How to get cited by AI

1. Answer the question first, in plain words

Engines lift the clearest sentence, not the cleverest. Put the direct answer up top, then add depth.

• Lead each key page with the question as a heading and the answer directly beneath it.
• Use short lists and tables. They are easy for a model to extract and quote.
• Write the way people ask, not the way brochures sell.

2. Mark it up with structured data (schema)

JSON-LD tells the engine what a page is and how the facts relate. It makes you legible.

• Organization with sameAs: your identity and the profiles that confirm it.
• Service: what you actually sell.
• FAQPage: your answers, in a form engines quote directly.
• Define one Organization entity, referenced by @id, and reuse it sitewide so every page points at the same business.

3. Keep one consistent identity (entity clarity)

The engine has to be sure every mention points at one business. Ambiguity is uncited.

• Same name, same description, same category on your site, LinkedIn, Crunchbase and X.
• A split story, 'automation studio' in one place and 'AEO studio' in another, makes the engine unsure, and unsure means uncited.
• Wire the profiles into your Organization schema with sameAs so the engine can join them up.

4. Get corroborated by trusted sources (surround-sound)

Engines cite brands that other trusted sources already name. For a newer brand this often matters more than any on-page change.

• Get into the roundups and 'top X' lists the engines regurgitate for shortlist questions.
• Publish a founder article that makes the same argument and links back to your reference pages.
• Keep clean, consistent listings on chambers, associations and directories, each repeating the same one-line description.

Do not block the bots that cite you

Confirm robots.txt allows OAI-SearchBot, GPTBot, ChatGPT-User, PerplexityBot, Perplexity-User and Google-Extended, alongside Googlebot and Bingbot. Blocking them is the most common way businesses accidentally opt out of AI citation entirely.

Render your content server-side

The answer, the headings and the schema must be in the initial HTML, not injected by client JavaScript after load. Crawlers and AI fetchers may never run your scripts, so anything that appears only after render is, to them, not there.

A concrete example

Ask an engine 'how much does a dental implant cost in Bangkok?'. It wants a page whose first line answers exactly that, marked up as an FAQPage, on a site it recognises as the clinic that does the procedure. Give it all three and you are quotable. Miss one and a competitor gets the citation.