A pie chart is a circular visualization that divides a whole into slices, with each slice’s size proportional to its share of the total. Its donut variant is the same chart with a hole in the middle — handy for showing the total (or a headline number) in the center. Pies communicate one specific idea: this is how the whole splits into parts.
When to use a pie chart
Pie charts work when the part-to-whole relationship is the message and the parts are few:
- A dominant share. “Two-thirds of revenue comes from one product line” is instantly legible as a pie — the eye reads “most of the circle” without needing any numbers.
- A handful of categories. Traffic by channel, pipeline by stage, plan mix across three or four tiers. Around five slices is a practical ceiling; beyond that, group the tail into “other.”
- Simple compositions on a dashboard. As a small card answering “what’s the mix?”, a pie or donut carries its weight, especially with percentage labels doing the precise part of the job.
The slices must genuinely sum to a meaningful whole. If the categories overlap, or the values are rates rather than quantities, there is no “whole” to divide, and the pie shape is asserting something the data doesn’t support.
When not to use one
The classic misuse is the many-sliced pie. Humans judge angles poorly — far worse than lengths — so a pie with ten near-equal slices becomes a guessing game with a legend. If readers need to compare the categories rather than grasp the overall split, a bar chart or row chart answers “which is bigger, and by how much?” at a glance; the pie can’t.
Pies are also unsuitable for change over time (use a line chart or area chart) and clumsy for comparing compositions across groups — a row of pies is far harder to read than one stacked bar chart. And avoid 3D or exploded effects anywhere you find them: tilting the pie distorts the angles, which are the only thing the chart had going for it.
A fair summary of the long-running “pie charts considered harmful” debate: pies are fine for a few slices where the rough split is the point, and wrong for everything else.
Pie charts in Metabase
Metabase has a native pie chart type. Build a question that groups a measure by a single dimension — count of orders by channel, say — and choose the pie visualization. In the settings you can show each slice’s percentage, set a minimum slice size so small categories roll up into “other,” pick colors per slice, and add a donut hole, which displays the total in the center.
Pie charts support drill-through, so clicking a slice takes you to the records behind it, and on a dashboard a pie responds to the same filters as the other cards — one click can show the channel mix for a single region or month. See the pie and donut chart docs for the full option list.
Related terms
Further reading
Put it to work
- Marketing analytics — Overview
- Sales pipeline dashboard — Dashboard
- Conversion rate — Metric
