Adobe Analytics Data

Visualization Playbook.
The Art and Science of Data Visualization.
The two objectives of data visualization, analysis and storytelling, are essential
skills for an analyst.
• Data analysis is the science of exploring data in an attempt to derive
useful information.
• Data storytelling is the art of effectively conveying that information
to an outside audience.

Getting good at data visualization is both an art and a science. There are
a variety of factors to consider when making visualization decisions. This
playbook walks through use cases of when you should pick one visualization
over another. It provides a list of common visualization objectives along
with examples.

These are guidelines, not rules. And for many of these rules, you’ll probably
be able to think of counterexamples. That’s fine. Just be thoughtful about your
visualizations. Play around with them, and try a couple options to see what
works best.

1
Table of contents
I want to show how a metric changes over time. 3

… with 4 or more metrics. 4

… where dimension values should add up to a total. 5

… with metrics of a different scale. 6

I want to compare values within a dimension. 7

… where values add up to a 100%. 8

… where values add up to a total. 10

I want to explore how users are navigating through 10

my site.

I want to evaluate the performance of 11

a conversion funnel.

I want to monitor progress towards a goal. 11

I want to show the distribution of a variable. 12

I want to highlight a single metric value or change 12

in value.

I want to show where visitor traffic is coming 14

from geographically.

I want to explore how visit or visitors overlap. 14

I want to show how two metrics are correlated 15

based on different dimension values.

I want to see how different groups of visitors return 15

to the site over time.
2
I want to show how a metric changes over time.
Continuous datasets (or time-series data) is best represented with a line or area
chart. If you are able to group that data into meaningful categories (e.g., month),
then you could also use a bar chart.

Example: Site visits trended monthly over the last six months.

Line chart
This is the default way to display time
series data in Analysis Workspace. It can
handle the most complexity of any time
series visualization.

Area chart
This is a line chart with the space below
shaded in. Under certain conditions, an
area chart might be easier to read than
a line chart.

Bar chart
Instead of a continuous line, a bar chart
calls out each data point distinctly.
Use this when you want to call out your
data points.

3
I want to show how a metric changes over time.
…with 4 or more metrics.
Maybe you need to show how a lot of metrics are trending over time. This can
easily lead to an overcrowded visualization. There’s not a common “best” option
for displaying this type of data, though a standard line chat will usually get the
job done. If possible, spend a little extra time making your visualization as readable
as possible.

Example: Weekly visits by marketing channel.

Good
Line chart
The basic line chart is almost always
going to be your best option when
displaying four or more different
trendlines.

Bad
Bar chart
My rule of thumb with trended bar
charts is no more than two metrics.
But sometimes rules are meant to
be broken.

Unappealing
Area chart
In some cases, four or more metrics
in an area chart might be ok. But if
it’s hard to read, use another option.
Make sure to indicate this is not
stacked, since with area charts there
is no clear difference between standard
and stacked.

4
I want to show how a metric changes over time.
…where dimension values should add up
to a total.
Sometimes your metrics should add up to a total at the top. For this, you’ll use
the stacked version of area and bar charts. Note that these charts make it easy to
see changes in the total but make it more difficult to tell changes in an individual
dimension value.

Example: Weekly visits by marketing channel.

Stacked area chart

The stacked area chart looks cleaner
than the stacked bar chart, especially as
you increase the number of data points
along the x-axis.

Stacked bar chart

The stacked bar makes it a little easier
to see changes in individual dimension
items than the stacked area. If that’s
important, consider this chart.

5
I want to show how a metric changes over time.
…with metrics of a different scale.
What if the metrics you want to trend are on completely different scales?
A common example is when you want to display a whole number metric, like
page views, alongside a “rate” metric like Bounce Rate.

Analysis Workspace makes this possible with a few different visualizations.

However, you are limited to two total metrics in the chart.

Example: Page views and bounce rate trended weekly over the last 12 full weeks.

Wrong
Single axis line chart
This example shows page views and
bounce rate with a single axis. The
bounce rate line is impossible to read.

Good
Dual axis line chart
This shows the dual axis option. Now
both trend lines are clearly visible.
The y-axis labels and line color match
for additional clarity.

Good
Dual axis line chart
The area chart again is very similar
to the line chart. It gives a little more
weight to the trendline.

Bad
Dual axis bar chart
While you can create a dual axis bar
chart, you shouldn’t use it for time
series data. It’s too difficult to read.
Instead, use a dual axis bar chart to
compare dimension values (see below).

6
 Advanced
Combo chart
This may be the best dual axis
visualization option. Combining two
chart types — line and bar — further
differentiates the metrics, making it
clear they are not the same thing.

I want to compare values within a dimension.

Another common visualization goal is to compare dimension values for a given
metric. This gives us more visualization options since we now have control over
what goes into the x-axis.

Example: Visits by marketing channel.

Bar chart
The standard bar is the most common
way to compare dimension values. It’s
likely the one your stakeholders are
most comfortable with.

Horizontal bar chart

The horizontal bar charts are great
because of their efficient use of space.
Notice how the value titles on the y-axis
fit neater than in the bar chart above.

Example: Top 20 pages.

Bad
Bar chart
When you have this many dimension
values to compare, it’s best to use a
freeform table.

7
 Beware
Horizontal bar chart
It’s hard to tell, but this chart has 20
bars and only 10 page names. Since the
text won’t fit on the chart, every other
page name label is dropped. You can
fix this by increasing the height of your
visualization. (See next example.)

Not Bad
Horizontal bar chart
Manually increase the visualization
height to make sure all the values
are displayed.

I want to compare values within a dimension.

…where values add up to 100%.
Example: Visits by mobile device type.

Good
Donut chart
Pie charts and donut charts get a lot
of grief. However, when you’re looking
at four or fewer values, they can be an
easy way to show how given values
make up the total pie (or donut).
Tip: When looking at mobile device type in a
visualization, update the default “Other” to be
“Desktop/Laptop.”

8
Example: Visits by mobile device type by day of the week.
Note: Make sure to check the “100% Stacked” box in the gear option menu.

Good
100% stacked bar chart
This looks better when you have more
equity across your dimension values.

Good
100% horizontal stacked bar chart
In this example, tablet and mobile
phone visits are much easier to read
with a horizontal bar chart.

Example: Visits by marketing channel.

Note: In this case, we have 12 dimension values (i.e., marketing channels). Treemap is the only good visualization that shows how that
many dimension values add up to a total.

Bad
Donut chart
Do not use a donut chart when you
have more than four dimension values.
There are better ways of visualizing
this data.

Advanced
Treemap
An underutilized “donut chart”
equivalent that does allow for the
inclusion of many dimension
values is treemap.

9
I want to compare values within a dimension.
…where values add up to a total.
This is similar to the previous use case (where values add up to 100%), but here we
are more interested in the aggregated total than we are in the makeup of the total.

Example: Top 5 marketing channels by mobile device type.

Good
Stacked bar chart
The most important comparison here
is between the marketing channels.
“Other campaigns” is driving the most
traffic. Within that, we can see what
devices people are on when coming to
the site from these marketing channels.

Good
Horizontal stacked bar chart
It’s easier to see how mobile phone and
tablet are contributing thanks to there
being more horizontal space.

I want to explore how users are navigating

through my site.
Example: How are people navigating to and from the home page?

Good
Flow
While flow visualizations can get out
of control quickly, they can also be
useful for discovering how visitors are
navigating through your site.
Tip: For more relevant results, add constraints
to your flow analysis.

10
I want to evaluate the performance of a
conversion funnel.
Example: Evaluate visitors going from a product view to a cart addition to a checkout.

Good
Fallout
When you have a defined “happy
path” (e.g., shopping cart or conversion
funnel), you can use a fallout
visualization to assess the path’s
performance and see where visitors
tend to drop off.

Even Better
Fallout comparison
Add segments to compare fallouts from
different types of users.

I want to monitor progress toward a goal.

Example: Site visit goal.

Bullet
If you have predefined goals for
a given metric, a bullet chart is a great
way to visualize progress to that goal.
The x-axis can display as either a total
number or percentage.

11
I want to show the distribution of a variable.
When a mean or median doesn’t provide enough information (or is potentially
misleading), showing the distribution of a metric can be helpful.

Example: How many product views do visits with a cart addition have compared to visits without
a cart addition.

Advanced
Histogram
You can use the “View Data” option
to add two columns to the underlying
freeform table.

Good
Histogram
This shows the distribution of product
views across visits.

I want to highlight a single metric value or

change in value.
Example: Show visits and month-over-month change in visits.

Summary number
Use this when you want to call out
an important number. Make sure
to lock selection.

12
Summary change
Use this to show a percent change.
Select the comparison number first and
then the new number. Make sure to
lock selection.

Advanced
Key metric summary
Use this to show both a summary
change, summary number, and
trendline all in one visualization. Play
around with the options until it looks
how you want.

13
I want to show where visitor traffic is coming
from geographically.
Example: How many visits are we getting from across the United States?

Map (heatmap)
Use the heatmap version of the map
visualization to get a quick idea of
where your visits are coming from
geographically.

Map (bubbles)
Use the bubbles version of the map
visualization to see the actual number
of visits coming from different locations.

I want to explore how visits or visitors overlap.

Example: Show the overlapping marketing channel mix of visitors to our website.

Good
Venn diagram
This shows how the same visitors
come to the site through different
marketing channels.

14
I want to show how two metrics are correlated
based on different dimension values.
Example: How do marketing channels compare based on two metrics: bounce rate and
checkouts per visit?

Good
Scatterplot
This is a great way to show how
different marketing channels perform
based on two unrelated metrics.

Advanced
Scatterplot
Add a third metric to turn this into
a bubble chart.

I want to see how different groups of visitors

return to the site over time.
Example: How many weeks after a visit does a checkout occur?

Advanced
Cohort table
The conditional formatting helps us
quickly identify which cohorts are most
effective at driving checkouts after
a visit.

15
 Example: What desktop browser is driving the highest rate of return visits?

Advanced
Cohort table
Try using a dimension-based cohort to
compare return visits from the various
dimension values.

Example: When did visitors visit before and after a cart addition?

Advanced
Cohort table
Use a latency table to show how many
visits occurred leading up to and then
after the inclusion event.

Author
This document was created by David Geist,
data and insights business consultant at Adobe.

© 2023 Adobe. All rights reserved.

Adobe and the Adobe logo are either registered trademarks or

trademarks of Adobe in the United States and/or other countries.
16