Now that we know several of the common errors when making a pie chart, let’s look at the other side of the coin and see how we can improve this little chart (chartling?) in a way that will not make you look amateurish.
Here are some of the options you should consider
3D effects emphasize slices “closer” to the observer, distorting the message. A 3D chart will also take up much more space than a 2D version. “Exploding” a slice is meant to emphasize it, but it also makes comparisons harder. Also, people often explode them all, so nothing is emphasized, and only a chart harder to read remains. If you want to look smart, avoiding these two effects is the least you can do.
Assume that, in any chart, a legend is a necessary evil. This means that it should be avoided at all costs. Pie charts rarely if ever need them, because you can label the slices directly.
Since you don’t need color to identify each slice (because you’re labeling them directly) you can use color to emphasize or group. Color is too important to be spent in menial tasks like identifying slices.
It’s not easy to compare slices in a pie chart, but some “anchor points” help. That’s why the core of your message should refer to quarters (one, two or three), thirds (one, two) or halves.
Most charts have two modes: labeled and context. “Labeled” means that each category has a visible key to identify it, using a legend, direct labeling any other option. “Context” means that you show dozens or hundreds of categories without identifying them, except for some kind of tooltip. The purpose of context charts is to capture the overall patterns, specially trends over time in a line chart.
Obviously, you can have an in-between option. That’s when you keep all categories but only identify the more meaningful ones. Combine this with the suggestions above and you’ll see that the common advice to restrict the number of slices loses much of its relevance.
Even if you accept that you should user a bar chart instead of a pie chart, there are unique strengths to the pie chart that a bar chart can’t replicate, namely the display of the whole, and the combined value of adjacent slices.
There is no right sorting key, but alphabetic sorting is usually the wrong one. By default, you should sort the data in descending order, and leave an “other” slice as the last one, but above all the way you sort the slices must support the message.
It’s fair criticism to say of of pies and other simple charts like doughnuts or gauges that they can’t communicate complex ideas. However, it’s unfair to dismiss them because of that. There’s always a place to communicate simple ideas with simple charts. Better yet, they can be used as an entry point (what I call the “gateway chart”). Used this way, a pie chart is just the beginning of larger and more detailed message.
For instance, you can say that two thirds of the population in Europe lives in just a few countries. This is a simple reference that can be easily digested and entices curiosity, almost like click-bait: which countries? Where?
I rarely (never?) use pie charts as standalone charts. Not because I don’t want to, but because I end up finding more interesting alternatives. However, if the message requires several steps starting with a simple fact can be useful as a reference line to which detailed data can be compared.