Bar Chart vs Line Chart vs Pie Chart: When to Use Each

Bars compare, lines show change, pies show one whole. Here’s a quick test for choosing, with a worked example.

Some links on this page are affiliate links. If you sign up through them we may earn a commission, at no extra cost to you. It never changes what we recommend. How we review tools

You have the numbers, the meeting is tomorrow, and the chart menu offers you a bar, a line and a pie. You pick the one that looks nicest, or the one you used last time. Then someone in the room squints at it and asks, “So what am I looking at?”

The choice between these three charts isn’t about taste. Each one answers a different kind of question, and each one has a way of quietly misleading when it’s used for the wrong job. This guide gives you a one-question test for choosing, the evidence behind it, a worked example using the same data three ways, and a checklist to run before you hit send.

The one-question test for choosing a chart

Before you open any tool, finish this sentence: “The point of this chart is to show…” Your answer almost always falls into one of three shapes, and each shape has a natural chart.

If your point is…UseWhy it worksWatch out for
How items compare (“West sells more than South”)Bar chart, sortedBars share a common baseline, so the eye compares lengths and end points easilyCutting the axis above zero; alphabetical order
How something changes over time (“Churn has fallen for four quarters”)Line chartThe slope carries the story; the reader sees direction and turning points at onceJoining categories that have no natural order; too many lines
What share of one total each part takes (“Mobile is half of all traffic”)Pie chart, few slicesIt shows one whole and its parts, which is the only thing it’s built forMore than about five slices; comparing two pies
Decision table: start from the sentence, not the chart menu.

If your sentence doesn’t fit any of these, for example you want to show a spread of values or a link between two measures, none of the three is right. Our guide on how to choose the right chart covers histograms, scatter plots and the rest.

Why bar charts win most comparisons

There is real research behind the advice to “use a bar”. In 1984 the statisticians William Cleveland and Robert McGill ran experiments asking people to judge the relative size of values shown in different ways, then ranked the encodings by how much error each produced (the study is summarised in Cornell’s graphical perception course notes). In 2010 Jeffrey Heer and Michael Bostock repeated the experiments with crowdsourced participants and found the ranking broadly held: judging position along a common scale was the most accurate, and both angle (tested as a pie chart) and area (tested as bubbles) did worse than position.

That is exactly what a bar chart gives you: values as positions along one shared axis. In practice, three habits make bars work harder for you:

  • Start the axis at zero. A bar’s length is the value. If you cut the axis at 80, a bar of 90 looks twice as tall as a bar of 85. The UK’s Office for National Statistics makes this distinction explicitly: bars compare sizes, so they should start at zero, whereas lines are a different matter (more on that below).
  • Sort the bars. Largest to smallest, unless the categories have a natural order such as age bands. Alphabetical order makes the reader do the ranking in their head.
  • Go horizontal when labels are long. Product names and regions read comfortably along a horizontal bar; on a vertical column they end up tilted or truncated.

When a line chart tells the story better

A line connects points, and the connection implies that something flows from one point to the next. That makes it the right choice when your x-axis is time (months, quarters, years) and the wrong choice when it’s a list of categories. A line drawn from “Marketing” to “Finance” to “Sales” invents a trend that doesn’t exist.

Lines also behave differently from bars on the axis. Because a line encodes values by position rather than length, it doesn’t have to start at zero. In its post on the question, the ONS puts it plainly: “It’s ok to start a line chart at a non-zero value”. Starting at zero can flatten real movement, such as a few percentage points of GDP growth, into a nearly straight line. The ONS also warns that zooming in exaggerates small changes, so treat the axis range as an editorial decision and say what the range is.

A few rules keep line charts readable in a meeting:

  • Keep it to three or four lines. Beyond that, highlight the one that matters and grey out the rest.
  • Label lines directly at their end instead of using a legend the reader has to decode.
  • If you only have two or three time periods, a simple column chart is often clearer than a line with almost no shape.

For design detail, see our guide to creating effective line graphs.

The narrow job a pie chart does well

Pie charts get a bad reputation, but the problem is usually the job they’re given rather than the chart itself. Lisa Charlotte Muth of Datawrapper sums up the constraint: “One pie chart can only show one total and its shares.” Her guidance, paraphrased:

  • Keep to about five slices at most. Beyond that, a stacked bar labels more cleanly.
  • Pies work when the shares are close to familiar fractions such as a quarter, a half or three quarters, because the eye recognises those angles.
  • Don’t use two pies to compare two groups or two years. Each pie is its own total, and the reader can’t line up the slices. Use a stacked bar instead.
  • If you only have two values, you may not need a chart at all. “62% of customers renewed” works as a sentence.

The perception research points the same way: angles are harder to judge precisely than positions. So if your audience needs to tell 31% from 35%, a pie is the wrong tool. If they only need to see that one part is “about half”, a pie is fine.

A worked example: one dataset, three charts

Here is a made-up dataset to show how the same numbers tell different stories depending on the chart. Imagine quarterly revenue, in thousands of pounds, for three sales regions. These figures are illustrative, not from a real company.

RegionQ1Q2Q3Q4
North420410400390
South300320330340
West180230290360
Illustrative data for the worked example.
  1. Pie of Q4. Three slices of roughly 36%, 31% and 33%. The chart says “the regions are about equal”, and the reader can’t easily tell which slice is bigger. True, but not useful.
  2. Sorted bar of Q4. North, West, South, with visible gaps. Better: the reader can now rank the regions. But it’s still one moment in time.
  3. Line chart of Q1 to Q4. Now the story jumps out. West has doubled, from 180 to 360, while North has slipped every quarter. West is about to overtake North.

The data didn’t change. The question did. If the meeting is about where to invest next year, the line chart carries the message. Give it a title that states the finding, such as “West has doubled since Q1 and is closing on North”, rather than “Revenue by region”, and highlight West in your accent colour while the other lines stay grey.

A 30-second checklist before you send the chart

  1. Can you say the chart’s point in one sentence? If not, you’re still exploring, not presenting.
  2. Does the chart match the sentence: comparison → bar, change over time → line, share of one whole → pie?
  3. If it’s a bar chart, does the axis start at zero, and are the bars sorted?
  4. If it’s a line chart, is the x-axis really time, and have you flagged a non-zero axis?
  5. If it’s a pie, is there one total, five slices or fewer, and no second pie to compare?
  6. Does the title state the finding, not just the topic?
  7. Is the one thing you want people to see highlighted, with everything else muted?

Rule of thumb

Bars compare, lines move, pies divide one whole. If you can’t say which of the three your point is, you don’t need a different chart yet. You need a clearer sentence.

Your next step

Take the last chart you presented and run it through the checklist above. If it fails a step, rebuild it with the fix and compare the two side by side; the difference is usually obvious within seconds.

You don’t need special software for any of this: Excel, Google Sheets and PowerPoint can all draw a clean bar, line or pie if you switch off the defaults you don’t need. If you’d like charts that come out tidy by default, with sorted bars, direct labels and sensible axes, Datawrapper is built for exactly this kind of publication-ready chart. Try Datawrapper Our Datawrapper vs Flourish comparison and the tools page will help you decide whether it fits your work. And if your chart is heading for a leadership meeting, pair it with our one-slide executive summary.


Sources: Heer and Bostock, “Crowdsourcing Graphical Perception” (CHI 2010) · Cornell INFO 3312 notes on graphical perception, citing Cleveland and McGill (1984) · ONS Digital, “Does the axis have to start at zero? (Part 1 – line charts)” · Datawrapper, “What to consider when creating pie charts”.

Share

LinkedInX

One data story every morning

Five minutes. Free. Unsubscribe anytime.