Pie Charts vs Bar Charts: Why Ranking Slices Is Harder Than Comparing Lengths

TLDR

For the pie chart vs bar chart decision, use a pie only when the categories are mutually exclusive parts of one meaningful total and the main message is composition. Keep it to about five or fewer clearly different slices. Use bars when readers need to rank categories, compare close values, handle numerous categories, or read values against a common scale. If you are uncertain, a bar chart is usually the safer default.

The difference is not that pie charts are inherently bad. It is that they ask readers to compare angles and areas, while bar charts let readers compare lengths from a shared baseline. That distinction matters most when values are close. A pie may quickly show that one category dominates a total, but bars make it easier to determine which of two similar categories is larger and by how much.

Pie chart vs bar chart: the quick decision table

Question	Choose a pie chart when…	Choose a bar chart when…
Do the categories form a whole?	They are mutually exclusive components of one meaningful total.	They do not form a total, or the total is not the point.
How many categories are there?	There are about five or fewer.	There are many categories or long category names.
How different are the values?	The shares differ enough to distinguish easily.	Several values are close and require comparison.
Does rank matter?	Only the broad composition matters.	Readers need to identify first, second, third, or other ranks.
Does order matter?	Slices can be arranged by size without losing meaning.	Categories have a natural sequence, or sorting will clarify rank.
Are exact values important?	Direct labels can provide the needed values.	A common scale helps readers estimate and compare values.
Are multiple groups being compared?	Usually not; one composition is shown.	Yes; grouped or aligned bars support cross-group comparison.

This table is a decision framework rather than a mechanical rule. Chart effectiveness depends on the reader’s task and the structure of the data. A task-based study of basic visualizations found no universally best chart for every combination of task and data type, although bars were faster and preferred for ordering tasks. Read the visualization study and its task-based results.

What a pie chart communicates well

A pie chart answers a narrow question: how is one total divided among its components? Each slice represents a share of the circle, so the categories should collectively describe the same whole. Examples include the allocation of a fixed budget, the share of votes among mutually exclusive choices, or the distribution of units sold across product types.

The denominator is essential. A chart showing 40%, 35%, and 25% works as a composition only if those percentages refer to the same population, period, and measure. Before charting market shares, for example, establish whether the denominator is revenue, units, customers, or another measure. Revenue share and unit share can produce different market rankings, even when both are valid percentages.

A pie is strongest when the intended conclusion is visually broad: one segment is a majority, two categories account for most of the total, or one component is much smaller than the rest. It becomes weaker when the reader must compare several nearly equal shares. UK Government Analysis Function guidance recommends pies for parts of a whole, advises using no more than five categories, and cautions against using them when categories are similarly sized. The five-category figure is a practical design guideline, not a scientific boundary at which a sixth slice automatically makes a chart unreadable.

Why bar charts are better for ranking and close values

Bars encode values as lengths aligned to a common baseline. Readers can scan the endpoints to see which categories are longer, identify rank, and judge relatively small differences. Pie slices do not provide the same aligned reference: readers must compare angles, arc lengths, or areas around different positions in a circle.

Foundational graphical-perception research by Cleveland and McGill distinguished among visual encodings such as position, length, and angle. That framework helps explain why the encoding should match the task: comparing aligned positions or lengths is different from estimating angles and areas. Review the original graphical-perception research.

More recent evidence points in the same practical direction without establishing a universal cutoff. In one controlled 2025 study condition involving the ranking of close values, participants were less accurate and took longer with pie and donut charts than with bar charts. That finding supports bars for close-value ranking, but it does not tell us that a particular percentage-point gap always makes a pie unsuitable.

Consider categories with shares of 24%, 23%, 21%, 18%, and 14%. A pie may communicate that the total is fairly dispersed, but readers will have to work to rank the middle slices. Bars preserve the same values while making the order visible through aligned endpoints. The figures in this example are illustrative; the principle is to choose bars whenever the differences themselves are the message.

Five questions to ask before choosing

1. Do the categories form one meaningful whole?

Use a pie only if every slice uses the same denominator and the categories collectively represent a meaningful total. Mutually exclusive survey responses to one question may qualify. Percentages calculated from separate questions usually do not, especially when respondents can select more than one answer.

The presence of percentage signs does not make data suitable for a pie. A satisfaction rate, conversion rate, unemployment rate, and market share might all be percentages, but they have different denominators and do not combine into one circle.

2. How many categories are there?

As slices multiply, colors become harder to distinguish, labels compete for space, and small categories turn into thin wedges. About five or fewer is a useful working rule from UK government guidance. If a sixth category is important and distinct, the chart does not automatically fail, but bars will often accommodate it more cleanly.

Horizontal bars are particularly useful for numerous categories or long names. They give labels room and allow values to be compared without forcing the reader to match colors to a distant legend.

3. Are the values close together?

The closer the values, the stronger the case for bars. There is no evidence-backed universal difference—such as two, five, or ten percentage points—at which a pie becomes unacceptable. Ask instead whether readers can see the distinction without repeatedly consulting labels. If exact labels do all the comparative work, the circular encoding is adding little.

4. Is ranking or sequence important?

Sort categorical bars from largest to smallest when the task is ranking. Do not sort blindly when categories have a meaningful native order. Time periods, age bands, stages in a process, and ordered response scales such as “strongly disagree” through “strongly agree” should normally retain that sequence.

For a pie chart, UK Government Analysis Function guidance recommends ordering slices by size and placing the largest slice at 12 o’clock. Consistent ordering reduces visual searching, although direct labels are still needed when readers must retrieve values.

5. Can every reader access the labels and values?

Do not make color the only way to identify a category. Direct labels can connect names and values to the corresponding slice or bar, reducing trips between the chart and a legend. Accessible text or a data table should also communicate the chart’s information, particularly because a pie can become difficult to interpret when magnified.

Labels should not turn the visualization into a thicket of leader lines. If direct labeling cannot be done cleanly, that is evidence that the chart contains too many categories or that a bar chart or table would serve the reader better.

When pie charts remain acceptable

A pie chart remains a reasonable choice when all or most of these conditions are met:

• The categories are mutually exclusive parts of one meaningful total.
• The chart contains about five or fewer categories.
• The shares are different enough to distinguish without precise angle judgments.
• The intended message is composition, not detailed ranking.
• Slices are displayed in two dimensions and arranged consistently by size.
• Category names and values are labeled directly where practical.
• The same information is available in accessible text or a table.

For example, a four-category budget allocation with one clearly dominant expense can work well as a pie. The circle reinforces the idea of a fixed total, and the broad differences are visible. If the question changes from “Where does the budget go?” to “How many dollars separate each expense?” bars become more useful.

Common situations where bars are the better choice

Use bars instead of a pie when categories do not add to a meaningful whole, when people can belong to multiple categories, or when values are counts rather than components of a fixed total. Bars are also preferable for long lists, close values, explicit rankings, and comparisons across periods or groups.

Multiple pies are an especially weak format for comparison. If each year, region, or customer segment has its own circle, readers must mentally match slices located in different places and possibly represented by different totals. Aligned bars usually make changes and cross-group differences easier to see.

Avoid three-dimensional pies. Perspective can make front-facing slices appear more prominent and turns an already difficult angle-and-area comparison into a comparison of distorted shapes. A flat, two-dimensional display keeps the visual encoding closer to the underlying proportions.

Alternatives that preserve useful context

A table is the best option when exact lookup matters more than pattern recognition. It can also accompany a chart so readers have both a visual summary and precise values.

A 100% stacked bar can preserve the part-to-whole framing, particularly when comparing composition across several groups. It is not an automatic replacement for every pie: only segments touching a common edge have directly aligned baselines, so middle segments can still be difficult to compare. If comparison of every category is the priority, grouped bars, small multiples, or another aligned design may be clearer.

A donut chart creates room for a total or short annotation in the center, but it does not solve the close-value ranking problem. It still asks readers to compare curved segments, and removing the center can reduce the angle cues available in a conventional pie.

A final chart-selection checklist

• Confirm that every value uses the same unit, period, population, and denominator.
• Use a pie only for components of one meaningful whole.
• Prefer bars for ranking, close comparisons, or numerous categories.
• Sort bars by value only when the categories do not have a meaningful native order.
• Keep pies simple, two-dimensional, consistently ordered, and directly labeled.
• Provide accessible text or a data table rather than relying only on color.
• Test whether the chart answers the reader’s actual question without requiring labels to rescue the design.

Quick answers

Is a pie chart or bar chart better for percentages?

It depends on what the percentages represent. Use a pie for mutually exclusive shares of one meaningful total. Use bars for percentages with separate denominators, multiple-response data, ranking, or close comparisons.

Can a pie chart contain values that do not add to 100%?

Normally, the displayed slices should represent the complete whole. Small deviations caused by rounding can be explained, but categories that fundamentally do not constitute one total should not be forced into a pie.

Should every bar chart be sorted?

No. Sort by value when rank is the message. Preserve chronological, geographic, procedural, or ordinal order when that sequence carries meaning.

What is the practical default?

Start with bars. Move to a pie only when the circular part-to-whole framing adds something important and the composition is simple enough to read immediately. The best chart is not the one that looks most distinctive; it is the one that makes the intended comparison easiest.

References

1. va.gatech.edu
2. Data visualisation: charts – Government Analysis Function
3. Accessible charts: a checklist of the basics – Government Analysis Function
4. Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods: Journal of the American Statistical Association: Vol 79, No 387
5. Are pie charts evil? An assessment of the value of pie and donut charts compared to bar charts – Andrew Hill, 2025