Pick up any default Excel chart and look at the title. You'll see something like 'Revenue by region' or 'Q3 conversion funnel'. Now pick up the front page of the Financial Times or a McKinsey deliverable. The titles read 'European revenue lags the US by 23 points despite higher product spend' or 'Conversion drops 40% between sign-up and first transaction.'
The difference between the two is not skill at design software. It is a difference of intent. The first chart describes a topic. The second chart argues a claim. Until you internalise that shift, every chart you make will read like a draft.
The economics of a chart
A reader spends three seconds deciding whether a chart is worth reading. In that window, they ask: what is this telling me? If the chart's title is descriptive, the reader concludes 'this is a piece of data the analyst wants me to interpret myself' — and most readers don't have time to interpret. They scan past. The work goes unread.
If the chart's title states a claim, the reader's brain shifts into a different mode: judging whether the claim is true. Now the chart is doing the work it was meant to do: arguing a position with visual evidence.
Descriptive vs action titles
DESCRIPTIVE: 'Quarterly revenue by product line, 2023-2025'. ACTION: 'Enterprise revenue grew 4x while SMB stayed flat — concentration risk has tripled in two years.' Same data. Different verb. Massively different impact.
The five-second test
Show any chart you've built to a colleague for exactly five seconds. Then ask: what was it telling you? If they can paraphrase the claim, the chart works. If they need to study it longer, the chart is asking the reader to do the analyst's job.
Publication-grade graphics — the FT, the Economist, NYT Upshot, Bloomberg — pass the five-second test trivially. Their action title states the claim. The chart itself, stripped of decoration, supports the title. An annotation points at the specific data point that drives the conclusion.
The three failure modes
(1) Multi-claim charts: trying to show three findings in one chart means the reader sees zero. (2) Unsorted bar charts: alphabetic or random order makes comparison impossible. (3) Default Excel palettes: the rainbow-of-colours signals 'I didn't think about this'.
Exercise
Find the last chart you produced for work, a paper, or a presentation. Apply the five-second test with someone unfamiliar with the data. Then rewrite the title as an action title that makes a single claim, and check whether the chart's design still supports that title — or if it needs simplifying.