MKTG 631

Marketing Analytics
Module1 – Review of Marketing Process and
Statistics

Jinhee Huh
Review of Marketing Process
What is Marketing Analytics?

• Optimize marketing outcomes using data from a variety of sources

• First party data
• A company’s own internal data

• Second party data

• Another company’s first party data shared directly from the source

• Third party data

• Other firms
• Scraped data
Why Marketing Analytics Matters Now More than
Ever
Consumer
Social network
demographic Mobile data
data

In-store Online shopping

purchase data data
Marketing
Data Web traffic data
Loyalty program
data
Online survey
Chat log
data
Survey data
 Marketing Process and Marketing Analytics
Customer
Margin
Regression
Company
Time series ROI Text analysis
analysis Financials Objectives Competitor Regression
CLV Choice model
Collaborator
Context

Product
Price Tactics Strategy
Regression
Choice model Segmentation Regression
Place ANOVA
Targeting Cluster analysis
Promotion Classification
Positioning
AirBnB’s Marketing Process
Marketing Process: Describe Your Business

• Describe your business and clearly identify goals and objectives

• Mission statement
• What is the purpose of your business?
• What is the strategic influence for your business?
• What is the desired public perception for your business?
• How does your mission statement clarify your strategy?
• How does your mission statement unify your team?

• Example: Airbnb
• Airbnb’s mission is to create a world where anyone can belong anywhere and we are focused on
creating an end-to-end travel platform that will handle every part of the trip
 Marketing Process: Situational Analysis
• Details the context for your marketing efforts by considering internal and external factors
that could influence marketing strategy

5Cs of marketing Customer Company Competitor

Collaborator Context
Marketing Process: Situational Analysis
• SWOT analysis
Helpful Harmful

Strength Weakness
Internal

• Assets and resources to • Resources and capabilities

improve business’ that may cause business to
competitive position have a less competitive
position

Opportunities Threats
External

• Situations or conditions • Problems that focus on your

arising from business’ weaknesses
strengths • Problems that can create a
• Set of positive externalities potentially negative situation
Marketing Process: Marketing Strategies & Tactics

• Marketing mix (The Four Ps)

WHAT
Product What are the expected benefits for customers to solve issues related to:
Variety, colors and sizes, quality, packaging and features, etc.

HOW MUCH
How to make the solution accessible in relation to:
Price List price, discounts, terms and means of payment, cost vs. profit compared to
the market, etc.
WHERE
Definition of places and accesses for:
Place Physical and online business reach area, location of stores, distribution and
transport of goods, differentiation of local competitors, etc.
How
Marketing strategies for:
Promotion Visual identity of product and categories, production online content, press
office, customer attraction and targeting, etc.
Marketing Process: Marketing Strategies & Tactics

• The Four Ps of Airbnb

Online marketplace for users to list, find, and rent vacation homes
Product ‘Entire place’, ‘Private room’, and ‘Shared room’

Serving different segments of travelers

Price Split-fee structure: the fee is split between hosts and guests
Host-only structure: the entire service fee is collected from hosts

Over 7 million listings

Place Available in more than 220 countries and 100,000 cities
Online marketplace

TV advertisement and social media marketing (ex. Instagram and YouTube)

Communicating the value proposition of “living like a local”
Promotion Various public relation activities and sponsorships
Logo and website design
Marketing Process: Marketing Strategies & Tactics

• STP

S T P
Segmentation Targeting Positioning

Select most Determine how to

Divide market into
attractive segments position your
smaller,
to focus on your product for each
homogeneous
marketing on target segment
groups
Marketing Process: Marketing Strategies

Location Activity Business Price Family

Entire house, kitchen

Marketing Process: Implementation & Evaluation

• Implementation: What are we going to do?

• Organization
• Activities and responsibilities
• Timetables

• Evaluation: How do we evaluate?

• Key performance indicators (KPIs)
• Metrics used to indicate whether marketing tactics to meet your objectives are performing well
and meeting your goals
Marketing Process: Implementation & Evaluation
Analytical Model can be Used at Every Step of
Marketing Process
• Principal Component Analysis
• Multidimensional Scaling
Marketing segmentation and • Cluster analysis
targeting • Classification analysis

Market research and hypothesis • Group comparison (One- or two-way ANOVA)

testing
• Linear regression
Consumer choice • Consumer choice model (Logit or discrete choice model)

• Association rule mining

Purchase decision process • Behavioral sequence (Markov Chains)

Social Media analysis • Text analysis

• Ridge/Lasso regression
Machine learning • Classification
• Clustering
Review of Statistics
Descriptive Statistics
• Measures of central tendency

• Number of tomato cans sold: 1, 2, 2, 3, 4, 7, 9

• Mean
• Sum of values of a variable divided by the number of observations:

• Median
• Middle value that has just as many observations above it and below it
• 1, 2, 2, 3, 4, 7, 9
• Mode
• Most frequent value of a variable
• 1, 2, 2, 3, 4, 7, 9
Descriptive Statistics
• Measures of variability
• Range

• Variance

• Standard deviation
• Square root of variance
Probability Distribution
• An arrangement of values of a variable showing their observed or
theoretical frequency of occurrence

Histogram
Probability Distribution
• Approximation of a histogram

Probability
Normal Distribution
• The most important continuous probability distribution!
• Probability density function

𝜇 − 𝜎 𝜇𝜇+ 𝜎

Sales ($)
 Central Limit Theorem
• If you take sufficiently large random samples from a population with
mean and standard deviation with replacement, then the
distribution of the sample means will approximately follow normal
distribution

Source: StackQuest
 Central Limit Theorem
• If you take sufficiently large random samples from a population with
mean and standard deviation with replacement, then the
distribution of the sample means will approximately follow normal
distribution

Source: StackQuest
 Central Limit Theorem
• If you take sufficiently large random samples from a population with
mean and standard deviation with replacement, then the
distribution of the sample means will approximately follow normal
distribution

Source: StackQuest
 Central Limit Theorem
• If you take sufficiently large random samples from a population with
mean and standard deviation with replacement, then the
distribution of the sample means will approximately follow normal
distribution

Source: StackQuest
Central Limit Theorem
• Implication

• CLT tells us that no matter what the distribution of the population is, the shape
of the sampling distribution will approach normality as the sample size
increases.

• The normality assumption is vital for (parametric) hypothesis tests

• Confidence interval
• t-test
• ANOVA test
What is Hypothesis Testing?
• Testing the results of a survey or experiment to see if you have
meaningful results
A B
 What is Hypothesis Testing?
• What is hypothesis and how can we test it?

The 8
likelihood
to check 6
the 4
website 2

A B
Consumers who got A promotion email are
more likely to check the website than the
consumers who got B promotion email ………

Reject or fail to reject

What is Hypothesis Testing?
• Null hypothesis
• The hypothesis that there is no difference between groups or variables.
• ex) There is no difference in the likelihood to visit between group A and group B.
•
• Alternative hypothesis
• Inverse or opposite of the null hypothesis
• ex) There is a difference in the likelihood to visit between group A and group B.

• Hypothesis testing and statistical model testing

• Statistical model diagnostics and interpretation
• Hypothesis testing is used to test statistical significance of estimates or overall model fit
• ex) Linear regression model: The coefficient is equal to zero
What is Statistical Model?
• Simplified, mathematically-formalized way to approximate reality and
optionally to make predictions from this approximation
• ex) linear regression, Poisson regression, logit regression, ANOVA, cluster
analysis, decision trees, time-series, classification, …

Coefficients of
t the estimated
Number of model
tomato cans sold 𝑑𝑖𝑠𝑐𝑜𝑢𝑛𝑡=$ 5

Prediction of
t
the real data
Discount

Model the number of tomato cans

with the discount amount
 What is Statistical Model?
• The model is an approximation of the real data

Number of
tomato cans sold

Discount
• A lot of statistics determine if a model makes a good or bad approximation of the
data.
• Model diagnostic
• Model fit
• Also, the statistics about the coefficient help with the model interpretation.
Statistics and P-value

• What is p-value?

• In the hypothesis testing, p-value is the probability of obtaining test

results at least as extreme as the results are actually observed, under
the assumption that the null hypothesis is true.

• Two-sided or one-sided p-value

 Statistics and P-value
• What is p-value?
0.5 0.5 × 0.5=0.25 The probability that the coin is a special coin =

0.5 0.5 × 0.5=0.25

0.5

0.5 The probability that the coin is a normal coin =

0.5 × 0.5=0.25
0.5

0.5
0.5 × 0.5=0.25

H1: The coin is a special coin

Cannot reject the null hypothesis
H0: The coin is a normal coin
Source: StackQuest
Statistical Testing Using P-value
• When do you need p-value? How does p-value affect model
interpretation?
t
𝐻 0: 𝑇h𝑒 𝑑𝑖𝑠𝑐𝑜𝑢𝑛𝑡 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑒𝑞𝑢𝑎𝑙𝑠𝑡𝑜0.
Number of
tomato cans
sold
𝑝− 𝑣𝑎𝑙𝑢𝑒=0.025
Discount
Multivariate Descriptive Statistics

• Covariance and correlation

• Measures of the relationship between two random variables
• Not measures of causality!
• Covariance
= 66 8 -7.5 -5 37.5
72 11 -1.5 -2 3
• Directionality of the relationship 77 15 3.5 2 7
84 20 10.5 7 73.5
• Correlation 83 21 9.5 8 76
= 71 11 -2.5 -2 5

• Range: [-1, 1] 65 8 -8.5 -5 42.5

70 10 -3.5 -3 10.5
255