AI search optimisation makes a business easier for ChatGPT, Perplexity, Gemini, Microsoft Copilot and Google AI Overviews to find, understand and cite. As an independent AI SEO consultant in the UK, I work on four things: consistent information about your business, content that answers questions directly, search-engine access (Bing as well as Google) and regular measurement of how AI answers describe you.
By George St. Georgiev, SEO consultant · Last updated 06/11/2026
What AI search optimisation is — and how GEO and AEO differ
AI search optimisation is the work of making your business a reliable source for AI-generated answers. It has two parts. Answer engine optimisation (AEO) shapes content so a system can lift a direct answer from it. Generative engine optimisation (GEO) makes sure AI models recognise your business, trust it and mention or cite it when they write an answer.
| Classic SEO | AEO | GEO | |
|---|---|---|---|
| Goal | Rank a page in the search results | Be the quoted answer to a question | Be named or cited in an AI-written answer |
| Where | Google and Bing results pages | Featured snippets, voice answers, AI Overviews | ChatGPT, Perplexity, Gemini, Copilot, AI Overviews |
| How it’s measured | Rankings, clicks, impressions | Snippet and answer ownership | Mentions, citations and accuracy across a fixed set of questions |
The three overlap. A page that is technically sound, clearly structured and from a recognisable source does well in all of them, which is why I treat them as one piece of work.
Why it matters now
People increasingly ask AI tools questions they used to type into Google. OpenAI said in February 2026 that ChatGPT had around 900 million weekly users. Google said in July 2025 that AI Overviews reached more than 2 billion users a month. When an AI answer recommends three firms and yours isn’t one of them, you never see the lost enquiry in your analytics.
There is also a practical link many businesses miss: ChatGPT’s web search relies heavily on Bing’s index, alongside OpenAI’s own crawler. A site that Bing hasn’t indexed properly is harder for ChatGPT to find and cite, however well it ranks on Google.
What I do — the five workstreams
1. Entity and brand consistency
AI models build a picture of your business from everything they can read about it. I make that picture consistent: the same name, description, services, people and contact details on your site, in your structured data (Organization or Person schema with links to your official profiles), on Google Business Profile, LinkedIn and the directories that matter in your sector. Conflicting details are one of the most common reasons an AI describes a business wrongly or leaves it out.
2. Answer-ready content
I restructure key pages so each question is answered in the first two or three sentences under its heading, in a self-contained passage of around 40–60 words, followed by the detail. Tables, steps and definitions are used where they make the answer clearer. This helps AI systems quote you accurately, and readers get to the point faster too.
3. Crawl and index access
AI systems can only use pages they can reach. I check that robots.txt doesn’t block the AI crawlers you want, that key content is in the server-rendered HTML rather than loaded only by JavaScript, and that the site is verified in Bing Webmaster Tools with IndexNow set up, so new and updated pages reach Bing quickly. Where it’s useful, I add an llms.txt file that points AI tools to your most important pages.
4. Mentions and citations
AI answers lean on sources other than your own site: industry lists, directories, review platforms, trade press and community discussions. I identify which third-party sources the AI engines cite for your topic, then work on getting your business accurately represented on them through corrections, listings and genuine outreach. I don’t buy links or post fake reviews.
5. Measurement
Every engagement starts and continues with measurement, described below. Without a baseline, “AI visibility” is just a claim.
How I measure AI visibility
I measure AI visibility by asking the same fixed set of questions, in the same way, across the main AI engines, and recording what each answer says about you and your competitors. Repeating the same questions over time shows whether the work is changing anything.
- Build the question set. 20–50 questions your customers actually ask, such as “best mortgage broker for contractors in the UK” or “how much does a technical SEO audit cost”.
- Record a baseline. For each question and engine: is your business mentioned, is your site cited, which competitors and sources appear, and is the description accurate?
- Make the changes. Entity fixes, content restructuring, indexing and citation work.
- Repeat on a schedule. Monthly checks against the same questions, with each result dated.
Example of a measurement log
| Question | Engine | Mentioned? | Your page cited? | Other sources cited | Date |
|---|---|---|---|---|---|
| “SEO consultant for mortgage brokers UK” | ChatGPT | No | No | 2 agencies, 1 directory list | DD/MM/YYYY |
| same | Perplexity | Yes | Yes | 1 agency | DD/MM/YYYY |
Illustrative format only, not real client data.
AI answers vary from one session to the next. One screenshot proves little; the pattern across dozens of questions, checked every month, does.
What I can’t promise
- No guaranteed placement. No one controls what ChatGPT, Gemini or Google’s AI writes. Anyone promising “guaranteed AI rankings” is selling something they can’t deliver.
- Answers change. Models are updated and retrieval sources shift, so visibility has to be measured repeatedly.
- It doesn’t replace SEO. Most AI engines rely on search indexes. A site with technical problems or thin content will struggle in AI answers too.
- Results take time. Third-party mentions and entity signals build up over months, not days.
Results
The case studies on this site show the SEO foundations AI visibility depends on: Wooof (monthly organic sessions up 40%, June 2024 – March 2025) and Pro Drillers BG (organic traffic up 45% within six months of relaunch). For AI-specific work I report against the measurement log above, so you see the before-and-after for your own questions.
Who it’s for
- Local and service businesses whose customers now ask AI tools for recommendations, such as “who’s a good … near me”.
- Regulated firms, including mortgage brokers, where an inaccurate AI description is a reputational risk. See AI search for mortgage brokers.
- Online shops and specialist brands competing with larger names that AI answers mention by default.
Questions
Is GEO replacing SEO?
No. Generative engine optimisation builds on SEO rather than replacing it. AI engines such as ChatGPT, Copilot and Google AI Overviews retrieve pages from search indexes before writing an answer, so a site that isn’t crawlable, indexed and trusted in search is unlikely to be cited. GEO adds entity consistency, answer-ready content and third-party mentions on top.
How long does AI search optimisation take?
Technical and indexing fixes, including Bing verification, can take effect within weeks of the pages being recrawled. Changes in how AI answers describe your business usually take two to six months, because they depend on third-party mentions and model or index updates. I measure monthly so you can see the direction early.
Do I need an llms.txt file?
An llms.txt file is optional. It is a plain-text file that points AI tools to your most important pages, and it costs little to add. It isn’t an official standard, and the main AI engines don’t publicly confirm they use it, so it should never replace crawlable pages, structured data and Bing indexing.
Can you get my business into ChatGPT’s answers?
I can make your business much easier for ChatGPT to find, understand and cite, and measure whether it is mentioned. No one can guarantee a mention, because OpenAI controls how answers are generated. The work focuses on what you control: your site, your entity data, Bing indexing and accurate third-party sources.
What’s the difference between GEO and AEO?
Answer engine optimisation (AEO) shapes content so a system can quote a direct answer from it, as in a featured snippet or voice answer. Generative engine optimisation (GEO) aims to get your business recognised, mentioned and cited in AI-written answers from tools like ChatGPT and Perplexity. In practice they share most of the same work.
How is AI search optimisation priced?
Pricing is on request after a free consultation. Initial setup, including the baseline, entity and indexing fixes and content plan, is quoted as a fixed project. Ongoing measurement and citation work is a monthly fee agreed in advance. See how pricing works.
Get an AI visibility check
Tell me your business and five questions your customers ask. I’ll show you how the main AI engines answer them today.
Get an AI visibility check · All services · Back to home
Sources
- eWEEK — ChatGPT nears 1 billion weekly users (reporting OpenAI, 27/02/2026)
- TechCrunch — Google’s AI Overviews have 2B monthly users (Alphabet Q2 2025 results, 23/07/2025)
- Search Engine Journal — ChatGPT search and Bing
- Yoast — How ChatGPT search uses Bing data
Last updated: 06/11/2026 · Reviewed every 3 months