The Five Stages of Grief Over the Death of Pie—err—Donut Charts

Help your client get through them as fast as possible.

I published this on the Site That Shall Not Be Named back in 2018. At the time, the infamous pie chart, bane of lucid communication for two centuries, seemed well and truly dead. After decades of (ab)use in corporate boardrooms, and a final sordid rumspringa in the early 2010s courtesy of the "infographic" glut, pies were at long last being deprecated as a data visual. I wrote this article after having had to disabuse some late-majority and laggard types at legacy companies, but I never expected it to be relevant to a "tech" audience for much longer.

However, a funny thing has happened in the last 5 years: the donut chart:

Someone got the brilliant idea that, if you punch a hole in the center of a pie chart, it suddenly becomes a respectable graphic. I have now had to educate a fresh crop of executives on the perils of radial graphs. So I have amended the old article in light of this new development and made it available on LinkedIn. Am I a mensch or what?

Pie charts have had a good run. They have been around since William Playfair created his?Statistical Breviary of 1801?using nothing but a pen, a ruler, and a compass.

But time has not been kind to this popular visualization tool. 200-odd years later, they are now increasingly seen as somewhat of an embarrassment by those in the know. And those in the know are growing in number. The real death knell of pie charts was data visualization high priest, Stephen Few’s gut punch titled “Save the Pies for Dessert”.

In the years following that brilliant piece, pies had their last hurrah as they were shamelessly exploited by every talentless graphic designer cashing in on the infographic fad of the early 2010s.

This only further ruined the already faltering reputation of pie charts. Now they are moribund, and to include one in your interface is a liability to your own reputation. Chances are, if you’re reading this, then you already read Few’s article, and you already know that pies are obsolete. Unfortunately, your client might not.

Pies, it seems, are well loved by many, many people. Some just love the way they look. Others legitimately believe they have practical merits. I won’t lie. I once loved pie charts too. When I learned that they were a bad thing back in 2011, I too was taken aback. I also tried desperately to make excuses for pies, to defend their besmirched honor, but I soon came around to reality.

Yes, the death of pies is like the death of a loved one, because people love them. Therefore, you can expect people to go through the proverbial stages of grief when they learn of their demise. Those people might include your client. What will you do when you hear those dreaded words:

“I’d like to see a pie chart on this screen.”

Allow me to walk you through the stages of grief over the death of pie charts, and how you should respond to each as your client experiences them, to help guide them to the light in the hopes that they too will reject this outdated and clunky technology.

Denial

Old habits die hard. People will have a built in set of excuses to justify their use of pie charts. They really do believe they are a useful tool and to believe otherwise not only would mean they were wrong, but that the world itself has stopped making sense.

“But they are so intuitive!”

The only thing intuitive about a pie chart is the fact that, when you look at it, you know it deals with a part-to-whole relationship. The actual data itself is not intuitive at all. Just consider this:

Which set of graphs made it easier to order the different values? QED.

“But they are so?pretty.”

Choosing bad designs because you think they are pretty is how we ended up with the hockey puck.

“But… our customers expect?them!”

Expecting something and wanting something are two different things. You are doing your customers a disservice by continuing to supply them with the tools of mediocrity. Thanks to unscrupulous software vendors in an arms race to see who can offer the most pimped-out pie charts, you can now make this:

Moreover, if you’re selling enterprise software, the customer doing the wanting might not even be a user. They might be some out-of-touch Karen who gets off on telling their employees what they’re going to use. Or they might be crusty old crank who spent years learning crappy software and sees their expertise as job security. Consider what your customers say they want, but do not take them at face value.

Anger

As their excuses start to feel hollow to them, they will feel betrayed, lied to. After all, somebody told them that pies were good. They were full of shit! Nah, that can’t be it. The real liar is the one trying to tell me that my precious pies are bad. The butthurt may flow. Be ready for it.

“Who does this Stephen Few guy even think he?is??”

Stephen Few knows more than either of us, maybe both of us combined, about data visualization. He invented the bullet graph, one of the most useful data visualization tools of the digital era.

I should also mention that the other big name in data visualization, Edward Tufte, also dislikes pie charts, calling them “dumb”, and saying they should “never be used”. If you don’t know, Tufte invented the sparkline graph, which is another one of the most useful visualizations available. It’s basically a micro line graph.

These two dudes comprise 90% of the data visualization expertise in the world. They both have published definitive works on the subject. Unless you have research that trumps theirs, it’s best to just take their word for it.

“Wait, look at this article by (insert hack) saying that pies are actually good. See??See???”

Are you going to trust some no-name over respected industry leaders? The guy who wrote the article you sent me is the same idiot who made this:

“Put in a pie chart or I’ll find someone who?will.”

Go ahead. And that hack will design their way onto a?Bad UX Roundup.

Bargaining

Clients attached to their pies will do anything to protect them. They will try to wear you down with seemingly reasonable compromises, just so you’ll let them have their pie and eat it too. Don’t give in. The only acceptable solution is no pie charts. Not toned-down pie charts. Zero pie charts. Be firm, and emphasize the parts of the design that they already agree with, so they’ll forget that they are giving up all the ground. Remember, compromise is the root of all evil in design.

“What if we just limited the number of?slices?”

What limit do you want to set? Four? Pies are still worthless with small numbers. See below:

Limit it to just two then? What’s the point?

“Ok, ok, fine! No 3D?pies.”

Nope. Sorry. 3D is not even the main problem with pies, just something that makes their uselessness more obvious. The pie, even in its cleanest, purest form, is still a lousy data visualization trope.

“What about a bar chart and a pie to supplement it?”

You mean like this?

This only wastes pixels by telling you the same information twice. Literally the only thing the pies would be for is telling the user that the information in the bars is part-to-whole, and, unless they are too stupid to be analyzing data, they will know by the percentages at the top. No pies!

"No way, I'd never use a pie chart. Donut charts are so much better!"

And THIS is the curve ball that you, the designer will be thrown more often these days. It's actually increasingly unlikely you'll be asked for a full pie, not so much because the one doing the asking has the slightest clue about human visual perception, but because donuts are everywhere now, with their handy numeric indicator in the center. And ubiquity equals orthodoxy in tech design. Everyone knows that!

If you're really unlucky, you'll have to deal with some of the "logic" deployed to defend donuts. This typically goes something like "people read donuts differently than they do pie charts. They read pie charts by section area, but they read donuts by segment length." This has the *ahem* ring of science because, yes, the big reason that bars and lines are superior to pies is because our visual perception is a lot better at reckoning line length than planar area. So, it would stand to "reason" that, if you make the ring of the donut skinny enough, you can trick the brain into reckoning each segment by its length in radians.

But, again, WHY? This just ends you up with a slightly less horrible pie chart. The viewer's brain still has to do a lot of heavy lifting to mentally straighten out those segments into lines and give them a common root point for comparison. In other words, they do nothing to obviate the radial reckoning issue while only partially (and theoretically) addressing the area reckoning issue. Donut charts are the ultimate form of bargaining, thus making this the hardest step to get your client through.

Depression

Eventually, the client may start conceding that pies (or donuts) have to go, if only because they are afraid of being seen as a dinosaur. That doesn’t mean they won’t resent it, though. They will feel as though something is missing. This is a dangerous time because, if anything should happen to the project, especially you leaving and being replaced by someone else, the client may take the opportunity to put their pies back in. This happened to me once with the other edible UI disaster, the hamburger chart, after I fought so hard to keep it out.

Make sure that you show the client how awesome the pie-less design is. With any luck, they will convince themselves they came up with the idea themselves, Inception style.

“The new design is so sterile and boring. This?sucks.”

As with all interfaces, design only exists to provide a stage for content. With data visualization, the content is the data. If the data is boring, pie charts won’t save it. But if the data is interesting, the only visualization which will make it boring is one which obscures it.

Take a look at this dashboard that won one of Stephen Few’s?competitions. It is intended for a teacher monitoring the performance of their students. I don’t even find the subject matter inherently that interesting, and yet it is presented so clearly that it actually?becomes?sort of interesting.

“It seems like there’s nothing but straight lines and right angles as far as the eye can?see.”

Well that’s just not true. While we are getting rid of unncessary roundness in the charts, we can add back that roundness where it belongs: in the parts of the interface intended to be “touched” by the user. It’s simple evolutionary psychology. People do not want to?touch?sharp objects because they can hurt themselves.

Buttons, in particular, should have round edges because they invite people to touch them, unlike sharp-edged “flat” buttons of yore. In fact any component on the screen that can be manipulated by the user should be rounded for this reason.

That same article also makes a case for rounded lines on flowcharts because they are easier to follow.

There are also bubble charts which display three dimensions of data at once:

So rest easy. There will be round stuff in your interface.

Oh, and it should go without saying that your bar charts should?not?have rounded tips as this harms readability.

Acceptance

Ultimately, all but those with the most severe cases of cranio-rectal herniation will come around to the virtues of bar charts and the vices of pies. Where they once saw an appealing graphic, they will now see a muddled hard-to-read mess. It should be smooth sailing until the next outdated design fad comes along.

“Man, look at this stupid infographic with all the pie charts. Who paid for this?crap?”

Wait till you see the cake chart.

Follow me for more UX and futurist content

If you made it this far, you either like what I had to say, or you're hate-reading. Either situation justifies a follow. So do it. I don't like playing to an empty hall.