Pie Chart Maker
Two columns — a label and a value — become a donut chart. Pie charts attract more criticism than any other format, much of it deserved, and there is a narrow set of cases where they are genuinely the best choice. This page covers both.
How to use it
- Paste two columns: Label and Value, with a header row.
- Keep the slice count low; combine small values into an Other row.
- Order rows largest to smallest before pasting.
- Pick a theme and export a high-resolution PNG.
What pie charts are actually good at
A pie chart communicates one thing better than a bar chart does: that the parts constitute a whole. The circle is a closed shape, and readers understand immediately that the slices sum to everything. A bar chart of the same data does not carry that implication without a label saying so.
It is also effective for a small number of slices at recognisable fractions. A slice at half, a quarter, or three quarters is read accurately and instantly, because those angles are familiar. Two or three slices with clearly different sizes is a case where a pie chart is fine and arguably better than the alternative.
The donut variant used here has a practical advantage over a full pie: the hollow centre gives space for a total or a headline figure, and it shifts the reader from judging area to judging arc length, which is marginally more accurate.
The angle-judgement problem
The criticism is grounded in perception research. People compare positions along a common scale accurately, lengths reasonably well, and angles and areas poorly. A pie chart relies on exactly the encodings people are worst at.
The failure shows up with similar slices. Two slices at 23 and 26 percent are effectively indistinguishable, whereas two bars at those values are obviously different. It gets worse when the slices are not adjacent, since comparing two angles at opposite sides of the circle requires holding one in memory while looking at the other.
Slice orientation matters too. A slice starting at the top is read more accurately than the same slice further round, because the vertical reference makes the angle easier to judge. This is why the convention is to start the largest slice at twelve o clock and proceed clockwise in descending order.
Value labels do fix the comparison problem, and at that point the chart is a decorated table. That is not automatically wrong for a slide, but it is worth being honest that the graphic is not doing the work.
How many slices, and what to do with the tail
Five or six slices is the practical limit. Beyond that the small slices become too thin to see, their labels collide, and the palette runs out of distinguishable colours.
Long-tailed data is the usual cause. Twenty products where the top four are 80 percent of revenue produces sixteen slivers that convey nothing individually. Combining them into a single Other slice is not hiding information, it is an accurate summary of a category that is individually negligible — and if the tail matters, it deserves its own chart rather than sixteen slivers on this one.
A related trap is data that does not sum to a meaningful whole. If the values are counts from overlapping sets, or metrics of different kinds, the slices sum to a number that means nothing and the circle asserts a completeness the data does not have. Percentages summing to more than 100 are the clearest symptom.
When to use a bar chart instead
Use bars when there are more than about five categories, when the values are similar enough that ranking them matters, or when precise comparison is the point of the chart.
Use bars when comparing across multiple wholes. Two pie charts side by side are very hard to compare, because the reader must judge angles in one, hold them in memory, and judge angles in the other. A grouped or stacked bar chart answers the same question directly.
Use a stacked bar when composition over time is the subject. A sequence of pie charts across quarters is one of the least readable things in business reporting; a single stacked bar chart with time on the X axis shows the same data and the trend as well.
The remaining case for a pie chart is a small number of parts of a single whole where the message is composition rather than comparison. That case is real, and it is narrower than the frequency of pie charts in the wild suggests.
At a glance
| Input format | Two columns: Label and Value |
|---|---|
| Style | Donut, hollow centre |
| Recommended slices | Five or six maximum |
| Requirement | Values must sum to a meaningful whole |
| Export | PNG at 2× display resolution |
Frequently asked questions
What columns does it need?
Two: a label and a value, with a header row and at least two rows.
Why are pie charts criticised?
Because people judge angles and areas far less accurately than lengths. Slices at 23 and 26 percent look identical; bars at those values do not.
How many slices should I use?
Five or six at most. Combine the long tail into a single Other slice rather than showing sixteen slivers.
Can I compare two pie charts?
Poorly. Judging angles in one and holding them in memory while judging another is unreliable. Use a grouped bar chart.
When is a pie chart the right choice?
A small number of parts of a single whole, where the message is composition rather than precise comparison.
Read more
Comparison and composition — People judge length against a common baseline far better than they judge angle or area. Most charting advice follows from that one fact.