Life of Pie
This is not a story about Richard Parker. This post won’t make you believe in God. In fact, it’s not about that Pi at all.* Nor is about the other kind of pie — Apple, Pumpkin or Shepherd. It’s about showing data, and the venerable pie chart.
Naomi Robbins issued a challenge on her blog Effective Graphs. Can this chart be improved upon?
I assume what’s behind this question is a general distaste among the Data Visualization community (yes, there is one) for Pie Charts. Their inventor, William Playfair has a lot to answer for. Tufte doesn’t like them, saying, “A table is nearly always better than a dumb pie chart.” Stephen Few thinks you should, “save the pies for dessert.” Naomi has an angular argument against pies, “We make angle judgments when we read a pie chart, but we don’t judge angles very well. These judgments are biased.”
Jorge Camoes has a different, more schizophrenic relationship with pie charts, “if you love pie charts and use them all the time you have a serious problem and you should take a data visualization crash course immediately.” But he does feature them a lot in his blog, and has some great examples of how to use them. Jon Peltier doesn’t like them, and lists a set of alternatives and how-to’s to avoid going down a pie route. Cole Nussbaumer is a hater too. Alberto Cairo is more phlegmatic. He doesn’t hate them, he thinks that circular graphics have their place.
Back to this chart. What’s wrong with it? Jess3 generally does great work, and I really don’t think this is that bad. Yes it uses pie charts**, which have perceptual problems, but there are only 3 slices to each pie, and it’s pretty readable.
The biggest problem, for me, is that it shows both data and insight, in that order. Insight is buried here, and that’s the biggest fault. Insight, (the point of what you are trying to say) should never be buried, it must always be highlighted. If you are showing data and insight, insight comes first.
I’m not a pie chart hater. Like everything in life, it’s about moderation. Sometimes there is a place for circular graphics. But, I do like square pie charts. According to Robert Kosara, “the square pie (or waffle) chart strikes a good balance between engaging the reader and not distorting the data.”
The last, most minor change is removing the legend and putting it directly next to the chart. This removes the need to glance back and forth to figure out what the chart is saying. So here it is, my version:
Eloqua Chart Makeover. @powerfulpoint.
*If you don’t understand the reference, you’ll have to see the movie, or read the book. It’s a great movie, and (I’m told) a great book.
**Strictly speaking, Donut Charts.
Related articles
- I give up, I am embracing pie charts
(simplystatistics.org) - Who Made That Pie Chart?
(The NY Times) - Excel Charts: Pie Charts
(Excel Charts) - Our Irresistible Fascination With All Things Circular
(Stephen Few) - Engaging Readers With Square Pie/ Waffle Charts
(Eager Eyes) - A Fun Pie Chart About Pie Charts
(Alberto Cairo) - Death to Pie Charts
(Cole Nussbaumer)
…
Gavin is a founding partner at fassforward consulting group. He blogs about PowerPoint, Presenting, Communication and Message Discipline at makeapowerfulpoint.com. You can follow him on twitter @powerfulpoint.
make a powerful point 
Reblogged this on Pros Write and commented:
I recommend Gavin’s post about pie charts — especially for his “makeover”!
Thanks! Much appreciated!
Reblogged this on Riz on it! and commented:
Bisa jadi bahan referensi untuk presentasi.
This is really help me to make my presentation better. I could consider to use a pie chart again. 😀
Hi Gavin,
I’m afraid I like neither the original, nor the remake. The biggest problem is that the headline is wrong: the LinkedIn referrals are not 16 times higher for B2B companies. The measures are percentages, not absolute numbers. The fact that LinkedIn share is higher for B2B companies may only mean that the other referrals have gone down. Without the absolute data, we cannot draw the same conclusion as the authors. For now, it seems that we can only say “LinkedIn referrals are 16 times more likely for B2B companies.” The message is still the same: move more resources to LinkedIn, but it also manages expectations better: referrals may not go up. Either I’m reading this completely wrong or eloqua, the creative agency, needs to replace one graphic designer with a statistician.
The second biggest problem is that the graph does not compare what the headline compares. The graphs are meant to compare parts of a whole, the headline compares shares across two data sets. In the original graph, one has to compare two slides that are far apart and, in the remake, two squares equally far apart. The lines joining the two do not help because they look like a bracket, rather suggesting that everything in between is part of a whole, which is how I originally read the lines. If one wants to compare LinkedIn vs LinkedIn, put them together.
The color scheme is harder to read in the remake because the darker blue is buried in the square. The shades of blue are too similar and clash with the red background.
What I like though is your focus in the main message. And yes, Life of Pi is a great book. Roger Ebert only as praise for the movie.
Let me see if I can turn all this talk into a chart.
Francis,
Your point about the percentages is a good one, I’ll have to think about that. I’m interested to see what you come up with. I hope it’s not a bar chart.
Gavin
Hi Gavin, I’m afraid I did! http://infodez.wordpress.com/2012/12/02/show-what-you-say/
Francis,
Quick work on the post! I have thought about what you said. I have 3 counter arguments, on the dataset, the purpose of the graphic and bar charts.
THE DATASET
Specifically, “The fact that LinkedIn share is higher for B2B companies may only mean that the other referrals have gone down.” Correct. But you make an assumption – that Eloqua and Jess3 used two different sized datasets in their work. While I don’t know the method or the raw data, but it’s more likely that they looked at equal size data sets. For the sake of argument, a dataset size of 100,000 clicks(or whatever unit of measure they use) of which LinkedIn gets 15,987 clicks in the B2B set, and 987 in the B2C set, makes the percentages and the insight correct.
THE PURPOSE OF THE GRAPHIC
Your point though, indicates a bigger issue, which always seems to be at the heart of these kind of online debates, about infographics, charts etc. The purpose of this chart is business. That is, Eloqua employed Jess3 to help them establish credibility and authority for their brand. They have to attract eyeballs to do this (The original piece – 40 Charts for Modern Marketers) helps Eloqua, which sells marketing automation and demand generation tools, to show (deliberate word choice – not prove) that they have credibility in this space. You could argue that showing is not proving, (and I agree) but they don’t need to reach the higher bar. That’s for scientists, lawyers and journalists.
The purpose is business, not art, or science, or journalism. That’s why design as an approach (which draws from all those other disciplines) works in this case. Design is not Art. That’s about pleasing the Artist. Design has a client. It’s about fitness for purpose, and the initial arbiter of that is the client.
PIE CHARTS
People’s view on Pie Charts is rooted in their discipline. If you take a scientific approach, or a journalistic one, there are a lot of problems with Pies since they shroud the truth. The usual counter to this is the bar chart, which has no real perceptual or accuracy problems in showing data. But the problem with bar charts is their ubiquity. They are common and (dare I say it) boring! Everything you do to make them accurate makes them less interesting. Everything you do to make them interesting makes them less accurate. Going back to the purpose, interesting plays a big role, since the general order of things is 1) attract eyeballs, and 2)demonstrate credibility and authority.