"Never Use Pie Charts"? — Four Common Critiques and the Cases Where Pie Still Wins

"Never use pie charts" is something close to received wisdom in data-visualization circles. Edward Tufte, Stephen Few, and many other leading voices have pointed out the chart's weaknesses. In practice, "when in doubt, use a bar chart" is common advice.

Yet look at real-world reports and presentations and pie charts continue to appear everywhere. Are they all simply wrong? Or are the critics missing something? Both, I'd argue. The critique is well grounded; and at the same time, there are messages that only a pie chart can deliver.

Four critiques of pie charts

First, let's lay out the criticisms. Using pie charts without understanding these would be reckless.

1. Humans struggle to compare angles accurately. Visual perception studies suggest we're good at comparing "lengths" and "positions" but considerably less precise with "angles." Subtle size differences between slices are harder to discern than subtle differences between bar lengths. Pie charts encode quantity as angle and run straight into this limitation.

2. Readability collapses sharply as elements increase. Two or three categories? Fine. Four is still okay. But beyond five, judging the size relationship between adjacent slices becomes hard at a glance. As a rough rule, once you exceed five slices, "which is bigger, second or third?" becomes surprisingly hard to answer from a pie chart.

3. Poor for time-series comparison. When you want to compare 2017 share with 2024 share, you end up with two pies side by side. The reader has to do the diff in their head. A 100% stacked bar chart conveys the change orders of magnitude better.

4. Terrible compatibility with 3D and stylization. A 3D pie chart distorts comparisons by making the front slices look bigger. Design fights data. 3D pie charts are about as close to a "never" as visualization gets.

Lined up like this, the urge to avoid pie charts is understandable. So why does the pie still get picked? Because it has strengths nothing else replaces.

Four cases where the pie still wins

The critiques are valid, but the pie has unique strengths.

Case 1: When you want to convey "majority or not" instantly. Whether some share crosses 50% is something a pie chart shows immediately. The "half" benchmark is built into the shape itself, so "more than half / less than half" reads before the brain even kicks in. A bar chart would need an extra 50% line drawn in.

Case 2: When you want the structure of "a part of the whole." When the message is "what percentage of the total market do we hold," a pie chart visually delivers the concept of "the whole." Bar charts invite comparison; they don't easily convey "wholeness." A pie's closed circle sends the strong message: "this is everything."

Case 3: When elements can be limited to two or three. "PC vs. mobile," "success vs. failure," "subscribed vs. not" — binary structures, or three-way splits like "for / against / neutral" — these are where the pie chart shines. Fewer elements mean larger angular differences and lower cognitive load.

Case 4: When the goal is "instant impression." Social posts, internal announcements, press release thumbnails — anywhere readers spend three seconds — the pie's "intuitive" character is an asset. If the goal isn't precise comparison but leaving an impression, the pie is a strong weapon.

The doughnut alternative

Want a pie's intuitive feel but also a total to display? Doughnut charts step in here. The empty center accommodates totals or KPIs and has become widespread in dashboards and executive reports.

Doughnut charts don't fix every weakness of the pie. Too many slices remain unreadable; angle comparison stays difficult. But the room for a center label gives doughnuts a unique strength: showing "the structure of the whole + the size of the whole itself" in a single figure.

Before debating pies, ask whether what you want to convey is a "comparison" or a "composition."

A practical decision flow for pies

To make this usable, here's a flow. When you're unsure whether to use a pie chart, walk through these questions in order.

First, is the goal "comparison" or "composition"? If comparison, a bar chart is the first candidate. If composition (a part of a whole), a pie chart enters the running.

Second, are there three or fewer elements? Three or fewer is where the pie shines. Four to five calls for caution. Six or more — a bar chart almost always communicates better.

Third, do you need to compare across time periods? If yes, give up on pies and switch to a stacked bar chart (especially 100% stacked).

Fourth, are there plans to apply 3D effects or heavy decoration? If yes, drop the pie. Distorting data with decoration is far less honest than a clean bar chart.

Between "shouldn't use" and "must not use"

Best practices in data visualization often arrive with absolutes. "Don't use pie charts." "Always start the Y-axis at zero." "Place legends above, not to the right." These rules are useful for learners, but in practice you'll always encounter situations where the absolute fails to communicate.

Pie charts are, indeed, easy to overuse. The habit of reaching for a pie without thinking is worth changing. Yet a pie chart used in the right place delivers its message faster and more forcefully than a bar chart ever could.

What matters is asking "should I use a pie chart here?" from scratch each time. Knowing the rules, then choosing based on the data and reader in front of you. That's the judgment of a real practitioner.