Data Visualization: A Marketing Superpower - Clark Boyd
- 2. digitalmarketinginstitute.com 2
Hi! I’m Clark Boyd
- Course Leader at Cambridge University and
Columbia University business schools
- Former digital strategy lead for American Express,
adidas, and General Motors
- Sign up for my free, weekly newsletter:
https://hitech.substack.com/
- Subscribe to the YouTube channel:
https://www.youtube.com/channel/UCtjxTUewG
1TMj5WydVD8bhA
https://www.linkedin.com/in/clark-boyd-digital/
- 3. digitalmarketinginstitute.com 3
COMING UP
1. The human visual system
2. Data visualization & marketing
3. A four-step process for success
○ The right data
○ The right visual
○ Remove the clutter
○ Show AND tell the story
4. Getting started
- 4. digitalmarketinginstitute.com 4
By the time you look over here and read this,
your visual system has already processed that
image.
You have counted the people, you know what
color their drinks are, and you can describe
the scene.
- 10. digitalmarketinginstitute.com 10
Or we can take the same data
and display it as a histogram
with distribution line.
Now we can talk about the
overall shape of the data, but
we lose the precision of
individual data points.
These trade-offs are always at
play with data visualization!
- 25. digitalmarketinginstitute.com 25
WHAT DATA WILL ANSWER YOUR QUESTIONS?
Tip:
When you create a
visualization, ask
yourself: What is
the first sentence
that comes to
mind when you
look at the chart?
Then, ask
someone else!
- 26. digitalmarketinginstitute.com 26
I made up a dataset to help prove a
point. The data is about penguin
populations, but it is entirely fictional.
I know we have Humboldts in London
Zoo - here I am taking pictures of them
in the rain - but the rest I have
fabricated.
Just getting that in early before some
pedant points out pernickety
imperfections in my penguin population
plots!
- 30. digitalmarketinginstitute.com 30
Species England France Germany Sweden Belgium
Rockhopper 33 32 42 33 27
Adélie 34 44 31 12 25
Humboldt 18 11 32 22 1
Gentoo 34 23 14 6 5
Macaroni 25 22 45 11 11
Depending on the audience, we can
display 5 separate charts, with the
raw data for those who like to dig
deeper.
- 36. digitalmarketinginstitute.com 36
This chart combines multiple
techniques very effectively. Over
here, we have the broad sweep of
the lower rankings. We need less
precision here and a log scale helps
display this.
- 45. digitalmarketinginstitute.com 45
The real pros (I’m not one
of them) know when to
break the rules.
I wanted to show the
gender breakdown at
Winter Olympic games
since 1924, so I dutifully
created a series of pie
charts. It was a
comparison task, over
many cyclical periods of
time.
- 46. digitalmarketinginstitute.com 46
The image was a lot bigger than this, so
the text was legible.
But anyway, I highlighted the female pie
section and you can see that it grows
considerably.
A few weeks later, The Economist worked
with the same data.
- 50. digitalmarketinginstitute.com 50
He’s down there in the bottom left. Once you see him, it seems so obvious. You can’t
“unsee” him.
Professor Steven Pinker calls this the “curse of knowledge” and it crops up all the time
in data visualization. Once you see something in the data, it will jump out at you from
your charts.