Key Takeaways — Chapter 9: Visuals and Data

The summary card. Read this to re-ground before the next chapter or a writing task.

The one idea

A figure does not speak for itself. Data is not a conclusion. A reader—especially a scanner or a non-expert—looks at your bars and sees shapes, not the insight you spent weeks earning. The caption and the surrounding prose are what make a figure mean something. The most important sentence attached to any figure is the one that says what it means. Once you cross this threshold, you stop labeling figures and start interpreting them—and your data begins to persuade.

Choose the tool by the reader's task

The deciding question is never "what looks professional?" but "will the reader see a shape, look up a value, or follow an argument?" And never deliver the same fact two ways—that redundancy is its own mistake.

Design clean (Tufte)

Maximize data-ink (the ink that represents data); delete chartjunk (3-D, heavy gridlines, gradients, redundant legends, rainbow color) because it obscures and sometimes distorts the data. Label directly instead of using legends. Start bars at zero (a bar encodes value by length, so a truncated axis lies). Pick the chart type that fits: trend → line, category comparison → bar, correlation → scatter; avoid pies.

The chart you make to find something (exploratory) is not the chart you publish to show something (explanatory). Don't paste your search; rebuild the one point for a reader who wasn't in your head.

Captions that interpret — climb the levels

• Level 1 (label): what it is — "Figure 3. Churn by cohort." (Where most writers stop. Not enough.)
• Level 2 (observation): what it shows — "Churn rises with onboarding time, 3% to 22%."
• Level 3 (interpretation): what it means — "Onboarding, not price, drives churn… intervene in the first week." ✅

Template: [finding], shown by [comparison]: [values], which means [implication]. A caption-only reader should leave with your actual conclusion.

Writing about data in prose — lead with the finding

Weak: "Figure 3 shows the relationship between onboarding and churn." Strong: "Slow-onboarding customers churn 7× as often (Figure 3)."

The formula: claim + key number + (Figure N). State the finding as the main clause, put the load-bearing number in words, demote the figure to a parenthetical citation. Interpret the number (is it surprising? why?); don't narrate the chart's wiggles.

Interpret, don't present

State the finding, then marshal the figures as evidence in an order that builds the case to an action. Cut exhibits that don't serve the argument. A warehouse of "Figure N shows…" is an inventory; a claim backed by chosen evidence is an argument.

The cautionary tale

The Challenger charts were accurate, but the decisive relationship—colder temperature, more O-ring damage; the forecast was colder than any prior flight—was scattered across many exhibits and never isolated into one clean figure (a temperature-vs-damage scatter, sorted, with the forecast marked outside all experience). The pattern that should have leapt out stayed invisible to time-pressured readers. Accuracy is not the bar; making the critical pattern unmissable is. (Tufte's analysis; verifiable history only. Returns in Ch 38, ethics.)

Themes this chapter surfaced: #5 structure-serves-the-reader (central) · #3 clarity-isn't-the-enemy-of-precision · #6 every-element-earns-its-place · #7 the-best-writing-is-invisible.

Threshold concept: A figure does not speak for itself; the caption and the prose make it mean something.

Feeds forward to: Ch 18 (posters), Ch 27 (data memos), Ch 30 (slides), Ch 32 (diagrams), Ch 38 (ethics).

Back to: Chapter 9 · Exercises · Quiz · Further Reading