Get cited and recommended by ChatGPT, Gemini and AI Overviews.

A large language model, the technology behind ChatGPT, Gemini and Copilot, reads a question, gathers the sources it trusts, and writes one synthesised reply. Your goal is to be a source it pulls from and a brand it recommends. That is a different job from classic SEO, and it has its own name in two halves: answer engine optimisation (AEO), getting picked as the direct answer, and generative engine optimisation (GEO), getting cited inside a generated response.

True SEO does the strategy and the implementation for both. We make your site the most complete, most quotable resource on the topics your buyers ask about, then we wire in the technical signals an assistant needs to read you cleanly and the off-site mentions it uses to trust you. The method that earns it is the same content-first approach behind our semantic SEO and topical authority service, applied to a new set of surfaces.

A citation can look like a named recommendation (“firms such as yours”), a linked source under an AI Overview, or a brand mention with no link at all. All three move buyers. The last one matters more than people expect, because an assistant that praises your expertise has already done your selling before anyone reaches your site.

Ignoring AI search costs you the click before the click, the moment a buyer forms a shortlist without ever reaching a results page. Pew Research Center found that when an AI summary appears, people click through to a website in just 8% of visits, compared with 15% when no summary shows, and only 1% click the sources cited inside the summary. Close to a third of UK Google searches now return an AI summary above the normal results, so this is mainstream behaviour, not a fringe case.

The traffic hit is real for the pages that drop below those answers. Ahrefs, measuring across a very large sample of searches, found AI Overviews cut click-through rates by about 34.5%. Authoritas analysis reported by The Guardian put the loss as high as 79% for sites that slipped beneath an Overview. Lose the answer slot and you lose the visit, the enquiry and the sale that followed it.

Picture a four-partner accountancy practice in the Midlands. It held page-one rankings for local tax searches, then watched form fills fall by a quarter over two quarters with no change to its own pages. The cause sat above its listing: an AI summary answering the question and naming two rivals. Nothing about the firm got worse. The route buyers took simply moved, and the firm was not built for the new one.

You measure AI visibility by testing real buyer prompts, tracking how often your brand is named, and watching AI referral traffic in your analytics. We run a fixed set of the questions your customers ask across the main assistants, then log whether you appear, whether you are recommended, and which competitors share the answer. That gives a share-of-voice figure you can move month on month.

We pair that with two harder signals. Your analytics now separates referrals from ChatGPT, Perplexity and other assistants, so we report visits and enquiries from AI sources directly. And we watch the AI Overview and AI Mode appearances for your priority queries, since being cited there is visibility even when the click stays on Google. AI referral traffic is still a small slice of the total for most sites, but it is growing faster than any other channel, and it tends to arrive better informed.

Reporting matters as much as measuring. A board wants to see which prompts you win, what changed, and what enquiries followed, not a screenshot of a tool. We build that into the monthly review so the AI work answers to the same standard as the rest of your SEO.

You correct a wrong AI answer by repairing the sources the model reads, since you cannot edit the model itself. Assistants repeat what their training data and the live web tell them, so a stale phone number, an old service list or a confused description usually traces back to inconsistent information across the places they trust. We audit those sources, align your name, address, services and descriptions across your site, Google Business Profile and the major directories, and remove the contradictions a model trips over.

Authoritative corroboration speeds it up. Getting accurate facts onto sources a model weights heavily, your own well-structured pages, recognised directories and reputable third-party mentions, gives the assistant a clean version to prefer over the outdated one. Fresh, dated content helps here too, because a recently updated page signals which version is current.

This is slower than flipping a switch, and we are honest about that. Models refresh on their own schedule, so a correction shows first in assistants grounded on live search, then in the rest as they update.