What gets quoted is a paragraph, not a page

For twenty years, the thing you were trying to get ranked was the page. You wrote a page, it ranked, someone clicked it. Everything about how sites get built still assumes that: the title, the blurb underneath it in search results, the internal links. The whole apparatus points at a page as the thing being chosen.

That assumption has quietly stopped holding.

In July 2026 Pillarbase analyzed 15.7 million AI Mode citations across 148 industries and resolved them to 4.6 million unique passages sitting on 2.7 million pages. The first number they published is the one that reframes everything else: 48 citations in 100 did not link to the page at all. They linked to one highlighted sentence inside it. Google is not sending a reader to your page and letting them hunt. It opens the page on the line that answers them.

Which raises the obvious question. What makes a sentence the one that gets picked?

The data answers with unusual clarity. Among the paragraphs that get reused across many different questions:

Typical length                                    117 words
States the answer in the first sentence           81 in 100
Makes sense on its own, without the page around   85 in 100
Ones that open by naming the question             reused twice as often

The single most-recycled paragraph in the whole study appeared 661 times across 483 different questions. One paragraph, answering hundreds of unrelated questions, for months.

Read those four lines again, because between them they describe a way of writing that almost nobody uses.

A hundred and seventeen words is a full paragraph. Not a snippet, not a bullet, not a one-line definition. What gets pulled out is a substantial block of prose: long enough to hold a complete thought, short enough to drop into an answer without cutting it.

The answer comes first. Four in five of the reused passages state their conclusion in the opening sentence. This is the reverse of how almost every business writes about itself. The standard service page builds: context, then credentials, then the offer, then finally what it actually does. A model reading that page finds nothing it can use until paragraph six, and by then it has moved on.

It stands alone. Around 85 in 100 recycled paragraphs lean on nothing that came before them. No “as we mentioned above”. No “this approach”. No pronoun whose referent is two paragraphs back. If a paragraph only makes sense in sequence, it cannot be lifted, and a paragraph that cannot be lifted cannot be quoted.

It leads with the question. Paragraphs that open by naming the question they answer get reused roughly twice as often as the rest.

That last finding needs a careful distinction, because it looks like it contradicts something else in the research. A separate controlled study measured pages built as a list of frequently asked questions and found they did slightly worse than the rest, not better. So opening with the question helps, and the question-and-answer format does not?

Both are true, and the difference is the thing worth understanding. A paragraph that opens by stating the question it answers, in ordinary prose, reads like a person explaining something. A block of frequently asked questions is a template: a list of collapsed rows, each with a question in bold and two lines under it, usually written to catch search terms rather than to answer anything. The first is writing. The second is furniture. Models appear to reward the first and be indifferent to the second.

Now the part that keeps this honest. Writing better paragraphs does not replace ranking.

The same data shows that pages with 21 or more highlighted sentences almost always ranked first on Google, and that among the paragraphs cited more than a hundred times, three in four came from pages in the top spot. The author is explicit that this is a correlation and not a proven mechanism, but the direction is not ambiguous. You do not write your way into an AI answer from page four of Google. What good passage construction does is convert a page that already surfaces into a page that gets used.

There is one more measurement worth putting alongside this. A separate study of 21,143 citations counted how many sources each system uses: ChatGPT names about 7 per answer, Perplexity about 16. Then it measured how much each of those sources actually shapes the words that come out, and the ratio flips: a page cited by ChatGPT carries roughly four times the weight of a page cited by Perplexity. ChatGPT cites narrowly and reads deeply. Perplexity cites broadly and stays on the surface. Which means a citation count is not a visibility score, and the pages that get genuinely used are a much smaller set than the pages that get listed.