughter, and
“SQL,” the abbreviation for Structured Query Language.
N
Napster – A pioneering peer-to-peer (P2P) file-sharing
internet software that emphasized sharing digital audio files
– typically audio songs – encoded in MP3 format. The
software was founded in June 1999, and Best Buy acquired it
on December 1, 2011.
NASDAQ – American stock exchange founded in 1971 that
was the first stock market in the United States to trade
online, using the slogan “the stock market for the next
hundred years.” The Nasdaq Stock Market attracted many
companies during the dot-com bubble.
Net Promoter Score (NPS) – An approach to valuing
brand impact through a scoring system that gauges the
loyalty of a firm’s customer relationships.
Nightingale, Florence – English social reformer,
statistician, and the founder of modern nursing. Nightingale
came to prominence while serving as a manager and trainer
of nurses during the Crimean War, when she organized care
for wounded soldiers.
Nonwords – Filler words, such as “um,” “ah,” and “you
know” that speakers use during presentations. They can
distract audiences and reduce the speaker’s
professionalism.
Numerical Comparisons Tool – A collection of a number
of effective techniques that bring greater context to
numbers using principles of behavioral economics. John
Kenny of FCB Chicago developed the techniques that
comprise the tool.
O
Online Video (OLV) – The general field that deals with the
transmission of video over the internet.
Optimism – A hopefulness about the successful outcome of
an analysis. One of three characteristics of the successful
analyst’s mindset.
Optimizely – An American company that provides A/B
testing tools, in which two versions of a webpage can be
compared for performance and multivariate testing.
Oscars Selfie – A “selfie” photo snapped at the 2014
Academy Awards show. Ellen DeGeneres organized the
selfie, which features several prominent film actors. When
the “Oscars Selfie” was tweeted as the show was still in
progress, it set a record with 3.4 million retweets.
Oracle DMP (formerly BlueKai) – A cloud-based DMP that
helps marketing organizations personalize online, offline,
and mobile marketing campaigns with information about
targeted audiences.
Over-the-Top (OTT) – A streaming media service offered
directly to viewers via the internet. OTT bypasses cable,
broadcast, and satellite television platforms – the companies
that traditionally act as a controller or distributor of such
content.
P
ParseHub – A free web scraping tool.
Peek Test – A test designed to improve the use of contrast
on a chart that involves the analyst taking their eyes off the
chart for a period and then looking at it to see where their
eyes are drawn. If their eyes are drawn to the intended
location of the chart, the chart passes the peek test.
Personally Identifiable Information (PII) – Any data that
could potentially be used to identify a particular person.
Examples include a full name, Social Security number,
driver’s license number, bank account number, passport
number, and email address.
Peterson, Eric – CEO at Web Analytics Demystified and the
author who introduced the “Pyramid Model of Web Analytics
Data.”
Pets.com – A now-defunct dot-com enterprise
headquartered in San Francisco that sold pet supplies to
retail customers. It began operations in November 1998 and
liquidated in November 2000.
PewDiePie – Felix Arvid Ulf Kjellberg, better known as
PewDiePie, is a Swedish YouTuber, known primarily for his
Let’s Play videos and comedic formatted shows.
Photoshop – A raster graphics editor Adobe developed and
published.
Photoviz – Same as dataviz but uses photos of real-world
objects rather than computer-generated visuals to represent
data.
Pie Chart – A circular statistical graphic which is divided
into slices to illustrate numerical proportion.
Pinterest – American social media web and mobile
application company. It operates a software system that
enables people to discover and save information on the
World Wide Web using images and, on a smaller scale, GIFs
and videos.
Plan Step – The first step in the Marketing Analytics
Process (MAP), during which time an analyst identifies a
singular Marketing Objective that will direct their analysis.
The analyst plans their approach with a planning document.
Planning Document – A document an analyst uses to
record their planned approach to an analysis project. The
document can take any form but must detail the Marketing
Objective that will guide their analysis, the connection
between the Marketing Objective and key questions the
analysis will investigate, the data to be collected, and the
sources of those data.
Playfair, William – Scottish engineer and political
economist who introduced several innovations in the
presentation of quantitative information by means of graphs
and charts.
Point Solutions – Data analytics and/or visualization tools
that typically center on a primary capability and feature
medium data flexibility.
Postpurchase Experience – Part of McKinsey & Co.’s CDJ
framework. It occurs when a consumer takes a product
home and uses it.
PowerPoint – A presentation program developed as part of
the Microsoft Office suite.
Preattentive Attributes – These are attributes that
determine what information catches our attention.
Predictive Analysis – One of the primary categories of
marketing data analysis. Closely linked to causal analysis,
predictive analysis uses models to predict the future value
of data attributes.
Prefrontal Cortex (PFC) – The cerebral cortex that covers
the front part of the frontal lobe. This brain region has been
implicated in planning complex cognitive behavior,
personality expression, decision-making, and moderating
social behavior.
Pre-Test Period – A period before an analyst conducts a
study, during which the analyst observes behaviors of the
treatment and control groups. By analyzing the pre-test
period, an analyst can determine a factor that equates the
behavior of the two groups to each other.
Primary Colors – The colors that cannot be formed through
a combination of any other colors, specifically: red, blue,
and yellow.
Probabilistic Model – A model that incorporates random
variables and probability distributions to estimate an event
or phenomenon. While a deterministic model gives a single
possible outcome for an event, a probabilistic model gives a
probability distribution as a solution.
Programmatic Media Buying – The data-intensive
algorithmic purchase and sale of advertising space in real
time through the use of marketing technology (i.e.,
“martech”).
Provide Examples – The fourth step in The McCandless
Method of presenting dataviz that offers substantiation for
the insight revealed in the third step. Typically, these
examples are highlighted through annotations or pulled
from patterns in the data.
Purchase Intent – The percent of consumers who intend to
purchase a product, rather than a competitor’s product or a
product from a set of competitive products, the next time a
need arises.
Python – An open-source general-purpose programming
language Guido van Rossum created and released in 1991.
Pyramid Model of Web Analytics Data – Web Analytics
Demystified CEO Eric Peterson developed this concept that
depicts the inverse relationship between digital data
availability and digital data value.
Q
Quanteda – An R package designed to conduct quantitative
analysis on text.
Questionnaire Bias – Bias in data collection that results
from unanticipated communication barriers between the
investigator and respondents, yielding inaccurate results.
Bias may arise from the way individual questions are
designed, the way the questionnaire as a whole is designed,
and how the questionnaire is administered or completed.
R
R – An open-source programming language the R
Foundation for Statistical Computing supports and that’s
used to analyze data and build dataviz.
Ranking – One of the primary categories of patterns found
in data. The position in a scale of achievement or status.
Regression Analysis – An analysis technique that involves
a set of statistical processes for estimating the relationships
between a dependent variable and one or more
independent variables.
Refined Execution – One of three critical elements of good
visual form. Refined execution is the deep attention to detail
and seeks to improve the legibility of a chart.
Relativity – One of the primary categories of patterns
found in data. The consideration of data observations in
relation or in proportion to something else.
RStudio – An integrated, visual-based development
environment for R.
Randomized Controlled Trial (RCT) – A study design that
randomly assigns participants into an experimental group or
a control group. As the study is conducted, the only
expected difference between the control and experimental
groups in a randomized controlled trial (RCT) is the outcome
variable being studied.
Rating Point – One percent of the potential audience.
Reach – Measurement of the size of the audience that
watches an ad. Reach (%) = Gross rating points (%) /
Frequency.
Refined Execution – A critical element of good visual form,
refined execution expresses the attention to detail needed
to ensure charts are as legible as possible.
Repeat Purchase Rate – The percent of consumers who
buy a product they purchased previously.
Report Step – The fourth and final step in the Marketing
Analytics Process (MAP), during which an analyst presents
the story they found during their analysis in a visual form.
Repositioning the Brand – One of six primary Marketing
Objectives. By repositioning the brand, an advertiser hopes
to better align its product with the needs of its target
consumers.
Return On Investment (ROI) – A ratio between net profit
and cost of investment. A high ROI means the investment’s
gains compare favorably to its cost.
RGB Color Model – An additive color model in which red,
green, and blue are combined to create a broad array of
colors. The name comes from the initials of the three
additive primary colors: red, green, and blue.
Royalty Relief – An approach to brand impact valuation
that calculates the net present value of the hypothetical
royalty payments an organization would receive if it were to
license its brand to a third party.
RStudio – A free and open-source integrated development
environment (IDE) for R, a programming language for
statistical analysis.
S
Saleh, Khalid – Co-founder and CEO of Invesp, an e-
commerce optimization company.
Sampling Bias – Also known as “selection bias.” A bias in
which a sample is collected in a way that some members of
the intended population have a lower sampling probability
than others.
Scheiner, Christoph – German priest, physicist, and
astronomer who played an instrumental role in the early
development of data collection methods.
Scripting Language – A programming language for a
special run-time environment that automates the execution
of tasks, such as R or Python.
Search Engine Optimization (SEO) – The process of
increasing the quality and quantity of website traffic by
increasing the visibility of a website or a webpage to users
of a web search engine.
Search Engine Results Page (SERP) – The links to pages
search engines display in response to a user’s query. The
SERP typically responds with ads (if relevant) and organic
results.
Search Engine Management (SEM) – A form of internet
marketing that involves the promotion of websites by
increasing their visibility in search engine results pages,
primarily through paid advertising.
Search Rank – The position that the brand’s website or
page is returned on the SERP.
Searchvolume.io – A free search trends analysis tool.
Second Moment of Truth (SMOT) – Part of Proctor &
Gamble’s three-step marketing model that describes the
process of brand marketing. It begins after the consumer
makes a purchase decision and brings that product home.
Here, the product is evaluated against the expectations set
during the consumer’s evaluation process.
Second-Party (2P) Data – Data an organization owns and
provides to a company.
Secondary Colors – The colors that are formed by mixing
primary colors (e.g., purple is a secondary color formed by
mixing red and blue).
Sensitivity Analysis – The study of how different variables or
values affect a model’s output.
Shape Contrast – Differences in shapes presented in a
data visualization that communicate the uniqueness of each
element instantly.
Share of Audience (SOA) – One way to quantify
relevance. Calculate SOA by dividing the number of
subscribers to a brand’s OLV channel by total subscribers for
brand channels in the category.
Share of Search (SOS) – One way to quantify relevance.
Calculate SOS by dividing the number of searches for a
brand by total searches conducted for brands in the
category.
Share of Voice (SOV) – One way to quantify relevance.
Calculate SOV by dividing the number of social mentions for
a brand by total social mentions for brands in the category.
You can also use SOV to calculate the share of advertising
messages a brand produces by swapping social mentions in
the calculation above with ads (or ad play time).
Shirky, Clay – American writer, consultant and teacher who
focused on the social and economic effects of internet
technologies and journalism.
Shuttleworth, Jamie – U.S. Chief Strategy Officer at
mcgarrybowen and co-author of the “You + Big Idea”
graphic, along with Ross McLean and Karl Turnbull.
Sign-Ins – The process of using usernames and passwords
to manage access to digital content.
Simons, Daniel – American researcher who, along with
partner Daniel Levin, introduced the Door Study to
demonstrate a concept they call “change blindness.” This is
the psychological phenomenon that occurs when an
observer doesn’t notice a change in a visual stimulus.
Singer, Natasha – A journalist on the investigative team
for The New York Times . She specializes in technology
coverage and has written extensively about location
tracking with partner Jennifer Valentino-DeVries.
Size Contrast – Differences in size presented in a data
visualization that communicate the relative difference of
elements in the chart instantly.
Smosh – American sketch comedy YouTube channel that
Anthony Padilla and Ian Hecox created.
Snapchat – A multimedia messaging app used globally. It
was released on July 8, 2011.
Snow, John – English physician considered one of the
founders of modern epidemiology, in part, because of his
work in tracing the source of a cholera outbreak in Soho,
London, in 1854.
Social Media – Interactive computer-mediated
technologies that facilitate the creation or sharing of
information, ideas, career interests, and other forms of
expression via virtual communities and networks.
SocialMention – A free, real-time social tracking tool used
to collect publicly available information about a brand
across social media posts, website blogs, news, videos,
articles, and other pieces of content. It’s particularly
valuable when conducting SOV analysis.
Sophisticated Use of Contrast – One of three critical
elements of good visual form. The sophisticated use of
contrast seeks to draw the audience’s attention to
important elements of a chart while sending less important
and complementary data to the background.
Spartan Test – A test that improves a chart’s legibility by
examining each element of the graphic to ensure it makes a
positive contribution to the visual by methodically deleting
unnecessary elements.
Split Complementary Color Themes – Harmonic color
themes created by using colors on either side of
complementary colors on the color wheel. This is among the
techniques that help to create harmonious color palettes.
SPSS – A software package for interactive or batched
statistical analysis. SPSS Inc. produced it until IBM acquired
the company in 2009. The current versions are IBM SPSS
Statistics.
Stata – A general-purpose statistical software package
StataCorp created in 1985. Most users work in research,
particularly in economics, sociology, political science,
biomedicine, and epidemiology.
Stimulate Demand – One of six primary Marketing
Objectives. By stimulating demand, an advertiser hopes to
increase demand for a product using tactics, such as
clarifying the need it fulfills, or introducing scarcity.
Store Visits – A measure based on tracking location data
that records when a consumer enters a defined physical
location.
Structured Data – Data that’s neatly organized and
conforms to a clear and consistent data format.
Structured Query Language (SQL) – A programming
language designed for managing data held in relational
databases. IBM designed it in the 1970s. Today, it’s the most
popular language an analyst can use to interact with a
relational database.
Subtitle – A brief line of text on a chart placed below the
headline that provides the insight that the audience should
take away from the chart.
SurveyMonkey – An online survey development cloud-
based software as a service company founded in 1999.
Sysomos – A Toronto-based social media analytics
company that Meltwater owns.
T
Tableau – American interactive data visualization software
company founded in January 2003 in Mountain View,
California. Salesforce acquired the company on August 1,
2019.
Tags – Strings of code that initialize when someone
performs an action in a browser, such as loading a webpage
or clicking an object displayed on that page.
Talk-Track – The narrative of the story you’re trying to tell
through the visuals.
Techie – One of three roles an analyst must play. The data
owner must understand how to collect and manage data in
ways that ensure data quality and promote efficiency.
Term Relevance – The relative number of times a term is
used in a body of text data, often visualized through a word
cloud.
Tertiary Colors – Colors formed by mixing a primary color
and a secondary color (e.g., blue-green, red-violet, yellow-
orange, etc.).
Test-and-Learn Attitude – An attitude analysts adopt that
accepts risk but mitigates its effect by allowing for programs
to “fail fast.” Through this attitude, analysts seek to improve
a brand’s marketing efforts by constantly optimizing
targeting, messages, and other important elements.
Test Period – The period when an analyst performs a study.
Third-Party (3P) Cookie – A digital cookie set by a domain
that’s not the domain of the webpage the consumer visits.
3P cookies are typically used for targeting personalized ads
to consumers and measuring conversions.
Third-Party (3P) Data – Data firms collect when they don’t
have direct relationships or agreements with the consumers
who generate the data.
Thompson, Stuart A. – A journalist who specializes in
technology coverage and is on the investigative team at The
New York Times . He’s written extensively on the subject of
location tracking with partner Charlie Warzel.
Tidy Data – A standard way of mapping the meaning of a
dataset to its structure. A dataset is messy or tidy,
depending on how rows, columns, and tables are matched
up with observations, variables, and types. In tidy data,
each variable forms a column and each observation forms a
row.
Time-Series Plot – A type of line graph that shows data
patterns over time.
Three-Step Model of Marketing – A Proctor & Gamble
framework that describes the process of brand marketing. It
involves a Stimulus, the First Moment of Truth (when the
consumer is at shelf, in-store making a purchasing decision),
and the Second Moment of Truth (following purchase when
the consumer experiences the product).
Trade Desk, The – A global technology company that
markets a software platform, including a DMP. Digital ad
buyers use the platform to purchase data-driven digital
advertising campaigns across various ad formats and
devices.
Treatment Group – The set of participants in a research
study that are exposed to some manipulation or intentional
change in the independent variable of interest.
Triadic Color Themes – Harmonic color themes that are
created by picking three colors from the color wheel that are
perfectly spaced from one another.
Trigger – Part of McKinsey & Co.’s CDJ framework. The
beginning of every consumer’s purchase path. Some sort of
stimulus that initiates a product need.
Turnbull, Karl – Founder and Chief Strategy Officer of
Cavalry and co-author of the “You + Big Idea” graphic, along
with Ross McLean and Jamie Shuttleworth.
TweetReach – A free tool for analyzing Twitter hashtags,
users accounts, and other activity.
Twitter – American microblogging and social networking
service based in San Francisco, California. It was created in
March 2006.
U
Unaided Awareness – The percent of consumers who
voluntarily mention the brand or product when prompted by
a generic question such as, “What brands of automobile
have you heard of?” Unaided awareness clearly reveals a
deeper consumer/brand connection than aided awareness.
Unicorn – A pre-IPO tech startup with a market valuation of
$1 billion or more. Also, a legendary creature that’s been
described as a beautiful horse with a large, pointed,
spiraling horn projecting from its forehead. Unicorns have
delighted children since antiquity.
Unique Page Views – A website analytics metric that
measures the number of unique pages a visitor views on a
website.
Unstructured Data – Data that doesn’t come in a
predefined, standardized format. It’s usually text-heavy and
sometimes contains dates and numbers that don’t fit a
uniform description, as well as awkward (but increasingly
important) data, such as images, sounds, and video.
V
Valentino-DeVries, Jennifer – A journalist who specializes
in technology coverage and is a member of the investigative
team for The New York Times. She’s written extensively on
the subject of location tracking with partner Natasha Singer.
Varian, Hal – An economist specializing in microeconomics
and information economics. He is the chief economist at
Google and holds the title of emeritus professor at the
University of California, Berkeley where he was founding
dean of the School of Information.
Venn Diagram – A visual chart that shows all possible
logical relations between a finite collection of different sets.
Voice of the Consumer – Data collected (typically through
online surveys) that measures consumer sentiment,
opinions, and/or thoughts about any number of topics,
including products, brands, and companies. Also known as
“voice of the customer.”
Visits – A website analytics metric that measures the
number of times people visit a website.
Visual Cues – An element of the concept of clear meaning
in good visual form, visual cues are design elements, such
as arrows, boxes, and shaded areas, that can be effective
additions to visuals by highlighting messages.
Visual Perception – The way our brain interprets data
visually.
Visual Polish – The expert application of small but
important details to a chart that improves its legibility.
Visual Website Optimizer (VWO) – A website
optimization tool that allows analysts to create A/B tests and
geo-behavioral targeting campaigns without having
technical or HTML knowledge.
W
Warzel, Charlie – A journalist who specializes in
technology coverage and is a member of the investigative
team for The New York Times . He’s written extensively
about location tracking with partner Stuart A. Thompson.
Web Scraping – Also known as web harvesting or web data
extraction. An analyst uses this technique to extract data
from websites that’s not provided through facilitated
downloads or APIs. Data is copied, or “scraped,” directly
from the website’s HTML code and typically requires the use
of a web scraping tool to make the collection process
efficient.
Website Analytics – The collection, reporting, and analysis
of website data. The focus is on identifying measures based
on organizational and user goals and using the website data
to determine the success or failure of those goals.
Wickham, Hadley – Statistician from New Zealand, Chief
Scientist at RStudio, and an adjunct statistics professor at
the University of Auckland, Stanford University, and Rice
University.
Wong, Dona – Senior Vice President, Digital Strategy,
Communications, at the Federal Reserve Bank of New York
and author of The Wall Street Journal Guide to Information
Graphics .
Word Cloud – A dataviz that highlights most frequently
used words in a document or paragraph.
Wordle – An online tool to create word clouds.
X
X+ Reach (Effective Reach) – The number of people who
will see an ad the most effective number of times. The most
effective frequency (e.g., “1+ Reach” means everyone who
saw the ad at least once).
Y
Yahoo – American web services provider headquartered in
Sunnyvale, California. Verizon Media owns the company.
YouTube – American online video-sharing platform created
in February 2005 and headquartered in San Bruno,
California. Google purchased the platform in November
2006.
Z
Zero Moment of Truth (ZMOT) – Google introduced the
concept to represent the time between the Stimulus and the
First Moment of Truth in P&G’s three-step marketing model.
During ZMOT, consumers use online and offline information
to inform their purchase decisions.
Zuckerberg, Mark – American internet entrepreneur and
philanthropist who co-founded Facebook, Inc. and serves as
its chairman, CEO, and controlling shareholder.
ABOUT THE AUTHOR

A spartner
Director of Analytics for Google, Kevin and his team
with major advertisers, creative agencies, and
media companies to develop digital solutions that build
businesses and brands. His approach mixes science and art
to deliver inventive, fact-based strategies that reduce
uncertainty and increase effectiveness in the marketing and
advertising programs they create.
In addition to his work at Google, Kevin has taught
graduate-level analytics courses for nearly 10 years at the
University of Chicago, the University of Notre Dame, and the
University of Illinois.
Kevin earned his BA in political science and economics from
the University of Notre Dame and his MBA and MPP from the
University of Chicago.
[1]
Regulating the internet giants : The world’s most valuable resource is no
longer oil, but data , The Economist (May 6, 2017)
[2]
Burtch Works, The Burtch Works Study, Salaries of Data Scientists &
Predictive Analytics Professionals (August 2020)
[3]
Burtch Works, The Burtch Works Study, Salaries of Data Scientists &
Predictive Analytics Professionals (August 2020)
[4]
Michael Friendly and Daniel Denis, Milestones in the History of Thematic
Cartography, Statistical Graphics, and Data Visualization (2001)
[5]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[6]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[7]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[8]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[9]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[10]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[11]
Friendly, Denis, Milestones in the History of Thematic Cartography,
Statistical Graphics, and Data Visualization (2001)
[12]
Adrienne LaFrance, The First-Ever Banner Ad On The Web (April 21,
2017)
[13]
Doubleclick’s Display Benchmarking Tool as cited by SmartInsights, US,
Europe and Worldwide Display Ad Clickthrough Rates Statistics Summary
(September 10, 2019)
[14]
Adam Hayes, Dotcom Bubble (June 25, 2019)
[15]
Leslie Berline, “Lessons of Survival, From the Dot-Com Attic,” The New
York Times (November 21, 2008)
[16]
John C. Wu, “Anatomy Of A Dot-Com,” Supply Chain Management Review
(November/December 2001)
[17]
Wu, “Anatomy Of A Dot-Com,” Supply Chain Management Review
(November/December 2001)
[18]
Erin Griffith and Dan Primack, The Age of Unicorns (January 22, 2015)
[19]
Griffith, Primack, The Age of Unicorns (January 22, 2015)
[20]
Max Roser, Hannah Ritchie, and Esteban Ortiz-Ospina, Internet (2019)
[21]
Caspar Llewellyn Smith, “Ellen DeGeneres' Oscars selfie beats Obama
retweet record on Twitter,” The Guardian (March 2, 2014)
[22]
Rhonda Richford, MIPTV: Ellen DeGeneres' Oscar Selfie Worth as Much as
$1 Billion (April 8, 2014) quoting Publicis CEO Maurice Levy
[23]
ALS Association, The ALS Association Expresses Sincere Gratitude to
Over Three Million Donors (August 29, 2014)
[24]
Kimberlee Morrison, “The Top 5 Viral Events of 2014,” Adweek
(December 24, 2014)
[25]
Gene Demby, “Combing Through 41 Million Tweets To Show How
#BlackLivesMatter Exploded,” NPR (March 2, 2016)
[26]
Hermione Hoby, “The woman in 10 Hours Walking in NYC: ‘I got people
wanting to slit my throat’,” The Guardian (December 17, 2014)
[27]
Ron Marshall, How Many Ads Do You See In One Day? (September 10,
2015)
[28]
Luca Bruna/Associated Press (2005), Michael Sohn/Associated Press
(2013)
[29]
Scott Brinker, Marketing Technology Landscape Infographic (August 3,
2011)
[30]
Brinker, Marketing Technology Landscape Supergraphic (2019): Martech
5000 (actually 7,040) (April 4, 2019)
[31]
Brinker, Marketing Technology Landscape Supergraphic (2019): Martech
5000 (actually 7,040) (April 4, 2019)
[32]
Domo (2019), BrandWatch (2019), Social Pilot (2019), Tinder (2019),
Word Development (2019), U.S. Election Atlas (2019), UK Election Atlas (2019)
[33]
Domo (2019)
[34]
Miniwatts Marketing Group, Internet World Stats (November 19, 2019)
[35]
Arthur Samuel, “Some Studies in Machine Learning Using the Game of
Checkers,” IBM Journal of Research and Development (1959)
[36]
Prannoiy Chandran, Disruption in Retail — AI, Machine Learning & Big
Data (July 23, 2018)
[37]
Demand Metric, Seismic, Content Marketing’s Evolution: The Age Of Hyper-
Personalization And Automation (March 2016)
[38]
Infogroup Personalization Study 2019, n=1,504 (June 2019)
[39]
Todd Spangler, YouTuber PewDiePie is More Liked Than LeBron James
Among Gen Z Males, Study Finds (November 5, 2019)
[40]
AudienceProject (January 17, 2019)
[41]
AudienceProject (January 17, 2019)
[42]
Christopher Elliott, Yes, There Are Too Many Ads Online. Yes, You Can Stop
Them. Here’s How (February 9, 2017)
[43]
Google, 2019 research review: New media channels are emerging
(December 2019)
[44]
Google, 2019 research review: New media channels are emerging
(December 2019)
[45]
Google/Ipsos, Global (U.S., CA, BR, U.K., DE, FR, JP, IN, KR, AU), “How
People Shop with YouTube” Study, 18–64-year-olds who go online at least
monthly and have purchased something in the last year, n=24,017 (July 2018)
[46]
Google/Magid Advisors, Global (U.S., CA, BR, U.K., DE, FR, JP, IN, KR, AU),
“The Role of Digital Video in People’s Lives,” n=20,000, A18–64 general online
population (August 2018)
[47]
Google/Ipsos, Global (U.S., CA, BR, U.K., DE, FR, JP, IN, KR, AU), “How
People Shop with YouTube” Study, 18–64-year-olds who go online at least
monthly and have purchased something in the last year, n=24,017 (July 2018)
[48]
Mark Dolliver, “US Time Spent with Media 2020: Gains in Consumer
Usage During the Year of COVID-19 and Beyond,” eMarketer (April 29, 2020)
[49]
Dolliver, “US Time Spent with Media 2020: Gains in Consumer Usage
During the Year of COVID-19 and Beyond,” eMarketer (April 29, 2020)
[50]
Dolliver, “US Time Spent with Media 2020: Gains in Consumer Usage
During the Year of COVID-19 and Beyond,” eMarketer (April 29, 2020)
[51]
Variety, Leading ad supported broadcast and cable networks in the
United States in 2019, by average number of viewers (in millions) (December
 26, 2019)
[52]
Netflix, Number of Netflix paying streaming subscribers in the United
States from 3rd quarter 2011 to 2nd quarter 2020 (in millions) (July 16, 2020)
[53]
comScore, Most popular online video properties in the United States as of
May 2020, ranked by unique viewers (in millions) (July 21, 2020)
[54]
We Are Social, Hootsuite, and DataReportal, Most popular social networks
worldwide as of July 2020, ranked by number of active users (in millions) (July
23, 2020)
[55]
eMarketer (April 2020); N.B.: Includes live, DVR, and other prerecorded
video (such as video downloaded from the internet but saved locally); includes
all time spent watching TV, regardless of multitasking; Gen Z are individuals
born between 1997 and 2012; millennials are individuals born between 1981
and 1996; Gen X are individuals born between 1965 and 1980; baby boomers
are individuals born between 1946 and 1964
[56]
Anthony Crupi, “Sleepy, Hollow: What You Won’t Hear At This Year’s TV
Upfronts,” AdAge (March 15, 2018)
[57]
eMarketer, Digital video penetration in the United States as of March
2019, by age group (February 18, 2020)
[58]
StatCounter Global Stats, Worldwide data for search engine referrals
(December 2019)
[59]
StatCounter Global Stats, Worldwide data for search engine referrals
(December 2019)
[60]
Lesley Fair, FTC Staff to Search Engines: Differentiate Ads From Natural
Results (June 25, 2013)
[61]
PPC, Hal Varian, Pay Per Click Management - Insights on the Google
AdWords Auction System (October 24, 2014)
[62]
Tris Heaword, Eye Tracking in 2017 for Google Hotel Searches: Why The
Old Rules Don’t Apply (March 17, 2017)
[63]
Google, 2019 research review: New media channels are emerging
(December 2019)
[64]
Google, 2019 research review: New media channels are emerging
(December 2019)
[65]
IAB Internet Advertising Revenue Report, 2018 Full Year Results,
prepared by PwC (May 2019)
[66]
Khalid Saleh, “US Digital Display Ad Spending – Statistics and Trends,”
Invesp (2019)
[67]
Geoffrey A. Fowler, “Goodbye, Chrome: Google’s Web browser has
become spy software,” Washington Post (June 21, 2019)
[68]
Yapmedia, Here’s Why You Need Social Media (2019)
[69]
Yapmedia, Here’s Why You Need Social Media (2019)
[70]
StatCounter, Social Media Stats Worldwide (January 2020)
[71]
Yapmedia, Here’s Why You Need Social Media (2019)
[72]
Luth Research, Auto Correct: The Marketing Of Car-buying Is Changing
(July 7, 2017)
[73]
N.B.: If you are, as I expect, a dutiful analyst reading this book, you’ll
want to understand how you can collect and analyze data to answer the
questions in the CDJ. Fear not! While we’ll simply identify those questions in
this lesson, Part 4 of this book will explore how we pursue and find those
answers in greater depth.
[74]
Fred Reichheld, “Prescription for Cutting Costs,” Bain & Company
(September 2001)
[75]
Cox Automotive, 2019 Car Buyer Journey Study (June 4, 2019)
[76]
Adblock Plus (2020)
[77]
J. Clement, “Facebook: advertising revenue worldwide 2009-2019,”
Statista (Feb 28, 2020)
[78]
N.B.: On mobile devices, such as phones and tablets, digital cookies can
be used on mobile web browsers just like they’re used on desktop and laptop
computers. Cookie technology, however, doesn’t work in apps, which comprise
the majority of mobile device internet usage. Mobile cookie data is, therefore,
viewed as less comprehensive and less reliable than cookie data from laptops
and desktops. As a result, a unique “device ID” is created to allow websites
and servers to identify your mobile device.
[79]
Eric T. Peterson, Web Analytics Demystified: A Marketer’s Guide to
Understanding How Your Web Site Affects Your Business (2004)
[80]
Jennifer Valentino-DeVries and Natasha Singer, “Your Apps Know Where
You Were Last Night, and They’re Not Keeping It Secret,” The New York Times
(December 10, 2018)
[81]
Valentino-DeVries and Singer, “How to Stop Apps From Tracking Your
Location,” The New York Times (December 10, 2018)
[82]
Stuart A. Thompson and Charlie Warzel, “Twelve Million Phones, One
Dataset, Zero Privacy,” The New York Times (December 19, 2019)
[83]
David Yanofsky, “Half of American Airlines’ revenue came from 13% of its
customers,” Quartz (October 27, 2015)
[84]
Kevin Hartman, “How to bring your marketing mix modeling into the 21st
century,” Think With Google (June 2019)
[85]
Nielsen MMM Meta-Analysis, U.S., n=20 studies from CPG clients, 2016–
2018
[86]
Hartman, “How to bring your marketing mix modeling into the 21st
century,” Think With Google (June 2019)
[87]
Hartman, “How to bring your marketing mix modeling into the 21st
century,” Think With Google (June 2019)
[88]
Hartman, “How to bring your marketing mix modeling into the 21st
century,” Think With Google (June 2019)
[89]
Alex Esber, et. al., “Measure Marketing Effectiveness: A Guide to
Implementing Incrementality,” FacebookIQ (September 11, 2018)
[90]
Esber, et. al., “Measure Marketing Effectiveness: A Guide to
Implementing Incrementality,” FacebookIQ (September 11, 2018)
[91]
A K Akobeng, “Understanding randomised controlled trials,” Archives of
Disease in Childhood (2005)
[92]
N.B.: In some cases, social listening and social media network analytics
tools will provide content analysis – term relevance, sentiment analysis, and
other modules – which can complement results earned from standalone
content analysis tools or be used as a substitute.
[93]
“How machine learning is liberating creativity,” Think With Google
(October 9, 2019)
[94]
N.B.: Many companies that offer enterprise platform products have
premium (i.e., paid) and free versions of their tools. For example, Google offers
a free version of Google Analytics as well as a premium version – Google
Analytics 360 (GA360). Typically, free versions of these tools feature much less
data flexibility than their premium counterparts. Free website analytics tools
are classified as point solutions rather than enterprise platforms.
[95]
N.B.: The concept of voice of the consumer is also referred to as “voice of
the customer,” although consumer in this context is used to broaden the
surveyed population from people who buy the products (i.e., consumers who
are customers) to those who don’t (i.e., consumers who aren’t customers).
[96]
N.B.: This assessment includes a sample of free tools available to
analysts and isn’t intended to be a complete review of the data analysis and
visualization tool market.
[97]
Avinash Kaushik, “Best Web Analytics 2.0 Tools: Quantitative, Qualitative,
Life Saving!,” Occam's Razor (October 19, 2010)
[98]
Dominic Field, Growing Up Digital, Part One: The Four Stages of
Marketing Maturity , (Nov 30, 2017)
[99]
Bain & Company and Google, Measurement Maturity Survey (January
2019); n=622 marketing executives; n=188 leading marketers who grew
market share and exceeded top 2017 business goal, n=115 lagging marketers
who lagged behind on these two measurements.
[100]
Gerald C. Kane, Doug Palmer, Anh Nguyen Phillips, David Kiron, and
Natasha Buckley, Coming of Age Digitally Learning, Leadership, and Legacy
(Summer 2018)
[101]
Field, Growing Up Digital, Part One: The Four Stages of Marketing
Maturity (November 30, 2017)
[102]
Bain & Company and Google, Measurement Maturity Survey (January
2019); n=622 marketing executives; n=188 leading marketers who grew
market share and exceeded top 2017 business goal, n=115 lagging marketers
who lagged behind on these two measurements.
[103]
Kane, Palmer, Phillips, Kiron, and Buckley, Coming of Age Digitally
Learning, Leadership, and Legacy (Summer 2018)
[104]
Jamie Shuttleworth, Ross McLean, and Karl Turnbull, Getting To What
Matters (2010)
[105]
Shuttleworth, McLean, Turnbull, Getting To What Matters (2010)
[106]
Dwight D. Eisenhower, Remarks at the National Defense Executive
Reserve Conference (November 14, 1957)
[107]
Andrew Clark, “Knob Creek Runs Dry,” The Guardian (Jul 20, 2009)
[108]
Michael C. Sloan, “Aristotle's Nicomachean Ethics as the Original Locus
for the Septem Circumstantiae,” Classical Philology (2010).
[109]
Samuel Taylor Coleridge, The Rime of the Ancient Mariner (1834)
[110]
Christie Schneider, The biggest data challenges that you might not even
know you have (May 25, 2016)
[111]
Kaggle.com website (2020)
[112]
Hadley Wickham, Journal of Statistical Software (February 20, 2013)
[113]
Michael Fassnacht, “The veil of statistics,” MarketingGeek (June 7, 2006)
[114]
Will Thalheimer, “How Much Do People Forget?,” Work-Learning Research,
Inc. (December 2010)
[115]
Daniel J. Simons and Daniel T. Levin, “Change Blindness,” Trends in
Cognitive Sciences (October 1997)
[116]
“New smart contact lens for diabetics introduced,” Science Daily
(February 22, 2018)
[117]
David McCandless in an email to participants in his Workshops Are
Beautiful seminar (November 29, 2016)
[118]
David McCandless, What Makes A Good Visualization (2009)
[119]
David McCandless, InformationIsBeautiful.net (July 2010), Clay Shirky,
Cognitive Surplus (2010)
[120]
Guillaume Cieutat, Number of Mainland China Stores in 2020 (March
2020)
[121]
Mark Fairchild, Color Appearance Models: CIECAM02 and Beyond
(November 9, 2004)
[122]
Color Blind Awareness, Color Blindness (2020)
[123]
Dona Wong, The Wall Street Journal Guide to Information Graphics
(January 4, 2010)