How to Choose the Right Chart
GUIDEPick the right chart in one step: start from what you want to show, not the chart you like.
Choosing a chart feels like a design decision, but it is really a logic decision. Every effective chart starts with one question: what relationship in the data do I want the reader to see? Answer that, and the chart type almost picks itself. Most data stories fall into five goals — comparison, composition, trend, distribution, and relationship — and each maps cleanly to a chart family.
Step 1: Name your goal
Before opening any tool, finish this sentence: "I want the reader to see…" The ending tells you the family.
| If you want to show… | Use |
|---|---|
| How separate categories compare | Bar chart |
| How parts make up a whole | Pie or stacked bar |
| How a value changes over time | Line or area chart |
| How values are distributed | Histogram |
| Whether two variables are related | Scatter plot |
Comparison — bar chart
When the point is "which is bigger?" — ranking products, comparing scores, showing votes per option — use a bar chart. The eye reads the length of a bar more accurately than any other encoding, so bars give the clearest, most honest comparison. Use horizontal bars when labels are long or there are many categories. If you are tempted by a pie chart here, see pie vs bar — bars almost always win for comparison.
Composition — pie or stacked bar
When the categories are parts of a single whole and you want to show their shares, use a pie chart (for a few categories, one snapshot) or a stacked bar chart (to compare composition across several groups or over time). Keep pies to a handful of slices; past about six, switch to bars. For composition that should be compared as percentages regardless of total size, a 100% stacked bar is ideal.
Trend over time — line or area
When a value moves across a continuous range — usually time — a line chart makes the trend and its turning points obvious. Plot multiple lines to compare how several series grow. Choose an area chart when the cumulative total is the message rather than the exact path. Avoid using a line for categories that have no natural order; those belong in a bar chart.
Distribution — histogram
When you want to see how one set of numbers is spread — are most values clustered in the middle, skewed to one side, or split into groups? — use a histogram. It groups values into ranges (bins) and shows how many fall in each. A histogram looks like a bar chart but answers a different question: a bar chart compares categories, a histogram shows the shape of one continuous variable. See bar chart vs histogram for the distinction.
Relationship — scatter plot
When you want to know whether two numeric variables move together, use a scatter plot. Each point is placed by its x and y value, so a rising cloud suggests a positive relationship, a falling cloud a negative one, and a shapeless cloud none. Scatter plots are also the best way to spot outliers. Remember that a relationship is not proof of cause.
Step 2: Check the details
Once you have the family, a few practical questions settle the exact chart:
- How many categories? Many categories rule out a pie chart and favour horizontal bars.
- Are the values parts of a whole? If not, do not use a pie chart, even for a few categories.
- Any negative values? Pies cannot show them; bars and lines can.
- How many series? Several series at once point to grouped/stacked bars or a multi-line chart.
- Is precision important? If readers must judge exact differences, prefer length-based charts (bar, line) over angle-based ones (pie).
Pick the chart that lets the reader judge values by length or position rather than angle or area. That single rule resolves most ties in favour of clarity.
Step 3: Build it and test it
The fastest way to confirm a choice is to make the chart and look at it. If it is hard to read, try the runner-up type with the same data — every maker below is free, instant, and exports PNG or SVG.