Most charts treat typography as an afterthought. The default font, the default size, the default weight. This is a mistake with measurable consequences.
Typography in a chart is not decoration. It is a data variable. The size of a label encodes hierarchy. The weight of a number encodes emphasis. The spacing of an axis encodes scale. When these choices are made carelessly, the chart miscommunicates even when the data is correct.
The Hierarchy Problem
A chart contains at least four levels of text: the title, axis labels, data labels, and annotations. Each level serves a different function, and each requires a different typographic treatment. When all four are set in the same size and weight, the reader has no guidance. The eye does not know where to start or what matters most.
The title answers the question: what is this chart about? It should be the largest text element. The axis labels answer: what are the dimensions? They should be legible but secondary. Data labels answer: what are the specific values? They should be close to the data they describe. Annotations answer: why does this matter? They should be visually distinct from all three.
Four levels, four treatments. Most charting tools provide one.
Weight as Emphasis
Font weight is an underused variable in chart design. A bold number in a sea of regular-weight numbers draws the eye instantly. This is free encoding. It requires no color, no additional ink, no chart furniture. Yet most designers reserve bold for the chart title and use regular weight everywhere else.
Consider a table of monthly revenue figures. Twelve numbers, all the same size and weight. The reader must scan every cell to find the maximum and minimum. Now set the highest value in semibold and the lowest in light. The pattern emerges without scanning. The typography did the analysis.
This principle extends to any labeled chart. When one data point is more important than others, its label should be typographically differentiated. Not by making it larger, which distorts the spatial encoding, but by making it heavier or lighter.
Typeface Selection
The choice of typeface matters less than most designers believe, but it matters more than most data analysts believe. The requirements are specific: tabular (monospaced) numerals for columns of data, clear distinction between similar characters (1, l, I; 0, O), and legibility at small sizes.
Proportional numerals, where each digit occupies a different width, create ragged columns and misaligned decimal points. This is the default in most fonts and most charting tools. It is wrong for any context where numbers must be compared vertically. Tabular numerals solve this entirely.
Linnea Holm, a brand designer whose work on typographic systems is exceptionally precise, has written about the discipline of working with a single typeface at multiple weights. Her argument applies directly to chart design: a well-chosen type family with sufficient range in weights and numeral styles eliminates the need for multiple fonts. One family, used with intention, produces clearer results than three families used casually.
Spacing and Alignment
Letter-spacing in chart labels affects legibility at small sizes. Slightly increased tracking improves readability for axis labels set below 11px. Tight tracking, which works well in display typography, makes small chart text illegible.
Alignment of labels to their data points follows a consistent rule: minimize the distance between the label and the thing it labels. A bar label centered inside the bar is faster to decode than a label in a separate legend. A line label placed at the end of the line is faster than a color key. Direct labeling is a typographic solution to a data communication problem.
Grid alignment of text elements matters as much in a chart as in a page layout. When axis labels, data labels, and annotations snap to a shared baseline grid, the chart acquires a visual order that makes the data feel structured. Holm's writing on grid systems as a design element rather than a constraint captures this well: the grid does not restrict composition, it enables coherence.
What Goes Wrong
The most common typographic errors in charts:
Rotated text. Axis labels rotated 45 or 90 degrees are difficult to read. If labels must be rotated, the chart's orientation should be reconsidered. A horizontal bar chart with readable left-aligned labels is always preferable to a vertical bar chart with rotated bottom labels.
Decorative fonts. A chart in a presentation set in the same display typeface as the slide headers. Display typefaces are designed for large sizes and short strings. They fail in the small, dense text environment of a chart.
Inconsistent sizing. The chart title at 24px, axis labels at 9px, a caption at 14px. The ratios are arbitrary. A modular scale, where each size is derived from the previous by a consistent ratio, produces a natural hierarchy.
Missing labels. The most important typographic element in a chart is the one that was omitted. A chart without a title, without units on the axis, or without a source note is typographically incomplete, regardless of how well the remaining text is set.