Extractability — Writing Passages AI Will Quote
Answer engines cite passages, not pages. Write the sentence you want quoted — and put it first.
Extractability is the third pillar of the AEO Canon — answer engines cite passages, not pages, so write the sentence you want quoted and put it first. Nine properties make a passage liftable; the biggest lever is leading with the answer.
Extractability is the third pillar of the AEO Canon: answer engines cite passages, not pages, so you have to write the sentence you want quoted and put it first. Access makes you readable and Alignment aims you at the right question; Extractability makes your answer easy for the machine to lift cleanly.
The AEO Canon · the cascade
Pillar 3 · Extractability — Answer engines cite passages, not pages. Write the sentence you want quoted — and put it first.
Why is the passage the unit of citation?
The passage is the unit of citation because retrieval-augmented engines break the web into chunks, retrieve the most relevant ones, and quote the few that best answer the query. Your page is the container; the passage is what competes. That has a liberating consequence — a single excellent paragraph can be cited even on an otherwise ordinary page — and a demanding one: your best answer has to be written as a standalone unit, not buried in an argument that builds to a conclusion.
So Extractability reframes writing. You are not composing an essay; you are placing self-contained answers an engine can drop into its response without rewriting. The pull quote, not the book.
Where on the page should the answer go?
The answer should go first — in the opening sentence under a question-shaped heading. The evidence is blunt.
Profound found 44% of ChatGPT citations come from the first third of a page, so a buried answer is usually a missed citation. And SE Ranking's analysis points to a 120–180 word passage as the sweet spot — long enough to fully answer one question, short enough to lift whole. Lead with the claim, size the passage right, and head it with the question.
What are the nine properties of an extractable passage?
An extractable passage has nine properties, and the through-line is that it must make complete sense quoted alone.
- 1
Answer-first
The opening sentence is a complete answer, not a windup.
- 2
Self-contained
It stands alone — no orphan pronouns, no 'as above,' no dependence on surrounding text.
- 3
Specific
Concrete claims and numbers, not vague generalities.
- 4
Evidenced
A statistic, source, or quotation in the same passage (this overlaps with Credibility).
- 5
Single-purpose
One question per passage, so its meaning — and its embedding — is sharp.
- 6
Well-headed
Sits under a heading phrased as the exact question it answers.
- 7
Right-sized
Roughly 120–180 words — complete but liftable.
- 8
Plainly written
Clear language a model can map cleanly to the query (and a human can read).
- 9
Structurally clean
Semantic HTML, real headings — easy to parse and chunk.
How do you apply Extractability?
Apply Extractability by rewriting your most important passages answer-first and self-contained — then testing whether each could be quoted out of context.
Apply the Extractability pillar
0 / 5
Each unchecked box is a place a competitor can beat you to the AI answer.
A useful test: copy a passage out of the page and read it cold. If it still answers a clear question completely, it's extractable. If it leans on the paragraph before it, rewrite it.
What are the most common Extractability mistakes?
The most common Extractability mistake is burying the answer — making the reader (and the engine) wait.
The buried answer
"There are many factors to consider when choosing a CRM, and every business is different. Before we dive in, it's worth understanding how CRM software has evolved…" — this tells the engine nothing. Compare: "The best CRM for a small agency is usually a lightweight tool like X or Y, because…" If your first sentence could open any article, it isn't answer-first yet. Other common misses: sprawling sections that answer several questions at once (muddy to retrieve), and orphan pronouns that break when the passage is lifted.
Where Extractability fits in the Canon
Extractability completes the Foundation layer: readable (Access), aimed right (Alignment), and now liftable. With the foundation solid, the next layer is Reputation — does the web vouch for you? — beginning with Authority.
For the mechanics underneath this, see what content format AI cites most and how AI engines choose what to cite. The full framework is The AEO Canon.
Frequently asked questions
- What does extractability mean in AEO?
- Extractability is how easily an answer engine can lift a self-contained passage from your page and present it as the answer. Because engines cite passages, not whole pages, the unit you optimize is the passage — a complete, quotable answer that makes sense out of context. The single biggest lever is leading with the answer in the first sentence.
- Where should the answer go on the page?
- First. Profound found that 44% of ChatGPT citations come from the first third of a page, so a buried answer rarely gets lifted. Lead each section with a complete, self-contained answer under a question-shaped heading, then support it — the inverted pyramid, applied to AI.
- How long should a citable passage be?
- Roughly 120–180 words. SE Ranking's analysis points to that range as the sweet spot — long enough to fully answer one question, short enough to lift whole. Place it directly under a heading phrased as the question it answers.
- What makes a passage easy for AI to quote?
- Nine properties — answer-first, self-contained, specific, evidenced, single-purpose, well-headed, right-sized (120–180 words), plainly written, and structurally clean. The through-line is that the passage should make complete sense quoted alone, with no orphan pronouns or dependence on surrounding text.
Last updated .