Radar chart

A radar chart (also called a spider chart) plots several numeric values for a single item on axes that radiate from a center point, then connects the dots to form a shape. Multiple items can be overlaid as overlapping shapes to compare them. Radar charts are good at one specific job — profiling something across several comparable attributes — and bad at almost everything else.

They're also one of the most-misused chart types in dashboards. The reason is that they look impressive and flexible, so people use them for data that doesn't fit. Getting this chart right is mostly about knowing when not to use it.

When to use it

Use a radar chart when all three are true:

1. You're profiling an item across multiple attributes. Rating a candidate on six skills, scoring a product on five features, evaluating an athlete on four metrics. The shape becomes the item's "profile."
2. All the attributes use the same scale. Every axis must go from the same minimum to the same maximum, in the same units, with the same direction (higher = better, or lower = better, but consistent). This is the single most important rule.
3. You have 3–8 attributes and 1–3 items. Below 3 attributes the chart is pointless; above 8 it's unreadable. Beyond 3 overlapping items, the shapes become a tangle.

Common uses: skill assessments, product comparisons, performance reviews, character-stat panels in games, feature parity analysis.

The critical rule: axes must be comparable

This is where radar charts go wrong more often than any other chart type. If the axes aren't comparable, the shape is meaningless.

Incompatible axes look like this:

• Revenue ($500k), Satisfaction (4.3/5), Response time (120ms), Headcount (42)

Four different units, four different ranges. Any tool will happily plot them, but the resulting shape is arbitrary — scaled by whichever axis has the biggest raw numbers. Revenue will dominate the chart simply because $500,000 is bigger than 4.3.

Compatible axes look like this:

• Technical (8/10), Communication (7/10), Leadership (6/10), Problem-solving (9/10), Teamwork (8/10), Creativity (5/10)

Same unit (rating out of 10), same range, same direction (higher = better). The shape now means something: a larger, more even shape is a stronger, more balanced candidate.

If your data has mixed units but you really want a radar chart, you have to normalize everything to a common 0–100 or 0–1 scale first. This works but is risky — readers may not realize normalization happened, and the "profile" they see is partly an artifact of how you scaled the numbers. Say so in a caption if you do it.

Example data

Radar data is in wide form — one row per item, one column per axis. Here are three candidates for a product manager role, each rated 1–10 on six competencies:

Candidate	Technical	Communication	Leadership	Problem-solving	Teamwork	Creativity
Alice	8	9	7	8	9	6
Bob	9	6	5	9	6	8
Carol	7	8	9	7	8	7

Plotted as three overlapping radar shapes, the differences jump out: Alice has a large, rounded shape (well-rounded generalist), Bob has a spiky one (deep technical but weak on people skills), Carol has the most balanced shape of the three. That insight would take longer to pull out of the table.

Most code libraries want the same data in long form instead — one row per axis-value combination:

Candidate	Attribute	Score
Alice	Technical	8
Alice	Communication	9
Alice	Leadership	7
...	...	...

Both contain the same information. Spreadsheet tools want wide form; charting libraries usually want long form.

How to read it

Scan a radar chart in this order:

1. Look at the overall size. A larger shape means higher values overall. Among multiple items, the biggest shape is the strongest performer across all attributes.
2. Look at the balance. A rounded, even shape means the item is consistent across attributes. A spiky shape means the item is strong in some and weak in others.
3. Look at which axes the shape reaches. Each axis is an attribute, and the shape's distance along that axis is the item's score. Look at where the shape extends furthest and where it falls short.
4. Compare shapes directly. When two shapes overlap, the one that extends further on a given axis is stronger on that attribute. Where shapes cross, that's where their strengths flip.

The area fallacy

One thing not to do: treat the filled area inside the shape as a "score."

The area of a radar shape depends on the order of the axes, not just the values. Swap two axes around and the area changes, even though the data is identical. If your attributes could reasonably be listed in any order (like the six skills above), the area is an artifact of axis ordering — not a real metric.

The exception is when axes have a natural order — days of the week going clockwise, hours in a day, months of the year. In those cases the area is meaningful because reordering would break the chart's logic.

If you want a single summary score, compute the average of the attributes and show it as a number next to the chart. Don't let readers infer it from the area.

Radar vs. small multiples — which to use?

When comparing several items across several attributes, radar is one option. Another is small multiples: a separate bar chart for each item, arranged in a grid.

• Radar is best when attributes form a meaningful profile and readers will compare 2–3 overlapping shapes. The overlap is the whole point.
• Small multiples is best when you have more items (5+), or when precise comparison of individual attributes matters more than overall shape. Each bar chart shares the same axis, so comparing, say, "Technical" scores across all candidates is far easier.

A quick test: if you find yourself squinting to compare two values on the same axis of a radar chart, small multiples will serve your readers better. Radar is a pattern-recognition chart, not a precise-comparison chart.

How to build one

In a spreadsheet (Excel or Google Sheets)

• Excel: select your data in wide form (items as rows, attributes as columns), then Insert → Charts → See All Charts → Radar. Excel offers filled and unfilled variants.
• Google Sheets does not have a native radar chart. You'd need a third-party add-on, or build one manually using a scatter plot with polar-coordinate math. For anything serious, export to another tool.

Excel's default radar styling is dated — thick lines, default colors, large markers. Clean it up by thinning strokes, using semi-transparent fills (40–60% opacity), and removing the data markers if the shapes overlap heavily.

With code (for a dashboard or report)

Most libraries have a radar chart type. The data pattern depends on the library:

// Long form — what most libraries expect (Chart.js, ECharts, Plotly, Recharts)
const data = [
  { candidate: "Alice", attribute: "Technical",       score: 8 },
  { candidate: "Alice", attribute: "Communication",   score: 9 },
  { candidate: "Alice", attribute: "Leadership",      score: 7 },
  // ... one row per axis-value per candidate
];

// Wide form — what some libraries also accept
const series = [
  { name: "Alice", values: [8, 9, 7, 8, 9, 6] },
  { name: "Bob",   values: [9, 6, 5, 9, 6, 8] },
  { name: "Carol", values: [7, 8, 9, 7, 8, 7] },
];
const axes = ["Technical", "Communication", "Leadership",
              "Problem-solving", "Teamwork", "Creativity"];

Always set the axis minimum and maximum explicitly (min: 0, max: 10 in this example). Letting the library auto-fit the axes will exaggerate small differences and mislead readers.

Tips

• Keep to 3–8 axes. Fewer is pointless; more is unreadable. If you have 12 attributes, group them into 5–6 higher-level categories, or switch to parallel coordinates.
• Keep to 1–3 overlapping items. Beyond 3, the shapes become a knot. If you need to compare 5 products, use small multiples instead.
• Use semi-transparent fills. Solid opaque fills hide whichever shape is behind. 40–60% opacity lets all shapes stay visible when they overlap.
• Label axis ends with the maximum value. Readers shouldn't have to guess whether the outer ring is 10, 100, or 1,000. A small "10" at each axis tip removes all doubt.
• Pick axis order deliberately. Related attributes should sit next to each other (all "hard skills" on one side, all "soft skills" on the other). Random order produces random-looking shapes.
• Don't use radar charts in presentations to general audiences. Like box plots, they require a learning curve. If your audience will see it once and never again, a bar chart conveys the same information with zero training required.
• Consider a parallel coordinates plot for more than 8 attributes or more than 3 items. Same idea — multiple numeric attributes per item — but laid out as vertical parallel lines. Handles more variables and items gracefully.