Color in Data Visualization: Choosing Palettes That Inform Without Misleading

Data visualization is one of the highest-stakes environments for color decision-making. The wrong palette can make a chart misleading, inaccessible to colorblind viewers, or simply unreadable when printed in grayscale. This issue covers the three main palette types (sequential, diverging, categorical), how to select and validate them, and the most common mistakes that turn an informative chart into a confusing one.

Data Visualization Charts Accessibility Color Systems

Highlights

Data visualization palettes have three fundamental structures: sequential (one hue, varying lightness — for ordered data with a clear low-to-high progression), diverging (two hues from a neutral midpoint — for data with a meaningful center, like profit/loss or temperature anomalies), and categorical (distinct hues for unordered groupings — for categories with no inherent order). Using a sequential palette for categorical data — or vice versa — creates false implied ordering and is one of the most common data visualization color errors.

Approximately 8% of men and 0.5% of women have some form of color vision deficiency. The most common is deuteranopia (red-green), which makes red and green indistinguishable at many lightness levels. A red/green comparison chart — still ubiquitous in financial reporting — is unreadable to roughly 1 in 12 male viewers. Validated colorblind-safe palettes (Okabe-Ito, ColorBrewer) use hue differences with sufficient lightness contrast to remain distinguishable under multiple deficiency types.

The grayscale test: convert your chart palette to grayscale and verify that all categories remain distinguishable. This matters because charts are frequently printed, photocopied, or viewed on low-quality displays. If two categories collapse to the same gray, they are not differentiable without the legend — and many viewers will not read the legend. Sequential palettes typically pass this test; categorical palettes frequently fail unless they were explicitly designed with lightness variation across hues.

Sequential vs. diverging vs. categorical: when to use each

Sequential palettes encode magnitude: the progression from light to dark (or vice versa) maps to low-to-high values. Use sequential when the data has a natural minimum and maximum without a meaningful midpoint — population density, sales volume, temperature (not anomaly). A single-hue sequential palette (e.g., light yellow → saturated orange → dark brown) is always safe; multi-hue sequential palettes (e.g., yellow-green → blue-green → dark blue) can increase discrimination at the cost of implying a direction change. Diverging palettes have two hues that meet at a neutral center. Use diverging when zero or the mean is meaningful — financial variance, survey agreement (strongly disagree to strongly agree), geographic deviation from average. The two endpoint hues should be perceptually equidistant from the neutral center in luminance. Categorical palettes need hues that are perceptually distinct and do not imply order. Maximum categorical discrimination: space hues at least 30-40° apart on the wheel, vary lightness slightly to add discrimination, and never use adjacent warm colors (yellow, orange, red) as separate categories — they look too similar under small chart element sizes.

Validating for colorblindness and grayscale

Run every data visualization palette through two validation tests before shipping. Test 1 — colorblind simulation: use a tool like Coblis or Figma's accessibility plugin to simulate deuteranopia, protanopia, and tritanopia. For each simulation, verify that all categories remain distinguishable. If two categories merge, replace one with a color that differs in lightness as well as hue — lightness difference survives all forms of color vision deficiency. Test 2 — grayscale: desaturate the chart entirely. Each category should remain distinguishable by lightness value alone. If you have 5 categories, you need 5 distinct lightness levels. The practical constraint: more than 4-5 categories in a single chart is usually a design problem, not just a color problem — the chart may need to be split or restructured rather than given more colors.

Newer issue

Color in Packaging Design: Substrate, Finish, and the Gap Between Screen and Shelf

2028-11-04

Older issue

Documenting Color Systems: What to Write Down and Why Most Teams Skip It

2028-11-18