We've already talked about How to Create Donut Charts, but should you really be using them? When is the right time to use them? Are they really different from pie charts? The short answer is "yes and yes, when it provides the best insight".
The reason might surprise you. It's not that donuts inherently communicate information faster and more clearly, but they force an analyst to be more simplistic with what the chart communicates. Historically, analysts would put an endless number of slices into a pie chart because they could. If you have multiple parts, why not show dozens of them?! Donuts make you rethink the value of those tiny slices, which results in a better visual.
I think that donut charts were a backlash in the rebellion against pie charts, although I have no historical proof of this. Pie charts are beloved by those that believe that they are the best and only way to show parts of a whole. "The Experts" (Edward Tufte and Stephen Few) came back to say that, in fact, using wedges as a visual measurement is a poor way to communicate information because we struggle to measure one angle versus another and the size of one pie slice versus another (Business Insider - The Worst Chart in the World).
I can see their point...
But just because pie charts are bad at showing some information, doesn't mean they're bad at showing all information. That's where the donut comes in.
Should you really be using them? When is the right time to use them?
Yes, when used to show just two or three slices or a measurement of distance to goal, they are a nice, quick way to show a high level indicator.
Are they really different from pie charts?
Yes, the main reason that donut charts are different from pie charts is the absence of information in the middle of the donut. This doesn't provide additional information in and of itself, but it does do two important things:
It provides a space to show the total sum of the pie or to show the actual percentage being measured (XX% of 100%).
It limits you're ability to see the angles of the pie and therefore forces the designer to use very few parts in the donut.
This second point is the most important to focus on, because it's one of the main reasons donuts work. There is evolving research that implies that while a multi pieced pie is unclear, a two pieced pie can be very clear. This study discusses the importance of area and arc length when comparing two pieces of a pie. It's not so much that a donut is an improvement on the pie chart, but that a donut forces an analyst to rethink how the end user will perceive the slices. A donut chart with 50 slices is just as bad as a pie with 50 slices.
Don't be afraid to use donuts. Just make sure that there aren't too many slices. No one wants a small piece of donut!