Adventures in Visualand: Data visualization today (Episode 4)

The three most common mistakes of 'chart junk' / bad dashboards

The greatest value of a picture is when it forces us to notice what we never expected to see.
— John W. Tukey, Exploratory Data Analysis, 1977, Pearson

Old Bling, New Bling. Still Not Da Thing

People can get bored with pie charts. So first they start using expanded pie charts, with multiple measures, combined inner circles and large arrays of colors. Then they get bored with complicated pie charts, too, and start using bubble charts, using color, size and multi-axis interpretation for each data set they analyze. But they get bored with bubble charts, too.

It is scientifically true: familiar visualizations, such as pie charts, bar or column graphs, are not as memorable as novel visualizations because novelty is ‘sticky’.

Therefore, they start using the new rave tools: geocoding and heatmapping, changing color scales, icon patterns, and categories simultaneously represented on the map. Then they get bored of geocoded maps, too, so they start using 3D representation, using depth and filter levels and network nodes for illustrating more and more complicated datasets.

But at one point, they stop. It is not about outdated models or novel types anymore. It is about efficiency. Why? Because dashboards can be complex and still have a clean and understandable design. Complicated does not automatically mean more value.

The Good, The Bad & The Unexpected

There are many negative examples of dashboards and whether they are called chart junk or just bling visuals, they all have a common feature; they won’t allow the viewer to discover any relevant information about the presented data, let alone reveal elements that may help a company build a competitive edge. We are talking about the ability to go in-depth, and bad dashboards are usually so complicated or flashy that you cannot grasp even the first layer of information, let alone what underlies it.

Here are just a few examples explained in this context:

Figure 5: Timeline


Figure 6: Example of an incorrectly used treemap


Almost all negative examples of dashboards have three main errors in common:


e.g. attributing different colors to many categories, choosing too varied or close nuances for comparison or using flashy colors solely for the purpose of grabbing attention, with no correlation to a specific trend within the data.


e.g. cluttering the dashboard with too many types of charts, including too many additional elements - text, images, icons, videos.


e.g. illustrating large datasets with graphic models meant for small and medium sets (column, bar or pie charts.

Read all the episodes:

Episode 1

Episode 2

Episode 3

Episode 5

Episode 6

Episode 7