8 rules for better

data storytelling
It was the end of 2020 when social media platforms faced an influx of Spotify Wrapped content—a summary of the user’s most-streamed music in the past
year. Those given the title of “top 1% fans worldwide” of their favorite artists proudly paraded their title like a badge of honor, while others shared their Most
Played Genre as a declaration of their musical preferences. In 2020 alone, 90 million users engaged with Spotify Wrapped. Clearly, users enjoyed the data
stories their streaming statistics told about them.

Spotify Wrapped is only one example that illustrates the increasing use of data visualizations in our daily lives. As the world watched the Covid-19 pandemic
unfold, data visualizations like John Hopkin’s Dashboard played a crucial role in informing the public on the extent of the pandemic. Data stories like National
Geographic’s coverage on 500,000 Covid-19 deaths elicited awareness of the deadliness of the virus. All these examples were in line with DataCamp’s
prediction of data visualization (and data stories) becoming mainstream in our 2021 Trends Report.

The difference between data visualizations and data stories is nuanced but important. Put simply, data visualizations are good at articulating what has
happened, but not necessarily why something has happened. This is where data storytelling comes into the picture.

Data storytelling is a methodology for communicating information with a compelling narrative. It is made up of three components—the data, visuals, and
narratives. Together, these components help data storytellers engage the audience, make data more memorable, and be more persuasive.

Why we need data stories

Data stories make the data more memorable

You might find it much easier to recall a funny story than the tenth digit of pi. 

Stories serve as a framework to connect disparate pieces of information coherently and elicit an emotional response.

Humankind has used storytelling to pass down culture and traditions. In a paper Memory, Imagination and Learning: Connected by the Story, it was
discovered that storytelling is the most effective technique for oral cultures to faithfully memorize their knowledge. In a sense, humans are hardcoded to
recall stories better than cold hard facts.

2
Data stories help us communicate better stories facilitate communication
between the storyteller and the listener. In fact, neuroscientist Uri Hasson Table of contents
observed that the brain activities of the listeners of a story are similar to the
brain activities of storytellers. This phenomenon, called the “speaker-listener
neural coupling.” indicates the achievement of successful communication. Know your audience
4

This demonstrates the potential of stories in eliciting mutual understanding.

Data stories drive actions

Begin with the goal in mind

6

When data are presented intuitively in the form of stories, they can be Choose the best visualization for your story
7

understood and acted upon quickly. Data stories also serve as a medium to
communicate data insights and inspire collaborative action in a way that a Do not lie with data stories 8
regular visualization or dashboard on its own could not.

Keep your visualizations minimal and avoid clutter

10

Who should learn data storytelling?

Add color to your stories

11

Data storytelling is a valuable skill set for both technical and non-technical
professionals. As companies build up self-serve capabilities in analytics, more Use texts appropriately
12

non-technical professionals will get access to a treasure trove of data.

However, the potential of that untapped data remains locked unless Develop a narrative around your data
13

employees are capable of telling compelling data stories to communicate

insights and drive action.

On the other hand, data practitioners who craft audience-specific data stories
can effectively convey the impact of their data science projects in the
language of the audience. This technique can be applied in the scoping,
implementation, and evaluation stages of the projects to alleviate the
skepticism of non-technical users. This paves the way to getting the stamp of
approval from decision makers and end users.

With the growing importance of data storytelling in the upcoming decade, all
employees must be equipped with the fundamentals of data storytelling. 

n this white paper, we outline eight rules for better data storytelling that will
help anyone craft impactful data-driven narratives.

3
 Know your audience

The Economist, a print international newspaper with an emphasis on data

journalism and analysis, recently ventured into Instagram. In charting this new

territory, it constantly redesigns charts for its audience instead of simply reusing

graphs meant for the magazine. The Economist’s feat in garnering millions of

Instagram followers within a few years attests to its ability to cater to its audience.

Here, we distill a few learning points we gathered from the Economist.

Know what is the priority of the audience

Data storytellers should know what matters to the audience, and present data

stories that are of importance to the audience. Otherwise, one runs the risk of

losing interest or wasting the time of the audience.

Particularly, a non-technical audience cares much more about the business impact

of a data science project rather than a 1% incremental improvement in accuracy.

As such, data practitioners should be mindful that the improvement of business

metrics has a higher priority than that of technical metrics for business

stakeholders. They can also benefit from data storytelling techniques in conveying

how AI projects can help with business metrics—whether it is an improvement in

customer experience, an increase in retention rates, or a decrease in cost.

An example of this in action is from The Economist and how it leverages Instagram

as a medium to an audience that it could not with its print version. Knowing that its

Instagram audience is younger than those of its regular readers, it highlighted

stories of interest to its younger audience, running the gamut from climate change

to celebrity news.

Fig 1: The Economist published data stories that are relevant

to its Instagram audience

4
Empathize with your audience

Data stories crafted with the audience in mind have a higher chance of retaining
the audience’s attention. Empathizing with your audience might influence the
content of the data story, the tone of the presentation, the length of the
document, and even the rate of speech. Here are some questions that might help
you understand and empathize with the audience better
Does the audience have the necessary prerequisite knowledge to understand
a particular metric?
How much time does the audience have to consume this data story?
What is the medium of presentation (written/oral) that the audience prefers?

Know their level of data literacy

The Economist demonstrated empathy when designing their data visualizations

on Instagram. On one occasion, it decided not to publish a chart (initially meant
for print) because it required niche prerequisite knowledge that an online
audience is unlikely to have. In another instance, it distilled only the main
takeaway of the chart and did away with the superfluous because a reader on
Instagram is likely to expect immediate gratuity than one on print media.

The Economist is not afraid to present complex charts that require high data
literacy to understand since it is targeted at highly educated business
professionals. Data storytellers should be aware of the data literacy of their
audience when designing charts to avoid underwhelming or overwhelming them.
Whenever possible, data practitioners should avoid technical jargons that might
confuse non-technical stakeholders, and instead use business metrics to convey Fig 2: How the Economist simplified

the impact of a data project. its chart for Instagram

5
2. Begin with the goal in mind
“Success in data visualization does not start with data visualization,” proclaimed "When it comes to the form and function of our data
Cole Knaflic, the author of the influential book Storytelling with Data. In the book,
Knaflic emphasized the importance of understanding the end goal and the visualizations, we first want to think about what it is we
context of a data story when starting a data visualization.

want our audience to be able to do with the data

Before creating your data story, you might find it useful to ask yourself the (function) and create a visualization (form) that will
following questions allow for this with ease."
Who are you communicating with? The more specific is the target audience, — Cole Nussbaumer Knaflic, Author of
the easier it is for the analyst to build a targeted data story that resonates Storytelling with Data: A Data
with and meets the needs of the audience. As mentioned previously, Visualization Guide

understanding the target audience also helps the analyst decide on the level for Business Professionals
of complexity of the data story
What do you want the audience to know or do? The main goal of building a
data story is to communicate a point and/or to give recommendations.
Knowing what exactly the point or recommendation is can help the analyst
stay focused on the main point instead of beating around the bush. A good
data story drives insights, which in turn drives decisions and calls for actions
How can you use your data to convey your point? With an understanding of
the audience and the intended action, we can start gathering the necessary
data evidence that supports the story arc.

The answers to these questions are ingredients for a coherent data story. Once
that is done, we can start building data visualizations that support the story.

6
3. Choose the best visualization for your story
The process of choosing the correct type of visualization for a particular In general, data visualizations can be used to show comparison, relationship,
insight requires an understanding of the different chart types. Here are the composition, and distribution. The following mindmap is a guide for
four most common types of charts. determining a suitable chart based on its purpose.

Bar charts are best suited to show comparisons of different categories Description Examples
Line plots are useful to show the changes of a variable over time.
Histograms show the distribution of one variable. This tells us how Bar charts Best suited to show comparisons A bar chart showing the reasons for
frequently a particular value occurs relative to others. of different categories. users to unsubscribe from the
Scatterplots can be used to show the relationship between two company s newsletter. Fig 3A
’ (

A bar chart showing the brea down of

variables.

company cost into different sources

(mar eting, operations, IT, etc
k )

Line plots Useful to show the changes of a A line plot of the company revenue
variable over time. over time. Fig 3B
(

A line plot of the app rating or user

satisfaction score over time.

Histograms Show the distribution of one A histogram showing the distribution

variable. This tells us how of user spending in the past quarter.
Fig 3A: A bar chart Fig 3B: A line plot frequently a particular value (Fig 3C
occurs relative to others. A histogram of the distribution of
employee satisfaction score of to
( 1 5)

from an H survey.

Scatterplots Used to show the relationship A scatterplot that shows the

between two variables. relationship between the time a user
spends on the app and the customer
lifetime value. Fig 3D
(

A scatterplot that shows the

relationship between the length of a
Fig 3C: A histogram Fig 3D: A scatterplot video advertisement and the
clic through rate.
k

7
 Such a claim ignores the fact that global temperatures naturally fluctuate from

4. Do not lie with data stories time to time due to events like El Ninos, volcanic activities, and ocean
conditions. When such effects are taken into account, and data are presented
from the 19th century to now, it is clear that the earth is warming.

Presenting cherry-picked data is fallacious and arguably unethical. Conceivably,

analysts who commit such an error run the risk of losing their hard-earned
As the saying goes, “if you torture the data long enough, it will confess to
credibility. Conversely, presenting the complete story, warts and all, is essential
anything.” Statistics can be manipulated to support any conclusion, but that
in gaining the trust of the audience.

certainly doesn’t mean that it should be.

It is the responsibility of the data storyteller to represent the data truthfully.

For example, multiple research projects corroborate the extent of climate
Apart from avoiding cherry-picking, data storytellers should strive to avoid data
change. Yet even a phenomenon as irrefutable as climate change can be
misrepresentation. To do so, the book How to Lie with Statistics offers a few
dismissed by climate change deniers with malicious data manipulation.
pieces of advice.
Presenting cherry-picked data, they pass pseudo-evidence as facts while
discrediting greening efforts as alarmists. “Global warming is not real because
global temperatures dropped in the years 2008 and 2011,” they claim. An
Ensure that axes start with zero since manipulating axes range can
observer who sees the chart for the first time might agree.

result in misleading correlations.

Fig 4: A cherry-picked chart on Fig 5: The complete chart on global

Fig 6: The same data points can convey
global temperature
temperature

dramatically different results

8
 Ensure that the axes scales are appropriate

Fig 7: By changing the scales of the y-axes, a forced correlation emerges

When needed to select a sample for a data story, ensure that the sample is

representative of the population. A sample that is too small can yield

extreme (very large or very small) results, whereas an unrepresentative

sample can yield incorrect inference about the population.

Use mean, median, and mode appropriately to ensure that the average is

truthful to the population especially when dealing with a skewed dataset.

9
5. Keep your visualizations minimal and avoid clutter
Clutter in visualizations takes up the cognitive load while not providing Calories per 100g
additional value. Removing distractions from a visualization focuses the 607
attention of the audience on the core message of the chart.

542
533

296
To quantify the amount of unnecessary data in a chart, Edward Tufte presented 260

an influential concept called the data-ink ratio, which is the proportion of ink
used to present the actual data to the total ink used to print the graphic. Good French
Potato
 Bacon Pizza Chili
graphics should have a high data-ink ratio.

Fries Chips Dog

To maximize the data-ink ratio, analysts should strive to minimize superfluous Fig 8A: A cluttered chart Fig 8B: The same chart, decluttered
words, colors, and lines without sacrificing clarity. In particular, here are some
ways one can keep visualizations minimal yet effective.

Use white spaces

Remove chart borders

Remove gridlines or axes

Clean up axis labels

Label data directly (as opposed to using a legend)

Remove data markers

Use special effects (bold, underline, italics, shadows) sparingly

10
6. Add color to your stories Fig 10 Use of color intensity

to highlight prevalence of

COVID-19 testing sites

Colors serve three fundamental use cases in data storytelling. We illustrate these
use cases using FiveThirtyEight’s colorful visualizations.

First, it is used to distinguish between groups that do not have an intrinsic order.
A prime example is the use of blue for Democrats and red for Republicans as in
Figure 10 below.

Third, it is used to highlight specific elements in the data. For instance, a

dark color can be used to highlight amongst other elements of lighter color.

Colors are a double-edged sword of data

storytelling. While they can inject life into
data stories, they can also potentially kill the
interest of the audience. To avoid this pitfall,
Datawrapper suggests the following best
practices.

Explain what your colors encod

Use colors sparingl
Use light shades for low values and dark
shades for high values
Fig 9 FiveThirtyEight's use of colors to Use intuitive colors (e.g. red for bad,
distinguish between republicans and democrats green for good
Use gray generously for less important
Second, colors are used to represent data values. Here, different shades of
data
similar colors represent the varying magnitude of a metric. In the following Fig 11: Georgia is highlighted since
Cater to color-blind audiences

example, we can immediately identify areas with darker colors as locations of it is the main character of the
busier Covid-19 testing sites. data story

11
 When the goal of the visualization is to drive a certain action, using actionable
7. Use texts appropriately chart titles can quickly call the audience to action. It is also helpful to add notes
to the footnote section of the chart for further details.

Texts clarify the meaning of charts and make them accessible. Too much text
adds clutter while too little text causes confusion. Here are some general rules
for using texts in your data stories.

Label data points when necessary

At times, it is useful to label important points directly on a graph to highlight a

certain point. In the chart below, we see that the points are labeled to
emphasize the diverging pattern of tickets received and processed.

Fig 12: Data points not labeled 


Fig 13: Data points labelled to
convey a point

Fig 14: Use actionable titles

Label axes and titles for clarity

Choose different font sizes

Without the axes and the titles, the audience is left guessing the meaning of the You may also notice the differences in font sizes in the charts above. In general,
chart, as is the case for Figure 14. Thus, it is crucial to add texts to name the font sizes need to be legible, and the difference in sizes depends on the relative
metrics being plotted to avoid miscommunication.

importance of the texts in conveying the message. In the chart above, the call to
action is the most prominent while the footnote is the least thanks to the choice
of word sizes.

12
8. Develop a narrative around your data
Cognitive psychologist Jerome Bruner suggests we are 22 times more likely to
remember a fact when it has been wrapped in a story. A narrative connects the
"The power of data storytelling is if we combine the
dots between facts and makes them more memorable.

right data with the right narrative and the right visuals,
The first step to data storytelling, according to Tableau’s whitepaper, is to find we have something that's very powerful that can really
the data story. To do so, one needs to identify the core elements of the story.
drive change and alter people's perspectives."
Who is the protagonist? — Brent Dykes, Effective Data Storytelling

What is the challenge? Author of Effective Data Storytelling: How to
What should the audience do at the end of the story?

Drive Change with Data, Narrative, and Visuals

With these elements identified, one can write out the storyboard, which helps
the data storytelling in identifying the key visualizations that support the
storyline. Metaphors and anecdotes are also particularly helpful in making the
story more personal and memorable.

Hans Rosling, a global health expert and the founder of GapMinder, developed a
narrative around global health data to effectively debunk myths about the
developing world in his influential TED Talk Stats that Reshape Your Worldview.

In this data story, the developing countries (the protagonists) overcame the HIV
epidemic and overpopulation (the challenge) with public health measures before
eventually reaching life expectancies on par with developed countries (the
ending). As the audience engaged with the visual narrative, Rosling is one step
closer to Gapminder’s goal of overcoming systematic misconceptions about
global trends.

13
Everyone will be a data storyteller

Pablo Picasso once espoused, “learn the rules like a pro, so you can break them
like an artist.” While these eight rules serve as a general guideline for data
storytellers, they are by no means the commandments of data storytelling.
Data storytellers should recognize that there are always exceptions to the rule,
much like how the Economist consciously broke the rules in data visualization
in creating masterful data stories.

One should also note that these eight rules to data storytelling are by no
means exhaustive. Instead of aiming to learn every single theoretical rule, one
should aim to start crafting and telling their first data stories as soon as
possible. The experience gained from data storytelling will pay dividends as
data storytelling becomes a necessary skill set for all in the coming decade,
and data insights become a cornerstone of organizations.

14
Build a Team of Data Visualization Data Visualization Courses
Experts with DataCamp at Your Disposal
Take your data storytelling to the next 
 Data Visualization with Python
level with Data Visualization best practices DataCamp provides a host of data visualization courses across the most popular Python
packages. Whether it’s matplotlib, seaborn, bokeh, plotly, or dash, you’ll be able to develop
DataCamp’s extensive data visualization curriculum can help the data succinct data visualizations and dashboards that are deeply customizable for your data stories.
storytellers of the future hone one of the key elements of data storytelling:
DataCamp’s proven learning methodology provides a cyclical process for
learning and retention. This learning methodology enables learners across the
data literacy spectrum to assess their skills and identify gaps, develop a
learning plan based on these gaps, practice skills, and apply them in a real-
world setting. Practitioners of any skill level can upskill on the latest data tools,
techniques, and concepts.

Assess Learn
Test your skills
Complete interactive
and track progress courses

Apply Practice
Solve real-world Practice with quick
problems daily challenges

Speak with a learning expert

15
 Data Visualization with R Tell your Data Stories with Tableau and Power BI
If you’re looking to sharpen up your data visualization skills in R, we also have you covered. R is Business intelligence tools were made with data storytellers in mind. Powering the analysts of the
considered one of the best tools for data visualization and reporting. Whether using ggplot2, future, business intelligence tools like Tableau and Power BI provide easy drag-and-drop
leaflet, or plotly for data visualization, or the shiny package for developing interactive interfaces to design and deploy dashboards that walk consumers through a data story.
dashboards, you’ll be able to design better visualizations that complement your data story.

16
 Upskill your team Track skills development with skill matrix
Track the data skills your team has today and map a path to the skills they need tomorrow.
Using the Skill Matrix, admin users can easily filter to identify individuals with the skills you
need to take on specific projects or teams with low use or data skills gaps. They can then
Create Custom Tracks create and assign custom tracks to help bridge these gaps and report on skill development.
DataCamp makes it easy for you to create bespoke learning
paths and assignments to meet the needs of all your roles,
teams, and departments.

Certify your people with certifications

Upskill with an industry recognized credential from the leader in data science education*.
Our Data Scientist Professional Certification lets businesses recognize qualified talent,
both internally as well as during the hiring process.

Set Assignments
Assignments are a great way to set clear, time-sensitive learning goals.
On average, courses assigned by Enterprise customers have completion
rates that are twice as high as unassigned courses.

*Based on competitive analysis and current as of June 2021:

Gain XP Course Chapter datacamp.com/why-datacamp
Take your team’s data literacy to the next level - 

Custom Track Assessment Project with customized, tailored data training
Speak with a learning expert
17