AI search optimisation (GEO) in London

AI visibility baseline

A fixed set of questions a London customer would ask about what you do, run in ChatGPT, Gemini, Perplexity and Google’s AI Overviews and AI Mode, captured with a date and a screenshot each. Who is named, who is not, and what is said about you.

Entity audit and fixes

Your name, address, categories, descriptions and key facts made consistent across the site, Google Business Profile, Companies House, LinkedIn and the directories AI engines draw on, so there is one version of you to cite.

Structured data that matches the page

Organisation, service, product, FAQ, person and breadcrumb markup written to describe what is visibly on the page, validated and monitored. Never markup for content that is not there.

Citable pages

Clear answers to the questions people ask, with definitions, figures and their sources, author names and dates, in HTML an engine can read. Menus, price lists and brochures moved out of PDFs and into pages.

llms.txt and crawler access

An llms.txt file describing your site for AI assistants, decisions with you on which AI crawlers to allow in robots.txt, and a check that your key pages are rendered on the server rather than hidden behind scripts.

Corroboration off your site

Reviews, directory listings, press and mentions that say the same things your site says, because engines weigh what others say about you more than what you say about yourself.

Monthly AI visibility report

The tracked questions re-run, the change in how often and how accurately you are named, the sources engines cite, and referral visits from AI assistants in GA4, alongside your normal SEO report.

How we do it

It helps to be clear about what these engines do. When you ask ChatGPT, Perplexity or Gemini a question about a London business, the assistant usually runs a search, reads a handful of pages it trusts, and writes an answer with citations. Google’s AI Overviews and AI Mode do the same with Google’s own index. So the work that earns a citation is mostly the work that earns a ranking: a page that answers the question clearly, a business the engine can identify with confidence, and other sources that agree. Our guide to AI search and GEO explains the mechanics without the jargon, and the AI Overviews guide covers what changes for a London business specifically.

The first job is the entity: can an engine tell, with confidence, who you are? We check every public description of your business for disagreements: the trading name against the legal name, the address on the site against Companies House and the Business Profile, the categories, the opening hours, the founders’ names, what you sell. Then we write structured data that describes exactly what is on each page, following the schema guide, and we work with you to fix the off-site sources that are wrong. An engine citing a directory for your prices is a sign the directory was easier to read than your site.

The second job is citable pages. An assistant quotes the page that gives the plainest answer with the least effort: a definition in the first paragraph, a price stated as a price, a figure with its source, a named author and a date, and HTML rather than a PDF or an image of a menu. We rewrite the pages that answer the tracked questions, and create the ones that are missing. The third job is access: pages rendered on the server, an llms.txt file describing the site, and a decision, made with you and written into robots.txt, about which AI crawlers may read it. We will be honest that llms.txt is a convention, not a standard any engine has committed to; it costs almost nothing to publish, so we publish one (this site has one), and we do not claim more for it than that.

Who does it: Muhammad Ali, our SEO Specialist, leads the work. Ayesha Jahangir, Generative AI Expert, and Thomas Edward Bennett, AI Content and Automation Specialist, run the baseline and the monthly captures with him. Momna Imran, Communications and Audit, keeps every recommendation grounded in evidence you can inspect. The whole team is named and pictured.

What changes and when

Technical fixes are usually re-crawled within two to eight weeks. Local search changes often show within one to three months. Competitive London head terms usually take six to twelve months. These are typical ranges from our own work, not a promise for your site, and we report progress every month so you can see which it is.

AI answers add their own uncertainty. The same question asked twice can produce two different answers, so we never judge a single capture; we track the trend across months on a fixed list of questions. Entity fixes and structured data tend to be reflected within the two-to-eight-week re-crawl window in Google’s results and within weeks in assistants that search live. Becoming the cited source for a competitive question takes as long as ranking for it, usually months, because it depends on the same authority. We report what moved, with the screenshots, and we say when something has not moved.

How it is measured

What we report: the share of tracked questions in which you are named, per engine; the accuracy of what is said (facts right, facts wrong, facts missing); the sources each engine cites for you and for the two nearest competitors; referral visits from AI assistants in GA4, by source; and the ordinary Search Console figures for the pages we changed. Where it comes from: dated, screenshotted captures we run by hand, GA4 and Search Console in accounts you own. How often: monthly, as a section of the standard report, with a review every quarter.

What we do not report: an "AI visibility score" from a tool with no stated method, or a share of Google impressions attributed to AI Overviews, because Search Console does not separate them and we will not pretend it does. If an engine your customers do not use starts naming you, we will mention it, but we will not count it as a result.

What it costs to find out

Nothing. The free SEO audit is a written review of your site by an SEO specialist from the team, delivered within two working days of your request. For this service we add one thing: we ask two assistants five of the questions your customers would ask and show you what they say about you today, with the pages they cited. It is often the most useful page in the review, because nobody has looked before.

AI search work is included in the higher managed SEO tiers and can be bought as a standalone project (baseline, entity and schema fixes, the first set of citable pages). What drives the price is the number of questions tracked, the number of pages that need rewriting, how many off-site sources need correcting, and whether we ship the technical changes or your developers do. Pricing is published on the pricing page; our terms are on the before you sign page.

Common mistakes we see on London sites

• The menu, the price list or the brochure as a PDF, or as an image, so the plainest facts about the business are the ones no engine can read.
• Three versions of the business name and two addresses across the site, the Business Profile, Companies House and LinkedIn, so the engine is not confident enough to cite any of them.
• FAQ or review markup for content that is not on the page. Engines and Google both treat it as a reason to trust the site less.
• Blocking every AI crawler in robots.txt because a plugin defaulted to it, then wondering why no assistant mentions the business.
• Pages with no author, no date and no sources, which read as anonymous to a system choosing what to quote.
• "GEO content" that is ordinary keyword stuffing with the words "AI" and "ChatGPT" added. The engines are reading for answers, not for vocabulary.
• Chasing a mention in an engine your customers do not use while the Google result they do use still shows the wrong opening hours.

How it fits with other services

This work sits on top of ordinary search. The crawl, rendering and speed fixes in technical SEO decide whether an engine can read you at all. The citable pages are content work with a stricter brief. The reviews and listings that corroborate your site are reputation management, and the press and mentions that engines weigh heavily are digital PR. On every managed retainer we baseline how AI engines describe you as a matter of course, because the work that earns a citation is the same work that earns a ranking.

Proof, stated plainly

We have not yet published an AI search case study. The field is new, the measurement is by hand, and we will publish a case only when a client agrees and every figure in it can be sourced to a dated capture. Until then the proof is the method on this page and the baseline we show you in the review. Ask us for a reference call; every case we publish names the client, the figure, the source and the date.

Get your free SEO auditA person reviews your site within two working days. Three fields, no call required.