Data Slide Design

Data Slide Design Skill

A data slide has one job: the room, glancing, reaches the same conclusion the presenter did — and most fail by neutrality. The chart arrives titled "Revenue by Quarter," un-annotated, gridlines intact, every series equally weighted: technically accurate, rhetorically mute, and the room reads it six different ways (or not at all). The design discipline: the title is the takeaway ("Enterprise drove all of Q3's recovery"), the chart is stripped to what carries that point (chart-choice picks the form; this skill stages it), the annotation layer points the eye (the highlighted series, the arrow at the inflection, the labeled delta), and the honesty pass keeps persuasion from shading into distortion — because a room that catches one axis trick re-audits every slide you ever show.

What This Skill Produces

• The takeaway title — the slide's headline claim, doing the interpretive work
• The staged chart — one chart, stripped, with the point's series highlighted and the rest dimmed
• The annotation layer — the callout, the delta label, the reference line: the eye-path to the conclusion
• The honesty audit — axes, baselines, and selection checked; the caveat line where the data needs one

Required Inputs

Ask for these if not provided:

• The data and the point — the chart (or numbers) and the one sentence it must prove; a chart without a committed point routes back to deck-outline-first
• The delivery mode — projected (bigger, barer) vs. reading deck (annotations can carry more) — the slide-density-rules fork applies
• The audience's data fluency — a room of analysts tolerates a scatter; a board wants the annotated line
• The data's soft spots — the caveat, the small n, the definition change mid-series (citation-hygiene: the as-of date and source line are non-optional)

Framework: The Design Rules

1. The title does the interpreting: "Churn spiked in month two — onboarding is the leak" as the title converts the chart from exhibit to argument; the neutral title outsources interpretation to a room that won't do it. The test: title covered, does the slide still make its point? (It shouldn't need to — the title is the point.)
2. One chart per slide, staged for its point: two charts split attention and slides are free (slide-density-rules); the surviving chart gets staged — the point's series in full color, comparisons dimmed to gray, gridlines lightened, the legend replaced by direct labels. Emphasis is a design layer, not a data change.
3. Annotate the path: the arrow at the inflection ("pricing change, March"), the bracket on the delta ("+$2.1M"), the reference line (target, last year) — annotations are the presenter's pointing finger, printed. Two or three maximum; an annotation crowd is a second wall of text.
4. The honesty pass, non-negotiable: bar axes at zero, zoomed line axes labeled, no dual-axis theater, the time window honest ("why does this chart start in March?" has an answer), the source-and-date line present (chart-choice rules enforced at slide stage) — and the caveat the data needs stated small but present ("n=340, pilot cohorts"). Emphasis persuades; distortion, once caught, un-persuades retroactively and permanently.
5. The glance rehearsal: three seconds, fresh eyes — what's the takeaway? If the answer isn't the title's claim, the staging failed: usually the emphasis is too even, the annotation is missing, or the chart form fights the point. Fix the stage, not the audience.