How do you structure a data story so it drives a decision rather than just presenting findings?
A data story has three components: a clear narrative arc (situation, complication, resolution), charts that each advance one argument rather than display all available data, and deliberate attention direction through annotation, color emphasis, and sequencing. The goal is that a viewer reading only the titles and callouts should understand the conclusion without reading every axis.
How to think about it
The narrative arc
Barbara Minto’s Pyramid Principle and Cole Nussbaumer Knaflic’s “Storytelling with Data” converge on the same structure:
- Situation — the stable context your audience already agrees on (“Our Q2 retention rate is 72 %, in line with the prior three quarters.”)
- Complication — the tension that creates a need for action (“However, the cohort that onboarded after the March pricing change shows 55 % retention at day 30, eight points below the historical norm.”)
- Resolution — the recommendation or decision that follows (“We recommend a targeted win-back campaign for that cohort while the UX team investigates the onboarding friction.”)
Every slide or section should carry one of these three roles. A slide that carries no role is filler.
One chart, one argument
Each chart should be buildable into a single declarative sentence that is also its title: “Retention dropped sharply for the March cohort.” If you cannot write that sentence, the chart is not ready. Exploratory charts (showing everything) belong in an appendix; presentation charts earn their place by answering a specific question.
Directing attention
Use pre-attentive attributes to guide the eye before cognitive processing kicks in:
- Color saturation: gray out the irrelevant series, saturate the one that matters.
- Annotations: a text callout on the chart at the inflection point removes guesswork.
- Bold or enlarged data labels on the key bar or line.
The viewer should not have to hunt for the point. If the insight requires explanation, the chart is not doing its job.
The “so what” test
After every chart ask: “So what does this mean for the audience?” If the answer is “it depends” or “interesting,” the chart does not belong in the deck. A chart earns its place only when the “so what” is actionable.
Sequencing
In a slide deck, build complexity progressively. Start with the headline metric, then reveal the breakdown, then show the root-cause evidence. Never front-load a complex four-panel chart — the audience will spend the first two minutes reading it while you are talking about something else.