Dashboard Design Mistake #4: Using charts and graphs that don’t work well on dashboards. Spoiler Alert: Save the pie for dessert.

You know which metrics you want to show on your dashboard, and many of them are actually a series of values – for example, the total monthly expenses for the six departments in your organization for the last 12 months.

You could just show tables of numbers but generally charts and graphs work much better because:

• They are compact: More information can be presented in a smaller area, making them particularly useful in dashboards.
• They are easy to scan quickly. Imagine trying to make sense of numbers in a 52-cell table representing total sales for the last 52 weeks. Then imagine how much easier it is to scan, say, a bar graph that presents the same information.
• They make it easier to see patterns. In the above example, unless you are a math savant seeing patterns in numbers versus a chart is much more difficult.

But there are two caveats:

• Levels of precision vary. A graph will never show values as precisely as a table of numbers, so if precision is paramount then having a table with a graph, or a table on its own might be necessary. Also some types of graphs allow users to “eyeball” the individual values they contain much more precisely than other types.
• Users have to know how to read them. Knowledge levels can vary among audiences, and it may be worthwhile testing some types of graphs with your users to make sure they are comfortable with them.

Not all graphs and charts work on dashboard

With the tables versus graph question answered your next decision is what type of chart of graph.

This is a really important question because not all graphs and chart types work well in a dashboard and some work well for some data but not other.

So it pays to understand what information you want to convey, and choose a chart or graphic that is suited to the task.

Here’s how some common graph types stack up:

Line charts

Compactness

Rapid scannability

Clarity

Precision

Familiarity

A great choice to display patterns of change across a continuum (usually, but not always, a continuum of time).

Bar charts

Compactness

Rapid scannability

Clarity

Precision

Familiarity

A great choice to enable users to rapidly compare items in the same category (e.g. customer satisfaction in several regions).

Pie charts

Compactness

Rapid scannability

Clarity

Precision

Familiarity

People are surprisingly bad at accurately comparing the sizes of the slices of a pie chart which is why top data visualization experts including Stephen Few tend to agree that they shouldn’t be used. This great article explains why. Instead of a pie they recommend a bar chart arguing it will generally do a better job of displaying the same data in about the same or less space.

Gauges

Compactness

Rapid scannability

Clarity

Precision

Familiarity

Most gauges convey just two pieces of information: a current value, and where that value falls along a scale that spans some variant of “bad to good”. Just showing the value and making its text colour correspond to a point on the bad-to-good continuum (i.e., red to amber/grey to green) scale would convey the same information in about one tenth of the space.

Since real-estate is almost always at a premium on dashboards, most data visualization experts strongly recommend you don’t use them. However, if you have the space or if you have stakeholders who insist on using them, consider making vertical the normal reading for the gauge so that it is easy to scan for exceptions - particularly if you have multiple gauges.

Sparklines

Compactness

Rapid scannability

Clarity

Precision

Familiarity

While their lack of a scale means that users will not be able to eyeball individual values with any kind of precision, the incredible compactness of sparklines often makes them the only choice for showing trends when you have a large number of metrics on your dashboard. But you should definitely make sure that your audience understands how to read them.

Scatterplots

Compactness

Rapid scannability

Clarity

Precision

Familiarity

They are not generally useful on dashboards since they are not easy to scan quickly and the relationships between two quantitative measures (what a scatterplot shows) tend to not change very frequently. Plus, a surprisingly large number of people don’t know how to read them. However, they can be great for an interactive report for knowledgeable readers.

Maps

Compactness

Rapid scannability

Clarity

Precision

Familiarity

When deciding whether or not to use a map on a dashboard, the main question to ask is, “Is geography an important part of the story?” If not, a bar chart would probably show the same data in less space, and with greater precision.

Here are some chart types that experts agree (most anyway) never work well on dashboards because data can always be represented in a more scannable, clear, precise and familiar way with one of the above charts:

• Radar graphs
• Bubble charts
• Anything with a third axis (surface graphs, 3D line graphs, 3D bar graphs, etc.)

Again, they may work well for Infographics, reports or in-depth analysis just not dashboards.

Have a chart you love to use on dashboards? Disagree with any of the recommendations? Would love to hear from you.

Next in this series: Common dashboard design mistake #5: the misuse of colour.