Direct Labeling
Direct labeling puts a category’s name on or beside the mark that represents it, instead of in a separate legend. The bar labeled “Returns” is the bar named Returns; nothing has to be matched.
A legend splits one reading task into four. The reader looks at the mark, looks at the legend, matches the color or symbol, and returns to the mark — while holding the mapping in working memory. Repeat that for every series and the cost is no longer trivial.
The readers a legend fails hardest are predictable. Anyone who cannot reliably separate the legend’s hues loses the mapping entirely, which is why WCAG 1.4.1 requires that color never be the only visual means of distinguishing an element. Anyone scanning quickly, or reading a small rendering, pays the cost every time.
The accuracy argument is separate and often missed: a direct label sits at the mark’s own position, and position along a common scale is the most accurately read visual property there is. A direct label removes a separate matching step when it is readable beside the mark; dense or overlapping labels can introduce their own reading difficulty.
A legend remains reasonable when labels would collide — many overlapping series, or a dense scatterplot where there is no free space beside any point. The usual remedy is to label the few series that carry the argument and leave the rest muted, rather than labeling everything or nothing.
References:
Discover more from Insightful Data Lab
Subscribe to get the latest posts sent to your email.
