This entry is part of the Chart Index, the reference library for the Chart Design Field Guide.

The radar chart — also called a spider, web, or polar profile — arranges multiple variables on radial axes emanating from a centre, then connects the values along the axes into a polygon. Each shape's silhouette becomes a profile — an entity's footprint across several dimensions, ready to be compared visually with other entities' silhouettes.

It is the form most often used in performance reviews, product comparison spec sheets, sports analytics, and competitive intelligence decks. It is also one of the most criticised chart types in the perceptual literature. Tufte, Few, and Cleveland have all written against it; analysts continue to use it because, despite the criticisms, it produces a striking single-entity profile that no other form matches.

What it is

A radar chart maps several quantitative variables to radial axes radiating from a common centre. Each axis runs from a minimum (at the centre) to a maximum (at the outer ring). An entity's values across these dimensions are plotted as points on the axes; the points are connected to form a polygon. Multiple entities can be plotted as overlapping polygons, but the form is most readable for one entity at a time.

The hero chart above scores two releases of the same product on eight capability areas, each normalised to 0–100 with larger meaning better. The blue polygon (v3) bulges towards speed, accuracy and support and pinches at cost and maturity; the grey polygon (v2) is the mirror image — mature and cheap, slower and narrower. That is the one thing the form does well: two silhouettes, one glance, a sense of shape. What it does not do is let the reader say by how much v3 leads on coverage. Randomise the scores and watch the polygons flex — and notice how hard it is to say which spoke moved most.

When to use it

Radar charts are the right choice when:

  • You are creating a profile or silhouette of a single entity across several dimensions.
  • The reader's question is "what is this entity's shape across multiple attributes?"
  • The dimensions are comparable — same scale, same direction-of-good.
  • You have 3–8 axes. Fewer than 3 and the form collapses; more than 8 becomes unreadable.
  • The audience is familiar with the form or you have room to introduce it.

When not to use it

  • Precise comparison across many entities. Overlapping polygons become uninterpretable past two or three.
  • Heterogeneous scales. If your axes mix percentages, dollars, durations, and counts, the form's common-scale assumption breaks.
  • Unordered axes. Radar charts imply that adjacent axes are related; arbitrary axis order produces meaningless shapes.
  • Audiences expecting precision. Few warns against radar in business reporting because executives misread the encoding.

Design principles

Order axes deliberately

Adjacent axes in a radar chart visually correlate — the polygon's shape connects them. Group related axes (Speed and Accuracy adjacent; Cost and Maturity adjacent; soft attributes like Ease of use and Support adjacent). Random order produces a shape whose silhouette means nothing. The two figures below plot the same eight scores; the only difference is the order of the spokes.

Spokes grouped by theme
Performance spokes at the top, commercial spokes at the bottom, experience spokes on the left. The polygon reads as a story: strong on the engineering side, weak on the commercial side.
Spokes in alphabetical order
Identical scores, spokes sorted A to Z. The polygon is now a jagged star, and the notches say nothing about the product — they are artefacts of the alphabet. The shape the reader remembers is entirely a design decision.

Keep the overlap to two or three profiles

For a single entity, a single colour suffices. For multiple entities, use distinct hues but keep saturation low and the fills translucent. Two or three overlapping polygons work; four or five become a quilt in which no individual profile can be traced. The pair of figures below adds three competitors to the v3 profile.

One profile: the silhouette is legible
A single translucent polygon over quiet rings. The eye reads the shape immediately and can name the strong and weak spokes.
Four profiles: nothing can be traced
The same product against three competitors. Every polygon crosses every other; the multiply blend keeps the overlaps visible but the individual silhouettes are gone. Past three entities the form has stopped working.

Normalise all axes to the same range

Every axis must run from the same minimum to the same maximum (typically 0 to 100, or 0 to 1). Mixed scales make the polygon's shape misleading. If your real data spans different ranges, explicitly normalise and document the normalisation. Pin the outer ring to the scale's maximum rather than to the largest value in the data; otherwise a product that scores 60 everywhere fills the chart and looks perfect.

Show the axes with quiet rings

Concentric rings in the background (typically at 20, 40, 60, 80 and 100 per cent of max) provide reference without dominating the data polygon. Use light grey strokes; no fill. A polygonal grid emphasises the spokes; a circular grid emphasises the scale — either is fine, provided it stays lighter than the data.

Label axes clearly outside the radar

Axis labels sit outside the outer ring, not on top of the data polygon. Each label should be horizontal where possible (rotating labels degrades readability). Give the frame enough side margin that the longest label is not clipped — Ease of use needs more room than Cost.

Offer the bar chart when precision matters

The radar's silhouette is memorable; its values are not. When the reader needs to say which release leads on which capability, and by how much, the same eight scores as paired horizontal bars answer the question directly: every value sits on a common baseline and every gap is a length the eye can measure.

The hero data as paired bars
The same two releases, eight capabilities, on one common scale. The reader can see at once that v3 gained most on speed and coverage and gave up most on cost and maturity — the sizes of the gaps, which the radar only hinted at, are now the message.

Annotate notable extremes

If the entity excels or struggles on a specific axis, mark it with a short annotation. Speed: 88, exceptional or Cost: 46, weakest dimension. The polygon shows the shape; the annotation explains the meaning.

Resist using it for time-series

Some implementations animate the polygon over time. The form is poor for trend reading — radar is for profile, not trajectory. For trends, use a line chart per dimension as small multiples.

Anatomy

The composition of a radar chart
An anatomical guide

A radar chart's anatomy is concentric reference rings, radial spokes, a data polygon with a vertex on every spoke, and axis labels arranged around the perimeter. The form's reading is visual silhouette; the annotations support it.

Reading list

  • Few, S. (2005). Keep Radar Graphs Below the Radar. The definitive critique.
  • Cleveland, W. & McGill, R. (1984). Graphical Perception. The ranking of perceptual channels that underpins the critique.
  • Munzner, T. (2014). Visualization Analysis and Design. The taxonomy of profile-comparison chart forms.