Telling data stories with AI

1. Telling data stories with AI

Dashboards are neutral by design — they show conditions, they don't argue for anything. Data stories are different. They take a stance. When you want data to drive a decision, change a mind, or rally a team, you're telling a story — and that's a fundamentally different skill from building a dashboard.

2. What changes when you tell a story?

Data stories use specific levers to shape how an audience reads the data. Color and emphasis.

3. What changes when you tell a story?

Titles and labels.

4. What changes when you tell a story?

What you lead with and what you set aside.

5. What changes when you tell a story?

The order you reveal things in. That's the power — and the responsibility — of telling a story with data.

6. Same chart, two stories

Remember the chart from Chapter 1 — Americans' time use across their lives? In one form, the headline read "enjoying one's own company", the palette was a friendly blue, and the chart framed “time spent alone” as positive. Now keep the data exactly the same. Change the color to red. Swap the title for "ALL ALONE", and the same chart becomes a story about loneliness. Two stories, and not a single number has changed.

7. Editorialize, but don't deceive

That's the line. Telling a story with data means making editorial choices — what to emphasize, what context to provide, and what framing to use. That's fine, and often necessary — as long as the underlying data isn't distorted, cherry-picked, or stripped of context that would change the conclusion. Editorial framing is honest. Hiding contradicting data isn't.

8. AI can draft the story for you

Let's say we're the sales leader at a leather goods company - the one called Iron and Grain. We've got this quarter's sales data, and we want a celebratory slide deck — something that highlights our top performers and encourages the rest of the team. We hand Claude the data, tell it the goal, the context, the scope, and with examples, and ask for a theme that fits the brand.

9. AI can draft the story for you

A few minutes later, we have a complete deck. Charts, headings, narrative copy, even a creative touch like using numbers instead of names in the leaderboard. It is ready to present.

10. State your angle, every time

Here's the catch: AI doesn't know what story we want unless we tell it. Watch what happens when we hand the same data to two tools without specifying an angle. Claude pauses and asks us — what narrative do you want? Celebratory, cautionary, neutral? Gemini just picks a data story, a framing, usually the most positive read of the data, and produces output based on that. Both behaviors are common, and the fix is simple: always state the angle in your prompt. Tell it who the audience is, what decision you're supporting, and what tone is appropriate. Otherwise AI picks for you — and it may not be the angle you want.

11. What to check beyond accuracy

You already know to check AI's calculations. With data stories, there's more. Four questions to ask. One: did AI editorialize beyond what you intended? Two: did it leave out context that would change the conclusion? Three: does the tone match the audience? Four — uncomfortable but necessary — did any biases creep in that you didn't notice? It's your name on the deck. Read it the way a skeptical reader would.

12. Let's practice!

Time to take some data and tell a story with it.

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.