Color for Data Visualization: Clarity Over Aesthetics

Data visualization requires a different relationship with color than brand or interface design. In brand design, color communicates identity and emotion. In data visualization, color communicates information — and getting it wrong means readers misinterpret data. The stakes are higher and the constraints are more specific. Understanding the grammar of data color is essential for any designer who works on dashboards, reports, analytics products, or data journalism. The three fundamental scales of data color are categorical, sequential, and diverging. Categorical scales assign distinct colors to unordered groups (product categories, countries, demographic segments) — the primary requirement is discriminability: each color must be distinguishable from every other color, at small sizes, under varied display conditions, and by people with common color vision deficiencies. Sequential scales encode magnitude on a continuous spectrum from low to high — effective sequential scales maintain perceptual uniformity (equal visual steps for equal data steps) and work in both color and grayscale. Diverging scales have a meaningful center point and two opposing directions — budget surplus vs. deficit, temperature above vs. below average — requiring two sequential scales meeting at a neutral midpoint. Accessibility for color vision deficiency is not optional in data visualization — approximately 8% of male readers have some form of color vision deficiency, and red-green deficiency is the most common. The most reliable categorical palette for color-blind users avoids relying on the red-green distinction as a primary differentiator. The Okabe-Ito palette (developed by Masataka Okabe and Kei Ito) is the most widely recommended accessible categorical palette: eight colors chosen to be discriminable by people with deuteranopia, protanopia, and tritanopia simultaneously. For sequential scales, single-hue progressions (from near-white to a saturated hue) are generally safer than multi-hue ramps, which can create false hue-based distinctions in the data. The most common errors in data visualization color are using too many colors (most human vision can reliably distinguish 5-7 categories, not 12), using the same hue for different data types in the same visualization (which implies relationship where none exists), and using fully-saturated colors in a table or chart (which creates visual noise that competes with the data). The principle of restraint applies with unusual force: the best data visualization color is invisible as design — it only becomes visible as information.