The One-Idea-Per-Paragraph Discipline

Programmers have a rule that a function should do one thing. Nobody thinks this rule is about making programs shorter. It’s about making them possible to reason about. When a function does one thing, you can name it honestly, test it in isolation, and reuse it without dragging in surprises. The same rule exists for prose, and it’s just as underrated: a paragraph should contain one idea.

I want to be careful about what “one idea” means, because taken too literally the rule produces nonsense. One idea doesn’t mean one sentence, or one fact. An idea usually needs support: a claim, a reason, maybe an example. All of that belongs together, because it’s all in service of the same point. The rule isn’t “keep paragraphs small.” It’s “keep paragraphs coherent.” A six-sentence paragraph that builds a single argument obeys the rule. A two-sentence paragraph that makes two unrelated claims breaks it.

The test I use is naming. If you can give a paragraph an honest one-line name, it contains one idea. If the honest name would need an “and” in it — “why our tool is fast, and also our pricing” — the paragraph is two paragraphs wearing a trench coat. This is exactly the test programmers use on functions, and it fails in the same telltale way.

Why does this matter for a web page more than for a book? Because of how web pages get read. Someone skimming your page on their phone doesn’t read paragraphs; they sample them. They read the first line of a paragraph, decide whether it concerns them, and either continue or jump to the next block. This sampling strategy only works if paragraphs are honest units. If your paragraph opens with a line about pricing and then quietly pivots to security compliance in sentence four, the reader who skipped ahead after the first line never sees the security part. You wrote it, but you shipped it somewhere nobody looks.

That first line deserves special attention. Journalists used to call it the topic sentence, back when everyone was taught to write one, and the discipline mostly survives in journalism because journalists face the most impatient readers on earth. The first sentence of a paragraph is a promise about what the paragraph contains. If the paragraph delivers the promise, the reader learns to trust your first sentences, and a trusted first sentence is what makes skimming actually work. A page where every paragraph opens with its point can be understood at two speeds: fast, by reading only the openings, and slow, by reading everything. That’s not a compromise. That’s a feature the writer built on purpose.

The discipline pays off before publication too, and this is the part I find most interesting. When you force yourself to give each idea its own paragraph, you discover how many ideas you actually have. Sometimes it’s fewer than you thought: two paragraphs turn out to be the same idea in different clothes, and you merge them. Sometimes it’s more: one bloated paragraph splits into three, and one of the three turns out to be the most important thing on the page, previously buried in the middle of a block where nobody would have found it. Either way, you learn something about your own argument that you didn’t know when it was all fused together. The constraint isn’t a cage. It’s a diagnostic.

I see the failure mode constantly in the audits my tool runs. A typical homepage paragraph tries to do four jobs at once: state what the product is, differentiate it from competitors, reassure the reader about security, and ask for the signup. Each of those jobs, given its own paragraph, would take about two sentences and land cleanly. Fused, they cancel each other out. The differentiation dilutes the definition, the reassurance interrupts the pitch, and the ask arrives before the reader has absorbed any of it. The writer put more in and the reader got less out.

There’s a deeper reason this happens, and it’s worth naming: fusing ideas together is easier than ordering them. When ideas share a paragraph, you never have to decide which comes first, which depends on which, or which one you’d cut if you could only keep two. Separation forces sequence, and sequence forces priority. That’s uncomfortable. It’s also the actual work of writing. A page whose paragraphs each hold one idea, in an order that makes sense, is a page where someone made every hard decision. You can feel it when you read it, even if you can’t say why.

So the rule I’d give anyone writing for the web is simple to state and hard to follow. Before you publish, walk through your page paragraph by paragraph and name each one in a few words. Where you can’t, split or cut until you can. When you’re done you’ll have something rarer than polished prose: a page you fully understand. Your readers will understand it too, including the majority who only ever read your first sentences. Especially them.