Why not Pie?

In pie chart’s defense…

from vectorcove.com

Pie charts get a bad rap

They’re the chart that people love to hate. Using them properly can be tricky. And there’s plenty of bad examples out there in the wild. Despite how easy they are to get wrong, I love pie charts… and many users love them too. There are certain situations in which pie charts excel 🙌. Let’s look at some of things that can contribute to successful pie chart implementation.

There will be interactions

One factor that often gets omitted from pie chart love/hate discussions is interactivity. Many of the bad examples of pie charts depict a static chart with no dynamic behavior. Point being, pie charts can be hard to understand or downright misleading when you’re left with nothing but a series of arcs and a couple of labels. It can be difficult to accurately interpret what you’re seeing (especially with several similar values, or worse — 3D skewing 🤢). In contrast, when a pie chart has interactivity, you don’t have to guess what the difference is between those two slices/arcs that are similar in size… just mouseover or click for more info. Mousing over for more information is a common and expected interaction on the web. We often see it with unfamiliar icons and buttons. It makes perfect sense on chart elements as well and helps give users the specifics on-demand.

The dataset is small

Not this.

As a recommendation, I’d say pie charts should stay somewhere in the ballpark of 4–6 slices. Datasets with over 6 slices should consider using some threshold to aggregate values into an “Other” category. One caveat, if “Other” becomes the biggest slice on the pie chart, it can unintentionally steal too much of the focus. So consider removing unimportant values altogether and/or using a title that implies that only a subset of data is being displayed, such as: Top 5…

There will be annotations

Children/Dependents, Finance & Insurance, and Personal & Medical are similar in value, but the annotations help clarify!

One of the arguments against pie charts is that human perception isn’t very accurate when judging the difference between curves and arcs. This is true and there’s been plenty of research to back it up. But how often do you see a data visualization of any sort without an axis, label, or some other “data ink” used to clarify the message? Pie charts are no different. But instead of X or Y axis ticks, show specific values and other supplemental info. Looking at the above example, theres a total in the middle, along with labels and values to help clarify the data that you’re looking at. Could you have swapped this pie chart for a column/bar chart? Maybe? But, why not pie?

Proportion, proportion, proportion

Query Builder (with User Flow) by Andrew Lucas. Note the interactivity and annotations.

Pie charts are great at showing proportion. Further, they’re great at articulating changes in proportion over time (+1 for interactivity). With the above example in mind, as you modify the query, you get an up-to-date reflection of how your query results compare to a given point of reference, or the total available data. The more specific your query is, the smaller your result set, the more generic your query is, the larger your result set. Getting a live preview of what your result set looks like is powerful. Thanks pie chart.

When visual interest is more than a “nice to have”

I agree with the mindset that form follows function. And I’m definitely in favor of minimal and bare-bone visuals that “get the job done”. However, an attractive product is also a product that is likely to impress potential users and gain new business. With that said, once you’ve nailed down the functionality, pie charts add a lot of visual appeal. I’d say their symmetry and the inherent positive connotation of circles is a part of the reason why.

Circles protect, they endure, they restrict. They confine what’s within and keep things out. They offer safety and connection. Circles suggests community, integrity, and perfection.
Because they are less common in design they work well to attract attention, provide emphasis, and set things apart. ~Steven Bradley

Pie charts are a good mechanism for adding some visual interest within a sea of “better-suited” visualizations. Screw you bar charts.

The question is simple (i.e. users don’t need pin-point accuracy to make a decision)

Sometimes simple questions need simple answers. Yes or no? Is it a lot or a little? Is it more than half? Is it less than half? Who has the most? Who has the least? Lets say you’re building a blog and you want to use your various social media accounts to drive traffic/readers to your blog. You take a trip to your blog’s analytics dashboard and ask the following question…

“Which platform is generating the fewest leads?”

I think the above clearly answers that question. However, its a fair point to say that if the arc for Snapchat was a bit smaller it would be hard to determine which slice is smallest. To my earlier point—mouseovers to the rescue! On the other hand, maybe the answer becomes “both”.

In short. Have a slice. Your user’s might love it. Follow some of the tips above and you’ll have a better shot at using pie charts to convey your message in a clear and intuitive way.

Thanks for reading 😃! If you want to hear more about data visualization design and engineering follow me on Medium & Twitter @stevoscript.