Short answer: AI search engines cite the pages they can retrieve and quote. A page gets cited when the engine's crawler is allowed to fetch it, it sits in the index the engine searches, it matches the question closely, and it contains a passage that answers the question on its own. Brand reputation matters, but only after those four gates are passed.
Is there a published formula?
No. OpenAI, Google, Perplexity, Microsoft and Anthropic do not publish how they rank sources inside an answer, and anyone who claims to know the exact weights is guessing. What is documented is the plumbing: which crawler feeds which product, and what makes a page eligible. That is enough to work with, because most brands that are never cited fail at the plumbing, not at the subtle ranking stage.
What are the four gates a page has to pass?
1. The right crawler must be allowed in
Each answer engine reads the web through specific user-agents. ChatGPT search uses OAI-SearchBot; OpenAI states that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers. Perplexity uses PerplexityBot, which it says surfaces and links websites and is not used for model training. Claude's search uses Claude-SearchBot. Google's AI Overviews are part of Google Search, so they rely on Googlebot.
The common mistake is blocking the wrong bot. Blocking GPTBot keeps your content out of OpenAI's model training but does not remove you from ChatGPT search. Blocking OAI-SearchBot does. Our AI crawler list separates answer-engine crawlers from training-only ones for exactly this reason.
2. The page must be in the index being searched
Answer engines do not read the live web for every question. They search an index first and then read a handful of results. If your page is not indexed, because of a noindex tag, a wrong canonical, a redirect chain or a page that returns an error, it cannot be retrieved. For Google AI Overviews, Google says a page must be indexed and eligible to be shown with a snippet. Nothing more exotic is required.
3. The page must match the question
Retrieval works on meaning, not just keywords, but it still needs the topic stated plainly. A page titled "Solutions" with a vague hero line is hard to match to "best invoicing app for EU freelancers". A page whose title, headings and first paragraph say what it is about is easy to match. This is classic on-page SEO, and it has become more important, not less.
4. The page must contain a quotable passage
An engine that has retrieved your page still has to lift something from it. Passages that get quoted tend to be self-contained: a short paragraph that answers one question completely, names the subject instead of saying "it", and includes a concrete fact. A 900-word introduction that reaches the answer in the eighth paragraph gives the engine nothing clean to quote.
What about brand authority and mentions?
Authority still matters. When several pages pass all four gates, engines tend to prefer sources that are consistent and widely referenced: a brand that is described the same way on its own site, in directories and in third-party articles is easier to trust than one with contradictory descriptions. Mentions on other sites also create more pages that can be retrieved for questions about you.
But authority cannot compensate for a failed gate. A well-known brand whose pricing page is blocked to OAI-SearchBot will not be quoted on pricing questions in ChatGPT search, however famous it is.
How does this compare with classic SEO?
| Stage | Classic search | AI answer engines |
|---|---|---|
| Access | Googlebot allowed in robots.txt | The engine's own crawler allowed |
| Index | Page indexed in Google | Page indexed in the engine's source index |
| Match | Title, headings, content relevance | Same, plus a clear statement of topic |
| Selection | Ranking on the results page | A passage quoted inside the answer |
| Outcome | A click from the results page | A citation link, sometimes a click |
The first three rows are the same work. The last two change what "good content" means: a page can rank well and still offer no passage worth quoting.
What can you check on your own site today?
- Confirm your robots.txt allows the answer-engine crawlers you care about. Our free checker does this in seconds.
- Make sure key pages are indexable: no stray
noindex, correct canonicals, 200 status. - Rewrite the opening of your key pages so the first paragraph answers the main question directly.
- Add one self-contained passage per question you want to be cited for.
- Check that your brand is described consistently on your site and your main profiles.
- Run a single page check to see structure, schema and answer-readiness findings for one URL.
Frequently asked questions
Do AI engines prefer big brands?
They prefer sources they can retrieve and trust. Big brands pass those tests more often because they are indexed widely and mentioned often, but a small site with a clear, quotable page can be cited for a specific question.
Does structured data make AI engines cite you?
No provider says schema markup guarantees citation. Structured data helps machines understand what a page is about, which supports matching and trust. It does not replace a quotable passage.
How do I know whether I am being cited?
You have to ask the engines. Run the questions your customers ask in ChatGPT, Perplexity and Google, and record which sources appear. ProvenVisible's tracking does this for a fixed set of prompts when you start a run; you can also do it by hand.