How to Get Cited by ChatGPT, Perplexity and Google AI Overviews

When someone asks an assistant for a recommendation, the assistant does not rank pages the way a search results page does. It assembles an answer and cites the sources it used. Those two things come apart: a page can rank in the ordinary top ten and never be quoted, and a page that ranks poorly can be quoted constantly because it answers the question cleanly.

This guide is about the second thing. It covers what answer engines actually look at, what makes a passage quotable, and the work that changes your odds — in the order it is worth doing.

Answer engines retrieve, then summarise

Every major assistant follows roughly the same shape. Your question becomes a set of searches. Those searches return candidate pages. A model reads the candidates and writes an answer, attaching citations to the passages it drew from. Anything that was not retrieved cannot be cited, and anything that was retrieved but is hard to lift a clean sentence from will be skipped in favour of a competitor that is easier to quote.

That gives three distinct failure points: you are not retrieved, you are retrieved but not understood, or you are understood but not quotable. Most advice online only addresses the first one. In practice the third is where most businesses lose.

Be retrievable

Let the crawlers in

Check your robots.txt actually allows the crawlers that feed assistants, not just Googlebot. GPTBot and OAI-SearchBot for OpenAI, PerplexityBot, ClaudeBot, Google-Extended for Gemini and AI Overviews, and Applebot-Extended for Apple Intelligence. A file that blocks them makes every other step pointless. Ours lists them explicitly.

Answer in the language people use

Assistants match the phrasing of the question. If every page on your site says "digital solutions" and buyers ask "who can run my Google Ads", the words do not meet. Write the service names people actually type, and keep the internal jargon for your sales deck.

Cover the question, not just the keyword

A page targeting "local SEO services" competes with a page titled "what does a local SEO agency do and what does it cost". The second one answers a question and gets quoted. The first one announces a service and gets ranked, if you are lucky.

Be understood as an entity

An assistant has to be confident it knows who you are before it will put your name in an answer. That confidence comes from consistency across the sources it reads, not from any single page.

• One name, spelled the same way everywhere. If your site says "Wildrank Technologies" and a directory says "WildRank Tech", you are two weaker entities instead of one strong one.
• One address and phone number, matching character for character across your site, your profiles and your listings. Mismatches read as uncertainty.
• Structured data that states the facts machine-readably: legal name, founding date, headquarters, and the services you offer. A crawler should not have to infer your industry from prose.
• A consistent entity across profiles. Link your site from every profile you control, and link those profiles back from the site. This is what sameAs is for.

Be quotable

This is the part most sites get wrong. A model needs a self-contained sentence it can lift and cite. Long introductions, throat-clearing and "in today's fast-paced digital landscape" give it nothing to take.

• Put the answer in the first sentence under the heading. Not after context, not after a story. The heading asks the question; the sentence answers it.
• Make each section stand alone. Assume the reader arrives mid-page. If a section only makes sense after the previous four, it will not be quoted.
• Use concrete nouns and numbers over adjectives. "We respond within one business day" is quotable. "Lightning-fast turnaround" is not.
• Define your terms once, plainly. If you have a specific way of doing something, name it and describe it in one sentence a stranger could repeat.

What does not work

• Keyword-stuffed pages written for a crawler. They are retrieved less often and quoted almost never, because there is no clean sentence in them.
• Claiming to be the best. Every competitor claims it, so it carries no information. Specifics carry information.
• Blocking AI crawlers as a protest. You keep your content and lose the citation; the assistant answers from someone else's page.
• Padding an article to hit a word count. Length is not the signal — the number of self-contained answers is.

The order to do this in

1. Fix robots.txt so the AI crawlers are allowed. Ten minutes, and nothing else matters without it.
2. Resolve the entity: consistent name, address and phone across site and profiles, with sameAs links both ways.
3. Add structured data for the organisation and each service, so facts do not have to be inferred.
4. Rewrite the top of your money pages so each section opens with a direct answer.
5. Publish content that answers the questions buyers actually ask, in their words.
6. Re-check monthly. Assistants change what they cite, and a citation you had is not a citation you keep.

None of this is separate from ordinary good SEO. It is the same work with a different reader in mind — one that reads a page looking for a sentence it can reuse.