answer engine optimization: what actually makes a page citable

answer engine optimization, generative engine optimization, and llm seo are three names for one job: making a page that an AI assistant can retrieve, quote without mangling, and attribute to you. most of that job is ordinary technical seo done properly. a small part of it is genuinely new. this post separates the two, because the gap between them is where most of the advice in this category is being sold.

write passages, not just pages

a retrieval system pulls a chunk of your page, not the whole thing, and the model sees that chunk without the surrounding argument. so a paragraph that begins "this is why it matters" is useless when lifted, because the referent is gone. a paragraph that begins "a referring domain is one linking site, however many pages on it link to you" survives the trip.

the practical test is to read any paragraph on its own with the rest of the page covered up. if it still states a complete, attributable claim, it is citable. if it depends on the sentence before it, rewrite the opening clause. this is also just better writing, which is why it is the one habit here worth adopting regardless of what assistants do next.

• answer in the first paragraph. then expand. the inverted pyramid was built for exactly this.
• put numbers in text. a figure that exists only inside a chart image is invisible to retrieval and to screen readers.
• say what you measured and what you inferred. a claim with its provenance attached is safer to repeat, and systems that summarise are increasingly picky about that.

llms.txt: worth publishing, not worth believing in

llms.txt is a plain-text file at the root of your site describing what the site is and what its important pages cover. we publish one, at /llms.txt, alongside a longer /llms-full.txt. here is the shape:

# CrawlGraph

> one-sentence description of what this site is and who it is for.

## Core pages

- [Gap analysis](https://example.com/gap-analysis): what the page does,
  stated as a capability rather than a slogan.
- [API documentation](https://example.com/docs/api): what the free tier
  covers and where it stops, in the same words the page uses.

## What this tool does not do

- rank tracking, keyword research, live re-crawl.

now the honest part: no major assistant vendor has committed to honouring it, so nobody can tell you it works, and anyone who does is guessing. we publish it because it costs an afternoon and because the writing exercise pays for itself. stating in one line what each page is for, in the same words the page uses, surfaces the pages you cannot describe. those are usually the pages that are cannibalising something else on your own site.

one detail that matters more than the file: the section listing what the tool does not do. being explicit about scope is what stops a summary attributing a capability to you that you do not have, and a confidently wrong citation is worse than none.

structured data, and the rule that matters

schema markup is not a requirement for being quoted, and it is not the lever the category sells it as. it is cheap, it removes ambiguity, and it has exactly one rule worth enforcing: the markup must say what the visible page says. markup that disagrees with the rendered text is a manual-action risk with search engines, and it hands an assistant two versions of your claim with no way to choose.

we enforce that structurally rather than by discipline. every faq on this blog is declared once as data and read twice: the visible question-and-answer section and the FAQPage json-ld are generated from the same array, so they cannot drift apart. that is a small architectural decision that removes a whole class of error, and it is the kind of thing worth copying.

be retrievable in the first place

none of the above matters if your page is not a candidate. retrieval happens over crawled pages and the link structure between them, which is the same raw material a search index is built from. that has two consequences worth stating plainly.

first, the technical basics are load-bearing rather than hygiene: server-rendered html, a page that does not require javascript to show its own text, honest canonicals, and no accidental noindex. second, credible inbound links still do the work of separating you from the millions of other pages that mention your subject. we have written up what the link data actually shows about that in whether seo is dead, answered from the link graph, which is the strategic half of this question rather than the mechanical half you are reading.

the underrated one: ship an interface, not just pages

the aeo advice nobody gives is that assistants can do more than read your marketing pages: they can call your product. we publish an mcp server, crawlgraph-mcp on npm, which lets an assistant run a backlink lookup or a gap analysis directly and get structured results back rather than prose scraped off a page.

this is not a ranking trick and we would not claim it lifts anything in search. it is a distribution channel with different economics: a person asking their assistant to check a domain's backlinks can be served by your product inside the conversation, with no page view and no click involved. if you sell something an assistant could invoke, that interface is worth more attention than another checklist post about schema.

what to ignore

• anyone quoting a weighting. no vendor publishes how heavily its retrieval layer uses any signal. specific percentages in this category are invented.
• submission services. there is no index to submit to and no allowlist to buy onto.
• "aeo is a new discipline" pricing. six of the seven items above are ordinary seo. the pitch that it is a separate practice is usually the product.

faq

What is answer engine optimization?

Answer engine optimization is the practice of making a page easy for an AI assistant to retrieve, quote correctly, and attribute to you. It overlaps almost entirely with ordinary technical SEO and good writing, with one shift in emphasis: the unit that gets used is a passage rather than a page, so a claim that reads correctly when lifted out of its surroundings is worth more than a page that only makes sense read top to bottom.

Is answer engine optimization different from SEO?

Mostly no, and treat anyone selling it as a separate discipline with suspicion. The retrieval systems behind AI answers are built on crawled pages and the links between them, so crawlability, clear structure, and credible inbound links carry over unchanged. The genuine differences are narrow: passages are retrieved rather than whole pages, citation replaces the click as the outcome, and a page that only states a claim in a chart or an image cannot be quoted at all.

Does llms.txt actually do anything?

No search or assistant vendor has committed to honouring llms.txt, so treat it as unproven rather than required. It is cheap to publish and it has one immediate benefit that does not depend on adoption: writing a plain-text summary of what your site is and what each important page covers forces you to state your own scope in a way that a machine could use, and that exercise usually exposes pages whose purpose you cannot articulate.

How do I get cited by ChatGPT or Perplexity?

There is no submission process and anyone offering one is selling. Being cited follows from being retrievable and being worth quoting: publish the answer in crawlable HTML rather than only in an image or a PDF, state the claim in a self-contained sentence near the top, show the working behind any number you assert, and earn links from domains already trusted in your subject so retrieval ranks you among the credible candidates.

Do I need structured data for AI search?

It is not required for a page to be quoted, but it is cheap and it removes ambiguity about what your page claims. The rule that matters more than any schema type is that the structured data must say exactly what the visible page says. Markup that disagrees with the rendered text is a manual-action risk with search engines and gives an assistant two conflicting versions of your claim to choose between.