Why a square filled doughnut is actually a waffle

At least once a week someone at Pellucid Analytics has a rant about a bad visualization. And while we admit we’re visualization nerds, we’re not the only ones who get upset about poorly presented data; just check out Junk Charts for some truly terrible charts.

However, you probably have better things to do than to gnash teeth over a badly labelled axis, so you may not be familiar with one of the biggest points of contention in data visualization: pie charts.

I doubt you have seen a pitchbook without at least one pie chart. In our audit of over 2,000 pitchbook charts we found that pie charts were one of the most commonly used visualizations.

So what’s the problem?

Too often, pie charts are used to communicate a point that they just aren’t designed to deliver.

Edward Tufte, a thought leader in the area of visual communication and visualization has long taken pie charts to task. In his book, The Visual Display of Quantitative Information, Tufte writes, “A table is nearly always better than a dumb pie chart; the only worse design than a pie chart is several of them for then the viewer is asked to compare quantities located in spatial disarray both within and between charts [...] Given their low density and failure to order numbers along a visual dimension, pie charts should never be used.”

I think he’s on to something; although I wouldn’t go as far as to say pie charts should never be used.

It’s widely believed that William Playfair introduced the first pie chart in his 1801 book,Statistical Brevity. Two hundred years ago data analysis was extremely limited, and probably pretty inaccurate, so a pie chart would appear a skilful way of expressing meaning from crude data collection. Much has changed since then, and there are many ways we can extract meaning from a dataset. In that light, it seems odd to gravitate toward a charting technique designed in a different time and under different data restrictions.

Pie charts should not be cut from pitchbooks entirely, just more thoughtfully used.

I met with a client recently whose standard ECM pages would have caused Tufte to go into a seizure. The book was like a pie explosion—pies upon pies upon pies—trying to reveal complex relationships across different investor classes and across time. They weren’t aligned on the page (my own go-to rant topic), they were impossible to read, and there wasn’t a clear message at all. It was a complete mess.

In my view, pie charts should not be cut from pitchbooks entirely, just more thoughtfully used. As with any data visualization, the first question should always be, “What am I trying to show”, and then the design should flow from there.

So, what is a good alternative to a pie chart?

Doughnuts are a good place to start. Yes, actual doughnuts. Homer Simpson doughnuts. If you’ve visited the Doughnut Plant in New York, you may be familiar with the superiority of its square filled doughnut. Called, “the most astonishing thing of all”by the New York Times they are what every doughnut dreams it could be. The reason they are so good? The filing is in every bite, not just toward the center.

Enough doughnut diversion, the point is this: A regular doughnut is to a square filled doughnut, as a pie chart is to a waffle chart.

A waffle chart is basically a square version of a pie chart but, in my view, raises the bar regarding aesthetics; they look like something out of a design studio, not a spreadsheet. A waffle chart uses a grid of differently colored squares (hence the name) to show proportion instead of pie slices, making it easier for the viewer to cognitively process the various data relationships. Also, since it’s less commonly used than its pie chart cousin, the waffle chart can serve as a differentiator, attracting client attention.

What does this look like in a pitchbook? Here is an example of a waffle chart showing the breakdown of institutional shareholders for a client:

Contrast this to the pie chart with the same information.

The difference is striking; the waffle chart looks pretty impressive. Moreover, even without value labels, it still delivers the necessary information.

You can even go further and break a waffle chart out into its components.

Not all pie charts should turn into waffles. As always, it depends on the information you want to convey and then critically assessing if your default go-to chart is doing the data justice.

What’s been your experience with waffle charts? Or doughnuts? I can talk about those, too. Email me at adrian.s.crockett@gmail.com.