The shift: from searching to asking
For twenty years, a buyer's first move was a search — and the result was a page of links where ten brands got a chance at the click. That behaviour is being replaced. Buyers now ask AI engines a question in plain language — “which hospital should I trust for cardiac surgery?”, “which private bank is best for wealth management?” — and the engine returns one synthesised answer naming a small number of brands.
There is no page two of an AI answer. If your brand isn't in it, you were never considered — and because the conversation happened inside ChatGPT or Gemini rather than on your website, nothing appears in your analytics. The loss is real and invisible at the same time.
The definition
AI SEO — formally, Generative Engine Optimisation (GEO) — is the discipline of structuring a brand's credibility, content and evidence so AI engines can read, trust and cite it in their answers.
The five engines that matter today are ChatGPT, Perplexity, Google AI Overviews, Gemini and Claude. Each composes answers differently, but all of them decide what to cite the same way: from sources they can parse and consider credible.
How engines choose what to cite
When an engine answers a trust question, it looks for three things. First, machine-readable structure — schema markup, clear entities, and pages organised so the engine can extract facts without guessing. Second, verifiable evidence — credentials, certifications, third-party corroboration the engine can cross-reference. Third, answer-shaped content — material that actually addresses the questions buyers ask, rather than the keywords they used to type.
Notice what isn't on that list: ad spend. There is no paid placement inside an AI answer. Citations are earned or they are absent — which is why no honest firm guarantees them.
Why strong SEO isn't enough
A common assumption: “we rank well on Google, so we'll be fine in AI answers.” It doesn't transfer automatically. Traditional SEO optimises signals — keywords, backlinks, page speed — that engines used to rank links. Generative engines synthesise answers, and they weight different things: whether your claims are structured enough to extract, whether your credentials are verifiable, whether third-party sources corroborate you. A brand can dominate search results and be absent from AI answers in the same category. The reverse is also true — which is the opportunity.
What the work actually involves
A serious AI SEO engagement has four movements. A baseline: recording exactly how each engine answers about you today. A diagnosis: mapping the questions where you're absent, weak or out-positioned. The build: gap-targeted content plus the technical and entity structuring engines require. And proof: continuous tracking across all five engines, so every change — or its absence — is visible. Anything sold without a baseline and tracking is faith, not strategy.
Glossary
- AI SEO
- The plain-language name for this discipline. Same thing as GEO.
- GEO — Generative Engine Optimisation
- The formal term: optimising for the synthesised answers of generative AI engines.
- AEO — Answer Engine Optimisation
- A closely related term emphasising engines that return direct answers rather than link lists.
- AI citation
- When an engine names or references your brand in an answer. Earned, never bought.
The logical next step
Before deciding anything, see your baseline: how the five engines answer about your brand today. It's confidential, no-obligation, and every finding is yours to verify.