Are pie charts the worst charts you can make? Some experts claim they are. I love pie charts, especially in the donut chart variant (meaning instead of a full circle, there's a circular empty space in the middle). Nevertheless, I sometimes find pie charts in professional presentations or reports that are completely unreadable.
I found one such chart in a new Deloitte report. It looks like this.
What's wrong with it? Several things.
Chart Type Selection
First of all, this data should not be shown through a pie chart (or donut chart). Pie/donut charts work well when we have few components. Ideally – only two. If three – it will still be acceptable, though not recommended. With four – that's already an absolute last resort. Meanwhile, here we have seven elements.
In pie charts, it's extremely difficult to compare the sizes of components. With two components – the comparison will be easy, but each additional element disrupts readability. Therefore, the more there are – the less readable the chart will be.
Category Division
An important element of this chart – and practically imperceptible – is the white division in the middle of the chart dividing it into four categories. Do you notice it? At first glance, it's hard to spot. But look closely – those thin white outlines on the pie chart that mark the boundaries between the categories cobalt, copper, lithium, nickel. This is a key element of the chart because it shows the scale. Meanwhile, it's so indistinct that the entire chart is unreadable – the categories merge into one green mass with two shades.
Category Order
To facilitate reading pie charts, the following rule has been adopted: we start with the largest values at 12 o'clock (assuming the pie chart is a clock face), and then, moving clockwise, we insert subsequent categories descending toward the smallest. Meanwhile, in the presented chart we see inconsistency. We start with the smallest value (cobalt 1%), then there's the largest (copper 62%), then 4%, and at the end 33%. Order? Alphabetical. But alphabetical order in data visualization has little value.
How Should This Chart Look Then?
There are several ways to rework it. I would rework it this way:
So instead of a pie chart – a bar chart, thanks to which it's much easier to compare category sizes. And additionally, the problem of category division disappears – thanks to the grouped chart, categorization is simple and readable. And instead of random order – order from the largest category to the smallest.
For accuracy – I'll clarify that because specific numbers in subcategories are not provided in the original chart, on my chart these values are approximate.
Do Such Errors Occur Often?
In the world of business presentations very often. Nevertheless, I'm surprised that in such a prestigious publication, issued by such a large consulting firm as Deloitte, such an unreadable chart appeared.
If you want to improve your chart-making and data storytelling skills, be sure to check out our data storytelling training. In it, we focus not on talking about numbers in the TED talk convention, but on talking about data in a typically business convention – with deep immersion in the substance, but also lightly, simply, and in a way understandable to recipients.
Piotr Garlej
