Intro to Bubble Charts by BECKON

1
As we increase social
spend, what happens to
downloads and revenue?
How does search affect
revenue and social
volume across brands?
What messaging, across
all channels, is resonating
most with our target
audience?
What kind of chart should you use to get answers
to critical business questions like these?
What happened to CPM
when we turned on
programmatic in those
two markets?
Which of our agencies
delivers the most net new
leads and social buzz per
dollar we pay them in fees?

2
Bubble charts are ideal for understanding the
relationship between three categories or dimensions.
They’re a close cousin of the scatter plot, but where a
scatter plot maxes out at two dimensions (whatever
is on the X-axis and the Y-axis), a bubble chart shows
three—the size of the bubble being the third value.
And—bonus!—we can illuminate a fourth category or
dimension by using bubbles of different colors.
A bubble chart
• Are visually cheerful
• Encourage curiosity about the data
• Use easy-to-understand shapes, colors,
sizes and patterns to enable insight
• Are statistically powerful
• Show relationships between outcomes
• Give a good sense of correlations, both
positive and negative
We love bubble charts. Why? Well, because they:

3
Let’s walk through a hypothetical example. Imagine
we’re marketers at a media and entertainment
company. We spend a significant amount of our
marketing budget on search advertising and social
media, with the goal of persuading folks to go to the
movies, the success of which is measured in revenue.
Going in, we’re pretty sure search volume and social
conversation volume have an impact on revenue—
we certainly hope so! But we’d like to more fully
understand the relationship. We also have anecdotal
evidence that different movie genres respond
differently in terms of search and social—we’d like
to know more about this as well.
We’ll use a bubble chart to plot search volume
against revenue, and show conversation volume
as a third factor. Plus, we’ll use differently colored
bubbles to represent different movie genres for
added context.
Let’s ﬂow our data into a bubble chart and
see what we can see!
Here are the measurable
items in question:
• Search volume
• Social conversation volume
• Revenue
How to read a bubble chart

HOW DOES SEARCH AFFECT REVENUE AND SOCIAL
CONVERSION VOLUME BY MOVIE GENRE?
BOXOFFICEDAILYREVENUE
GOOGLE SEARCH INDEX
$2BN
$0.00
0 15K10K5K
R2
= 0.92 ALL
$4BN
$6BN
$8BN
$10BN
$12BN
4
To start, we can see that search volume is on one
axis and revenue is on the other.
We see some bubbles are bigger than others, and
they’re sporting a lovely color palette. Also, some
bubbles are clustered together, some are hugging
the diagonal line that rises from left to right, and
some are hanging out on their own.
At the bottom left of the line we see a number:
R2
= .92

GOOGLE SEARCH INDEX
$2BN
$0.00
0 15K10K5K
R2
= 0.92 ALL
$4BN
$6BN
$8BN
$10BN
$12BN
5
Each bubble is a different movie. A bubble’s position
on the X-axis shows how much revenue it pulled in.
Its position on the Y-axis indicates search volume.
The size of each bubble corresponds to social
conversation volume; bigger means more volume.
And the different colors of the bubbles delineate
different genres.
What is the chart
showing us?

GOOGLE SEARCH INDEX
$2BN
$0.00
0 15K10K5K
R2
= 0.92 ALL
$4BN
$6BN
$8BN
$10BN
$12BN
6
There’s also a line rising from left to right. The line
and the label R2
= .92 at the lower left—called the
R-squared number—from the statistical backbone
of the chart.
• The line—called a trend line or linear regression
line—represents an equation describing the
best-fitting straight line through the various
X, Y points (that is, the center of each bubble).
• The R-squared number, simply put, is a measure
of how tightly the bubbles cluster along the line.
R2
values are always between 0 and 1 and are read
as a percent, so the range is always 0% to 100%.

GOOGLE SEARCH INDEX
$2BN
$0.00
0 15K10K5K
R2
= 0.92 ALL
$4BN
$6BN
$8BN
$10BN
$12BN
7
Ok, so back to our example. Looking at this chart,
we can say that there is a strong correlation between
search volume and revenue. Further, we can say that
different movie genres exhibit correlation nuances—
this can be seen by how closely (or not) bubbles of
each color hug the trend line. In fact, we can say that
92 of the time there is a correlation between search
and revenue, and that some genres show better
correlation than others.
This is great! We definitely have an argument for our
search budget!

8
With a solid answer to our initial question in hand, we can
dig deeper:
• For the movie genres that are not as closely correlated, is
there a marketing tactic other than search that is more closely
correlated to revenue?
• How can we reallocate search spend to maximize potential
revenue?
• How does the correlation between search and revenue change
if we look only at genres with high (or low) conversation volume?
• There are some “outliers” with big revenue. What useful info
can we learn about those?
• What are some other variables we might want to investigate
for correlations? Are my results different by age groups?
By gender? By region?
These follow-on questions spring from the example
we used here of a media and entertainment
company, but you can easily translate them into
questions that are critical for your own business. The
takeaway is this: By starting with a straightforward
question, usually stemming from “hunches” that you
already you have, it is easy to organize your thinking
about what to explore next, how to prioritize, and get
statistically relevant insight.
Ideas for further exploration

9
Always keep in mind that relationships between variables is
not the same as one variable causing the other variable. That’s
okay, though—confirming a relationship (or the absence of a
relationship) is in itself super useful.
Second, R-squared analysis, like all statistics, needs to be used
carefully. And there are caveats. For example, we’ve been
looking at a linear trend line here, but certain data may exhibit
characteristics that call for a different kind of trend line. Linear
trend lines will get you pretty far, just know that there are other
kinds, too.
Finally, keep it simple. Gleaning just one insightful correlation
that you can act on can deliver enormous value in terms of how
you allocate your budget, how you optimize your marketing
tactics and how well you understand your customers.
Correlation is not causation (and other tips)

LOOKING TO IMPROVE TRANSPARENCY
AND ROI IN YOUR AGENCY PARTNERSHIPS?
Get in touch at hello@beckon.com. We’d love to hear from you.

Intro to Bubble Charts by BECKON

Recommended

Recommended

More Related Content

Similar to Intro to Bubble Charts by BECKON

Similar to Intro to Bubble Charts by BECKON (20)

More from Amanda Roberts

More from Amanda Roberts (12)

Recently uploaded

Recently uploaded (20)

Intro to Bubble Charts by BECKON