When not to use a pie chart
The internet's hostility to pie charts is overdone, but the underlying complaint is real: people compare angles badly. That limitation defines a narrow set of cases where pies work well.
They must be parts of one whole
If your values do not sum to something meaningful, a pie is a category error. Revenue by region sums to total revenue and qualifies. Average response time by region does not sum to anything and does not.
Six slices is the practical ceiling
Beyond six, slices get thin, labels collide and the reader ends up decoding a legend. That is the point at which a treemap or a sorted bar chart does the same job without the effort.
Similar values defeat the format
Four slices at 26, 25, 25 and 24 percent are visually identical. If the ranking is the message and the values are close, bars are the only chart that shows it.
Never compare two pies
Two pies side by side ask the reader to compare angles across charts, which is the hardest visual task on the list. Use a stacked bar or a slope chart instead.
- Good: market share where one player dominates.
- Bad: eleven expense categories.
- Bad: two pies for this year and last year.
- Bad: anything where the numbers do not sum to a whole.
The honest summary: a pie chart is a good way to show that one thing is much bigger than a few others, and a bad way to show almost anything else.
Frequently asked
Are pie charts bad?
No, they are narrow. With six or fewer slices and one dominant share they communicate instantly. Outside those conditions a bar chart or treemap is better.
What should I use instead of a pie chart?
A sorted bar chart when ranking matters, a treemap when there are many categories, and a stacked bar when comparing composition across periods.