Pie charts are probably the most emotionally charged objects in business graphics. Many designers would ban them outright; many clients love them. The McKinsey position is pragmatic: pies and their cousins (doughnuts, Harvey balls, wheels) have a very narrow but legitimate role when used with discipline.
The key is to be honest about what the human eye is good at. We read length and position very accurately; we read angle and area much less accurately. That means pies are poor at fine comparison, but acceptable when the story is about a simple, high-level share picture.
This chapter explains where pies and doughnut-type charts fit in a consulting toolkit, and—equally important—when you should choose a different chart.
7.1 When (and when not) to use pie charts
Use a pie chart only when all of the following are true:
- You are showing one point in time, not a time series.
- The story is explicitly part-to-whole: how a single total splits into a few components.
- You have very few categories—ideally 3–5.
- The message is about big, obvious differences, not subtle ones.
Examples where a pie can be acceptable:
- “Three customer segments account for 80% of revenue.”
- “One category dominates the cost base.”
- “Regulated products are a small share of the portfolio.”
Avoid pies when:
- You need to compare across several entities or time periods (e.g., mix by year or by country). Stacked bars or 100% stacked bars are better.
- You have many small slices or a meaningful long tail. The chart becomes unreadable and underplays the importance of the tail.
- The decision requires precise readings or fine differences in share (e.g., 18% vs 21%). Bars make that clear; pies do not.
- You have more than one pie on a page and ask the reader to compare them. Humans are very bad at comparing separate angles.
A useful rule: if your title or storyline involves words like “trend,” “change,” “compare across,” or “rank,” you almost certainly do not want a pie. If it says “share of” or “portion of total,” and you only need a rough sense of magnitude, a pie might be acceptable—but bars or stacked bars are still usually safer.
7.2 Simple Pie chart (one point in time, few categories)
When you do use a pie chart, keep it brutally simple. A classic pie is a single circle, one time period, a handful of slices.
Good practice for a simple pie:
- Limit yourself to 3–5 meaningful slices. If you have more, consider grouping the rest into “Other” or moving low-importance categories to a table.
- Make sure all slices sum to 100% and label them explicitly with percentages. Do not ask the reader to infer values just from angles.
- Use direct labels on or just outside each slice. Legends that force constant back-and-forth scanning waste time.
- Order slices in a deliberate way:
- Often the largest slice starts at the 12 o’clock or 3 o’clock position and proceeding clockwise by size.
- Keep “Other” last, away from the main story.
- Highlight only one slice if you can—the one that carries your message. Use a stronger color for it and a neutral palette for the rest.
Typical McKinsey-style pie use: a small, single pie tucked into a corner of a slide to show that, for example, 70% of the cost base is in one bucket that your recommendations target. The pie is there to give a visual sense of dominance, not to carry the full analytical argument.
7.3 Doughnut chart and multiple doughnut chart (comparing mixes)
A doughnut chart is simply a pie with a hole in the middle. That hole has two advantages: it reduces ink and gives you a natural place for a central label (e.g., “Total revenue,” “100% of cost base”).
Use a single doughnut instead of a pie when:
- You want to show the same one-time composition, but with a central summary (e.g., total value, period, or a high-level message).
- You want a slightly more modern look while still conveying a simple part-to-whole picture.
A multiple doughnut chart places several rings side by side or concentrically to compare mixes across a small number of entities (e.g., “Client vs Competitor” or “2020 vs 2024”). They are useful only in very constrained situations:
- You are comparing 2–3 compositions at most.
- Each composition has few components (again, 3–5).
- The message is coarse comparison—e.g., “Client is much more dependent on Product A than competitors.”
Design guidelines:
- Keep the same order and color for segments across all doughnuts, so the reader can track components quickly.
- Ensure all doughnuts are the same size and clearly labeled by entity or period.
- Consider pairing with a small table or bar chart that provides absolute totals. Doughnuts show share; they hide level.
Multiple doughnuts can look appealing but are easy to overuse. If the reader needs to study the page to see differences in small slices, you probably should have used 100% stacked bars instead.
7.4 Harvey balls (qualitative part-to-whole indicators)
Harvey balls are small circles that are partially filled to show a qualitative degree of completion or relative strength: empty, quarter, half, three-quarters, full. They are closer to icons than charts, but they communicate a part-to-whole idea efficiently.
Typical uses:
- Capability assessments: each row is a capability, each column is an entity, and the Harvey ball shows maturity or strength (e.g., “low,” “medium,” “high”).
- Feature comparisons: product vs competitor across a list of features.
- Implementation or readiness dashboards: level of progress across workstreams.
Good practices:
- Use a small, fixed set of levels (e.g., 0%, 25%, 50%, 75%, 100%) and define them clearly in a legend (“Not present,” “Basic,” “Good,” “Leading”).
- Keep them small and consistent in size; they work best inline within a matrix or table, not as standalone, large graphics.
- Treat them as qualitative indicators, not precise percentages. They are there to give the eye a quick sense of “more filled vs less filled,” not to represent exact numbers.
Harvey balls are particularly helpful when numbers would imply a false sense of precision (e.g., scoring a capability “63/100”) but you still want a visual sense of relative position.
7.5 Wheel charts (circular category layouts)
Wheel charts place categories around a circle, sometimes with radial segments or colored bands. They are attractive but often misused as de facto pies.
They work best when the circular layout itself adds meaning, for example:
- Mapping capability areas or functions around a central concept (the operating model hub-and-spoke).
- Grouping related topics into thematic clusters around a core (“customer,” “operations,” “technology,” “organization”).
- Showing a framework rather than data—e.g., four quadrants, each with sub-elements.
In other words, wheel charts in consulting are usually conceptual layouts, not primary quantitative charts. Numbers, if present at all, are secondary.
Guidelines if you do use a wheel with data:
- Treat it as a high-level picture, not the main vehicle for precise comparisons.
- Limit the number of categories to avoid a cluttered, crowded circle.
- Use it sparingly, and only when the radial metaphor (spokes from a hub, circular cycle) genuinely reinforces your story.
If you are mainly trying to show “share,” “ranking,” or “trend,” a wheel chart is almost never the right answer. It is a framing device, not a default analytical visual.
7.6 Alternatives to pies: bars, stacked bars, Mekko, and others
In many cases where a pie is proposed, a different chart will communicate the same idea more clearly and robustly. As the design lead, you should gently steer teams toward these alternatives.
Common and stronger substitutes:
- Simple horizontal bar chart with percentages labeled.
- Great for showing one-time composition with exact values.
- Much easier to compare components, especially smaller ones.
- Stacked bar or stacked column chart for part-to-whole across entities or time.
- Shows both total and components together.
- Works where someone might suggest “a set of pies,” but is much easier to compare.
- 100% stacked bar or column chart for mix comparisons.
- Ideal when you care about share, not levels, and need to compare several periods or groups in one view.
- Mekko (Marimekko) charts for size and mix simultaneously.
- Column width = total size (e.g., market size by country).
- Column height segments = share breakdown (e.g., players within each country).
- Useful when part-to-whole is part of a larger portfolio structure story.
- Small-multiple bar charts (one small bar chart per entity).
- Helpful when you want to preserve simple bars but show a few entities side by side, each with the same structure.
When a team proposes a pie, ask:
- Do we need to compare mixes across time or entities?
- If yes → use 100% stacked bars or Mekko.
- Do we need precise values or fine differences?
- If yes → use bars.
- Is this mainly conceptual (framework, topics, capability areas)?
- If yes → maybe a wheel or simple radial layout, but keep numbers elsewhere.
In a mature McKinsey deck, pies tend to appear rarely and only in situations where a quick, single-moment share picture is genuinely all that is needed. For almost everything else involving composition, bars and their relatives will serve you and your clients better.
