A Sankey diagram or violin plot needs its own reading instructions, since daily habit won't decode it

What it is

Sankey diagrams, chord diagrams, violin plots, radar charts, streamgraphs—these chart types lie outside the common graphic grammar, and readers cannot guess the encoding rules from daily experience (is a chord's width traffic or headcount? Are a streamgraph's layers stacked or independent?). When an uncommon chart's reading logic exists only in the author's head, readers' decoding either fails or becomes guesswork. Reading instructions (an embedded "how to read this chart" example, annotation interpreting a concrete element, animation demonstrating the mapping) transfer the teaching cost from readers to the designer, and are the precondition for uncommon charts being read correctly at all.

Why it happens

Reading instructions work by compressing "learning a new chart" into "looking at one example": decoding an unfamiliar chart requires readers to complete a three-stage inference—visual element → encoding rule → numeric meaning—and every stage can go wrong (reading a chord's width as distance rather than traffic). An embedded instruction hands readers the completed inference chain—"this ribbon's width represents the number of people migrating from A to B; wider means more"—so readers match rather than infer. Effective instructions share three structural features: they target one concrete element that actually exists in the chart ("look at this ribbon from 'Mobile' to 'Desktop'") rather than abstractly stating rules; they demonstrate one complete act of interpretation (reading out a specific value or relation) rather than only defining encodings; and they sit adjacent to the element being explained (in-chart annotation beats footnotes, footnotes beat external docs). Uncommon charts also have an interactive option—dynamic demonstration: the graphic gradually walks through its mapping (one ribbon highlighted with a numeric callout, then "width = people" appears). This "teach by doing" mode yields the largest gains in reading accuracy for spatially complex charts (chord diagrams, streamgraphs).

Where it stops holding

Reading instructions have clear cost boundaries: they occupy chart space (limited canvas), add visual complexity, and are redundant for readers already fluent in the chart type. So the depth of instruction should scale with the audience's expected familiarity: for mixed audiences, make the full instruction visible by default with a collapse option for the fluent; for purely professional audiences, instruct once at first use. Another boundary is that "uncommon" is reader-relative: the same chord diagram is common in the graph-database community (no instruction needed) and uncommon in a board presentation (instruction mandatory). The designer's characteristic error is substituting their own familiarity for the readers'—"I think this is common" usually reflects the designer's circle, not the target audience.

Applying it

• When introducing a nonstandard chart, embed one reading example targeting a concrete element, placed adjacent to it.
• When space is tight, use an instruction card that is "expanded on first view, collapsible," defaulting to expanded so first-time visitors see it.
• For spatially complex charts (chord diagrams, streamgraphs), prefer dynamic demonstrations of the mapping over static annotation.
• Verification: ask a reader who has never seen the chart type to answer one numeric question with the instruction visible; a wrong answer indicates the instruction's wording is ambiguous, not a reader deficit.

Related

• Same group: U9.05.1 Reader graph literacy caps the usable complexity of a chart · U9.05.2 Conventions familiar to domain experts do not hold for outside readers · U9.05.4 Readers' prior positions shape their interpretation of the same graphic · U9.05.5 Chart density for decision-makers and for peers should not be identical
• Nearby: U9.05.2 Conventions familiar to domain experts do not hold for outside readers · U9.03.1 Annotations point to where readers should look
• Search terms: how to read · chart explainer · sankey diagram