Categorical vs. sequential vs. diverging: three palette types with different jobs
Data visualization palettes come in three structural types, and each has different design requirements. Categorical palettes assign one color per data category — think pie chart segments or multi-line charts. Their job is maximum distinctiveness: each color should be as different from its neighbors as possible in hue, and the palette must remain distinguishable in both screen and print contexts. Sequential palettes encode a continuous quantity — darker or more saturated means more. They should progress smoothly from low to high with no perceptual breakpoints. Diverging palettes encode deviation from a meaningful midpoint — positive vs. negative, above vs. below average. They use two contrasting hues that converge at a neutral midpoint color. Using a diverging palette when data has no meaningful zero (like revenue figures) creates false perceptual structure that misleads readers.
