Get named when somebody asks an AI who to hire.

AI search optimisation is the work of getting a business named in answers written by generative engines. It goes by several names: generative engine optimisation, GEO, LLM SEO, answer engine optimisation. They describe the same job. An engine reads a set of pages, decides which ones answer the question, and writes a reply that cites some of them. The work is making sure yours is one it can read, trust and quote.

AI search optimisation is not a replacement for search engine optimisation, and I will not sell it as one. Most of the work is the same work with the emphasis moved: crawlable pages, consistent listings, clear structure, real content, references from other sites. The real difference is how you write a page, because an engine lifts passages rather than ranking whole documents.

Werehouse Media includes AI search work in every search engagement rather than selling it as an extra. Charging twice for one body of work would be a pricing tactic, not a service.

How do AI search engines decide who to name?

Generative engines decide in two stages. The first is retrieval: the engine assembles candidate pages, either from an index it has already crawled or from a live search it runs the moment you ask. A site outside that pool cannot be named, whatever else is true about it.

The second stage is selection. The engine writes its answer from those candidates, and it favours pages that state a specific, checkable claim it can lift cleanly. Vague pages lose here even when they rank well.

The engines do not share one index, which is why "optimise for AI" is not a single job:

The cheapest useful thing most UK small businesses can do takes ten minutes and costs nothing: submit the site to Bing Webmaster Tools. Two of the five engines above depend on Bing, and most sites I audit have never submitted to it.

Does any of this actually work, or is it guesswork?

Some of AI search optimisation is genuinely measured rather than guessed. In 2024 a team from Princeton and IIT Delhi published a controlled study at KDD, the ACM's main data mining conference, testing nine content changes across roughly 10,000 queries. That paper coined the term generative engine optimisation and measured what actually moved visibility in AI answers.

Three changes stood out, each worth up to 40% more visibility on their measure:

• Citing credible sources. Linking claims to something verifiable.
• Adding statistics. Specific figures rather than adjectives.
• Adding quotations. Attributed words from a named person or body.

Two results matter as much for what they rule out. Writing in a more authoritative, confident, brand-like voice produced no significant improvement. Keyword stuffing produced little to none. A generation of SEO habits does not carry over, and the authors say so themselves:

“Our results show that techniques effective in search engines may not translate to this paradigm, and that content-level changes can substantially improve visibility in generative engine responses.”

Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, GEO: Generative Engine Optimization, KDD 2024.

The finding that matters most to a small business is about rank. When the researchers optimised every source in a pool, the gains landed on the pages ranking lowest in classic search. The source sitting at rank five gained 115% from citing sources, while the source at rank one went backwards. The authors explain it simply: classic search leans on backlinks and domain age, which a small firm cannot easily build, whereas generative engines read the page.

The caveat belongs in the same breath as the finding. This was one study, one engine configuration, in 2024, run on a benchmark rather than on your website. The results are evidence, not a forecast, and anyone quoting them as a promise is overselling. The paper carries DOI 10.1145/3637528.3671900.

How do I get cited by ChatGPT?

In the order that actually pays, which is not the order most people start in:

The first three steps cost almost nothing and have nothing to do with clever writing. That ordering is deliberate: the expensive work returns very little until the cheap work is done.

What does it cost?

Nothing extra. AI search work is part of every search engagement here, at the prices already published: a search and AI visibility audit at £495, or local foundations as a one-off project at £995. There is no monthly retainer. Every new website build includes the groundwork as standard.

Splitting the price out would be dishonest, because the technical half of AI visibility is the technical half of SEO. Selling one crawl audit twice under two names is how this industry earned its reputation.

Can you guarantee I will appear in ChatGPT?

No, and be careful with anyone who says otherwise. There is no ranking to buy, no submission form, and no relationship that gets a business listed. Ask the same assistant the same question twice and you can get different names, so there is no position to occupy.

Measurement is genuinely poor at the moment. There is no equivalent of Search Console for AI answers. What I do is unglamorous and honest: write down ten questions a customer would actually ask, put them to ChatGPT, Gemini and Perplexity by hand each month, and record who gets named. It is a small sample and I say so. It is still better than the dashboards being sold.

Why hire a UK studio for this?

Mostly because the alternatives are not in the UK. Search for an AI SEO agency and the results resolve to Kuala Lumpur, Mumbai, Singapore, Perth and a handful of tool vendors. None of that is a criticism of their work. It does mean a British business wanting someone in the same timezone, writing for a UK audience and citing UK regulators, has slim pickings.

The category being this new also means it is open. I would rather say that plainly than claim a decade of expertise in a field that was named in 2024.