Donut graphs are so pretty…pretty awful. Oh, they can look really good but they are nigh on impossible to read. So here I rant about the plague of donut graphs marching across our corporate reporting landscape.
In a previous post I outline four ways to tell if your pie chart is useless. Given that donut graphs are pie charts with the centre gnawed away it is worth revisiting those four clues. The first dead give away that your graph is useless is that it has a hole in the middle. We 'read' the proportion that the segment represents by looking at the angle at the middle – no middle, no angle, no understanding.
The next way to tell if the pie chart is useless is that the first segment doesn't start at 12 o'clock. We read pie charts the same way we read analog clock faces, so have the first segment start vertically at 12.
The next two tests of uselessness are that it has more than three segments and needs to have numbers added to the segments. I want to add a fifth marker of pie chart madness: the segments are all the same or very similar colours - I call this corporate branding gone bonkers.
Here are two examples that demonstrate all of these characteristics:
Origin Energy's Sustainability Report 2014. Hole in the middle, not starting at 12 o'clock, more than 3 segments, needs numbers and is rendered in the corporate colours resulting in a mysterious emphasis on generators and confusion between networks and government - are they the same colour? Hard to tell. The purpose of making a diagram is to illustrate an important aspect of the story, not to render a table into a diagram.
Data needs context to turn it into information. The audience filters that information through their own experiences so that it becomes knowledge. This last step is where the data is integrated into the reader's knowledge bank - see our information design process diagram below.
Use diagrams to illustrate an aspect of the story, not to render a tableTweet this
Data without context isn't information, it is still data and has no meaning. So, if you present a table of data - whether rendered as a (gasp) donut graph or left as a table - and you leave out when the data was collected, whether it represents a trend, is an anomaly or is utterly normal, then you are arming your audience with incomplete information and inviting misinterpretation.
Looking at the Origin Energy graph above, it is not clear when the data was collected. It is not clear whether it is typical or a blip in the data. It isn't clear whether it is a good allocation of resources or one that needs addressing fast. The red section for generators suggests that this is a bad proportion because red usually means alert / caution / stop / bad etc. But it could just be corporate branding gone mad.
Data without content isn't information, it is still data and has no meaningTweet this
BP's Sustainability Review 2013. It took me a while to work out that the "4" at the top of this graph is indicating segment number 4 which is so small as to be almost invisible. The white lines dividing the segments from each other make segment 4 even smaller so design over substance, once again, renders data less readable than a table. This donut graph has most of the same problems as the first example with addition of white lines dividing segments and an odd black ring around the outside.
Do I have a solution? Get better help. But even good help will struggle with not having data with context, with marketing departments that insist on using the branding colours regardless of outcome and with the desire for something to look good first and foremost.
My solution to bad diagrams: get better help!Tweet this
I could be reading this whole situation wrong though. Donut graphs are a great way to obfuscate. If you really don't want anyone to know what is actually going on then bad graphs are a good way to do it!
More on facing the truth...
- More tips on staying out of trouble with infographics
- Pie chart uselessness checklist