How to Get Your Business Cited by ChatGPT, Perplexity and Gemini

A growing share of commercial research now begins in an AI assistant rather than a search box. The mechanics of being recommended there are related to SEO but not identical, and the differences are worth understanding precisely.

The key difference: passages, not pages

A search engine ranks documents and shows you a list. An answer engine retrieves documents, extracts the passages that answer the question, and writes a response citing its sources. It never shows a ranked list at all.

That single difference drives everything else. A page can be excellent overall and still never be quoted, because its answer is distributed across six sections and no individual passage stands on its own. The unit of optimisation is the passage.

1. Let the crawlers in

This is the step most often missed, and it is binary: a page an AI crawler cannot fetch can never be cited. The relevant user agents include:

• GPTBot and OAI-SearchBot — OpenAI, ChatGPT and ChatGPT Search
• ClaudeBot — Anthropic
• PerplexityBot — Perplexity
• Google-Extended — Gemini and AI Overviews grounding
• Bingbot — Bing, and by extension Copilot

Note that Google-Extended does not affect normal Google Search rankings at all. It controls only whether your content may ground Gemini and AI Overview answers. Blocking it costs you nothing in classic search and removes you from AI citation entirely.

2. Write passages that survive extraction

The practical test: take any paragraph out of your page and read it alone. Does it still make sense? Does it say who is making the claim? If it depends on the previous three paragraphs, it will not be quoted.

What works:

• Answer the question in the first sentence, then explain
• Use headings that are the questions people actually ask
• Give specifics — ranges, timeframes, named technologies — rather than adjectives
• Put comparisons in real tables; they extract cleanly and are quoted often
• Keep each section to one idea

3. Make attribution easy

An answer engine that cannot confidently identify a publisher will often decline to cite it. Make identity unambiguous:

• Organization structured data with a stable identifier, address and contact details
• sameAs links to your verified social and directory profiles
• Named authors with real credentials where an article makes expert claims
• A visible, honest last-updated date

Recency deserves emphasis. Answer engines strongly prefer current sources, and a page with no date at all is treated as unknown rather than fresh.

4. Publish an llms.txt

An llms.txt file at the root of your domain gives AI systems a plain, accurate description of your organisation and links to the pages that support each claim. Support is not universal and it is not a formal standard, but it is inexpensive to publish, easy to keep accurate, and removes ambiguity for any system that does read it.

5. Keep doing conventional SEO

Most answer engines retrieve from a conventional index before generating. If you do not rank, you are not retrieved, and if you are not retrieved you cannot be cited. Crawlability, speed, internal linking and genuine authority remain the foundation. GEO sits on top of that foundation; it does not substitute for it.

What does not work

Hidden text aimed at AI crawlers is cloaking, and it is penalised by search engines. Stuffing prompts or instructions into page content in the hope of manipulating a model is unreliable and reads as spam to human visitors. Fabricated statistics are the worst option of all: once quoted by an assistant and traced back, they damage exactly the credibility the exercise was meant to build.

The durable version of this work is unglamorous. Say true things clearly, in passages that stand on their own, from a source that is easy to identify, on a site that loads quickly and lets the crawlers in.