Data Visualization Is Writing

A chart makes an argument. Everything in it — the axis scale, the color, what's left out — is a rhetorical choice. Treat it like prose.

The argument a bar chart makes

Every visualization has a claim. Usually it's implicit. "Sales went up." "These two groups are different." "The trend is accelerating."

The problem is that implicit claims are harder to scrutinize than explicit ones. When you write a sentence, a reader can push back. When you show a chart, the data feels like it speaks for itself — even when the designer made choices that shaped what the data is saying.

What you choose when you make a chart

The baseline. Starting a bar chart at anything other than zero visually amplifies differences. This is sometimes honest (when the actual differences are what matter) and sometimes misleading. Knowing the difference is an ethical skill, not just a technical one.

What's included. Every axis has a start and end. Every dataset has a time range. Every comparison omits other comparisons. What you exclude shapes what the chart proves.

The color. Color creates hierarchy. It suggests which series matters most, which bars to look at first, which line to follow. This is writing — it's emphasis and de-emphasis.

The title. Most charts are under-titled. "Monthly revenue" is not a title; it's a label. "Revenue grew 34% after the pricing change" is a title. It tells the reader what to see.

Making it a habit

Before I publish a chart, I write one sentence: this chart argues that...

If I can't finish the sentence, the chart isn't ready. If finishing the sentence reveals something I don't want to argue, I reconsider the chart.

The goal isn't to make charts that are neutral — there is no such thing. The goal is to make charts that are honest about the argument they're making.