How to choose the right chart type for your data

Most bad charts are not badly drawn — they are the wrong chart. A trend forced into a pie, a ranking crammed into a line, a five-way comparison squeezed into a single gauge. The good news: choosing well is mostly about answering one question before you touch any tool. What is the sentence you want the reader to walk away with?

Start from the sentence, not the data

Every strong chart proves one sentence. "Sales grew every quarter" is a trend, so it wants a line or area chart. "Marketing eats a third of the budget" is a share of a whole: pie, donut or treemap. "The Madrid office outsells every other region" is a comparison: bars. "Most visitors drop off at checkout" is a funnel. Write the sentence first and the chart type usually picks itself.

Comparing categories: bars, almost always

Vertical bars are the default for comparing quantities across a handful of categories. Switch to horizontal bars when the labels are long or when you are ranking more than seven or eight items — the labels stay readable and the ranking reads top-down like a leaderboard. Radar charts only earn their place when you compare several metrics across two or three subjects, like product profiles or skill assessments.

Change over time: lines, areas and combos

A line chart shows the shape of change; use one line per series and stop before five. Fill it into an area chart when the volume itself matters, and stack areas when you need parts of a total moving together. A combo chart — bars plus a line — is the classic answer when two units share one story, like revenue in euros against margin in percent.

Parts of a whole: pie, donut or treemap

Pies and donuts are fine at what they actually do: showing that one or two slices dominate a total. Past six slices, switch to a treemap, which keeps every rectangle comparable and labels legible even with twenty categories. If your parts also change hands — money moving from budgets to departments to projects — that is no longer a share picture but a flow, and it wants a sankey.

Flows and processes: sankey, waterfall, funnel, pyramid

Sankey diagrams trace where quantities come from and where they end up. Waterfalls walk a starting number to a final one through gains and losses — the P&L bridge. Funnels show survivors of a sequence of stages, and pyramids show layered structures. Scatter plots, finally, are the tool when the question is about the relationship between two variables rather than any single one.

• Comparison across categories → bar (vertical) or horizontal bar (long labels, rankings).
• Change over time → line, area, stacked area or combo.
• Share of a whole → pie or donut (≤6 slices), treemap beyond that.
• Flow between stages → sankey; running total → waterfall; stage drop-off → funnel.
• Relationship between two variables → scatter.

Still torn between two types? Build both — in MakeCharts you can switch the chart type from the toolbar without retyping your data, so trying the alternative costs about ten seconds.