The honest short version: answer engine optimisation (AEO) is making a page so easy to quote that ChatGPT, Microsoft Copilot, Perplexity or a Google AI Overview picks it as the source for an answer. Generative engine optimisation (GEO) is the same job under a different name. Nobody can guarantee you a citation, because no engine sells one or shows you a position. What you can control is whether the engines can reach your pages, and whether the page answers the question better than anyone else’s.
What is answer engine optimisation, and how is it different from SEO?
Ordinary SEO tries to put your page high in a list of ten links. AEO tries to get a sentence from your page lifted into the answer itself, with your site shown as the source. The two overlap far more than agencies selling “GEO” as a new discipline admit. Every engine we look at below finds pages through ordinary crawling and indexing, so a site that is invisible to Google and Bing is invisible to the answer engines built on top of them.
Where they differ is the unit that wins. A blue link wins on the whole page. A citation wins on one extractable passage: a heading that matches the question, followed by a sentence that answers it on its own. If your answer only makes sense after three paragraphs of preamble, an engine has nothing clean to quote.
If you are worried that AI-written copy will count against you, that is a separate question, covered in will AI content hurt your Google rankings.
Can anyone guarantee that ChatGPT will cite my business?
No. ChatGPT, Copilot, Perplexity and Google do not sell citations, do not publish a ranking for them and do not let you see a fixed position. The same question can produce different sources for different people on different days. An agency promising “position one in ChatGPT” is promising something it does not control.
What is measurable is whether you are cited at all, for which questions, and how often. Microsoft now reports exactly that for Copilot, which is the closest thing to a scoreboard this field has. More on that below.
How do ChatGPT, Copilot, Perplexity and Google choose which sources to cite?
None of them publishes its full selection logic, but each publishes the rules for getting into the pool in the first place. Here is what each company says in its own documentation.
- Google AI Overviews and AI Mode. Google’s page on AI features and your website (checked October 2026) says there are no additional requirements and no special optimisations needed. A page has to be indexed and eligible to show in Search with a snippet, and Google says you do not need AI text files, new markup or any special schema.org structured data. Traffic from these features is counted inside the normal Web performance report in Search Console. Our explainer on AI Overviews and zero-click search covers what that does to your clicks.
- ChatGPT. OpenAI’s crawler documentation (checked October 2026) separates three agents. OAI-SearchBot is used to surface websites in ChatGPT’s search features, and sites that block it are not shown in ChatGPT search answers. GPTBot collects content that may be used to train models. ChatGPT-User fetches a page when a user’s question needs it. Blocking GPTBot does not remove you from ChatGPT search; blocking OAI-SearchBot does. OpenAI says a robots.txt change can take about 24 hours to be picked up.
- Perplexity. Perplexity’s bot documentation (checked October 2026) says PerplexityBot exists to surface and link websites in Perplexity’s results and is not used to crawl content for AI foundation models. It recommends allowing PerplexityBot in robots.txt if you want to appear.
- Copilot and Bing. On 10 February 2026 Microsoft launched AI Performance in Bing Webmaster Tools (checked October 2026), a report of how often your pages are cited in Copilot, in AI summaries in Bing and in some partner integrations, with the grounding queries that triggered each citation. Microsoft’s own advice in that announcement is to keep content fresh with IndexNow, use clear headings, tables and FAQ sections, and support claims with evidence.
Read together, the pattern is plain. Be crawlable by the right agent, be indexed in Google and Bing, and give the engine a passage worth quoting. There is no secret tag.
How do I get my business cited by ChatGPT and the other AI engines?
This is the checklist we run on our own sites, in the order we do it.
- Phrase headings as the question people actually type. “How much does a website cost in the UK?” beats “Our pricing philosophy”. The heading is how an engine matches your passage to the query.
- Put the answer in the first words under that heading. Lead with the number, the price, the timescale, or a plain Yes or No, then add the caveat. A sentence that stands on its own is a sentence that can be quoted.
- Show a visible date. A “Last updated” line tells a reader, and an engine, that the figure is current. Update the date only when you have actually checked the content.
- Add FAQ sections with FAQPage schema. Google no longer shows FAQ rich results, so do not expect expandable answers in Google. We keep FAQ sections because Microsoft names them as a structure it favours, and because a clean question and answer pair is easy to extract. Our guide to schema markup explains what structured data can and cannot do.
- Let the AI crawlers in. Check robots.txt allows OAI-SearchBot, ChatGPT-User, PerplexityBot and Bingbot. Many security plugins and CDN rules block them by default without telling you. Whether you also allow training crawlers such as GPTBot is your choice; it is not what decides ChatGPT search.
- Get indexed in Bing, not just Google. Verify the site in Bing Webmaster Tools, submit your sitemap and ping changes with IndexNow, a protocol the IndexNow site (checked October 2026) lists as supported by five search engines: Microsoft Bing, Naver, Seznam.cz, Yandex and Yep. Few small businesses ever open Bing Webmaster Tools, which makes it an easy edge.
- Back every figure with a source. Link the primary source and say when you checked it, as this page does. A sourced number is safer for an engine to quote.
We run hirethecreatives.com this way. As of 4 October 2026, the AI Performance report in Bing Webmaster Tools showed 35 citations in Copilot over the previous 3 months for our pages. It is a modest number for a small studio, and it proves nothing about your site, but it is real and measured by Microsoft, not by us.
Do I need an llms.txt file?
It is optional, and cheap. llms.txt describes itself as a proposal (checked October 2026), first published on 3 September 2024, for a plain-text file at /llms.txt that summarises a site for AI agents. It is not an adopted standard, and Google’s AI features page says outright that you do not need AI text files to appear in its AI features. No major engine has publicly confirmed that it uses the file to choose citations.
We publish one anyway, at /llms.txt, because it takes an hour, costs nothing to host and does no harm. Treat it as housekeeping, not a ranking lever.
How do I check if ChatGPT cites my business?
There is no ChatGPT equivalent of Search Console yet, so checking is part manual and part data. This is the routine we use.
- Write down 10 to 20 questions a customer would ask before buying from you, in their words. “How much does a loft conversion cost in Preston?” is a question; “loft conversions” is not.
- Ask each one in ChatGPT with search switched on, in a fresh chat, and open the sources. Note whether your domain appears, and which page. Repeat in Perplexity, Copilot and Google, where AI Overviews show.
- Repeat on the same day each month. Answers vary, so one check proves little. A monthly log of the same questions shows the trend.
- Open the AI Performance report in Bing Webmaster Tools. It is the only first-party citation count available, and the grounding queries tell you which questions your pages are being used for.
- Look at your server logs or analytics. Hits from OAI-SearchBot, ChatGPT-User and PerplexityBot show the engines are reading you; visits referred from chatgpt.com, perplexity.ai and copilot.microsoft.com show people clicking through.
Google is the hard one. Search Console folds AI Overview and AI Mode traffic into the ordinary Web report, so there is no separate AI Overview figure, and every “AI Overview rank tracking” tool is sampling searches from the outside rather than reading Google’s data.
What is generative engine optimisation (GEO)?
Generative engine optimisation (GEO) is the practice of writing and structuring a page so that an AI system which composes its own answer, rather than listing ten links, picks that page as a source and names it. The “generative engines” are ChatGPT search, Microsoft Copilot, Perplexity, Google AI Overviews and AI Mode, and anything else that reads the web and writes a reply. The page still has to be found by a crawler and indexed first; GEO is about what happens after that, when the engine decides which passages to lean on.
In practice GEO and AEO describe one job. We use “answer engine optimisation” on this page because it is the phrase most UK businesses search for, but every step in the checklist above is GEO as well.
Where did the term generative engine optimisation come from?
From a 2023 research paper. The name was introduced in GEO: Generative Engine Optimization (arXiv 2311.09735, checked October 2026), by researchers at Princeton University and IIT Delhi, first posted on 16 November 2023 and published at the KDD 2024 conference. The authors framed the problem from the side of website owners: search engines that write answers decide which sources get seen, and content creators had no method for improving their chances. They built a large benchmark of user queries across many subject areas, called GEO-bench, to test which edits to a page made it more visible inside generated answers.
Is generative engine optimisation real?
Yes, as a measurable effect, but not as a separate trade with secret techniques. The paper’s own headline result, in its abstract, is that GEO methods can boost a source’s visibility in generative engine responses by up to 40%. That is a research result on the authors’ benchmark, measured with their own visibility metrics, not a promise about your site in ChatGPT next month. What has changed since 2023 is that the engines themselves now publish crawler rules and, in Microsoft’s case, a citation report, so the effect can be checked on real traffic rather than in a lab.
Where the hype runs ahead of the evidence is the idea that GEO needs its own agency, its own tools or a special file. None of the engines’ documentation, quoted earlier on this page, asks for anything beyond being crawlable, indexed and worth quoting.
Is GEO different from AEO and SEO?
No, not in what you actually do to the page; the three names describe where the result shows up. The table sets them side by side.
| SEO | AEO | GEO | |
|---|---|---|---|
| What you are trying to win | A high position in a list of links | A quoted passage inside an answer, with a source link | The same: being used and named as a source in a generated answer |
| Where it shows | Google and Bing results pages | ChatGPT search, Copilot, Perplexity, Google AI Overviews, voice assistants | ChatGPT search, Copilot, Perplexity, Google AI Overviews and AI Mode |
| Unit that wins | The whole page | One self-contained passage | One self-contained passage |
| Prerequisite | Crawled and indexed | Crawled and indexed, by the right AI agents too | Crawled and indexed, by the right AI agents too |
| First-party measurement | Google Search Console, Bing Webmaster Tools | Bing Webmaster Tools AI Performance; manual checks elsewhere | Bing Webmaster Tools AI Performance; manual checks elsewhere |
AEO is the older label, used for featured snippets and voice answers before ChatGPT existed. GEO is the academic label from 2023. Both rest on ordinary SEO, because an engine cannot cite a page it has never fetched.
What generative engine optimisation techniques actually work?
Adding sources, quotations and statistics did best in the original research, with a relative improvement of 30 to 40% over the unedited pages, while keyword stuffing offered little to no gain. The paper tested nine kinds of edit, from making text more authoritative in tone to inserting technical terms, and those three came out on top (full paper, checked October 2026). It is a striking result for anyone still writing for keyword density.
That lines up with what Microsoft tells site owners to do for Copilot: clear headings, tables, FAQ sections and claims supported by evidence. On a small business site it translates into something unglamorous: put the real price or timescale in the first sentence, link where the figure came from, and quote the rule or the source rather than paraphrasing it into mush.
How do you measure generative engine optimisation?
1 first-party report exists today: AI Performance in Bing Webmaster Tools, free, launched in public preview on 10 February 2026. Microsoft’s announcement (checked October 2026) lists four measures: total citations, average cited pages per day, the grounding queries the AI used to retrieve your content, and citation counts for each page. It covers Microsoft Copilot, AI-generated summaries in Bing and some partner integrations.
For every other engine, measurement is a monthly log you keep yourself, using the routine in the section on checking ChatGPT above. These are the numbers worth tracking:
- Citations and cited pages from Bing Webmaster Tools, month on month.
- Grounding queries, which show the questions your pages are being used to answer, often ones you never targeted.
- Share of a fixed question set: of the same 10 to 20 customer questions asked each month in ChatGPT, Perplexity and Google, how many name your site.
- AI referral visits in analytics from chatgpt.com, perplexity.ai and copilot.microsoft.com.
- AI crawler hits in server logs, which tell you the engines can reach the pages at all.
Be sceptical of any tool that gives you a single “GEO score” or a ranking position inside ChatGPT. The engines do not publish one, so a tool selling it is sampling answers from the outside and presenting the sample as a fact.
Do I need a GEO agency in the UK, or can I do this myself?
You can do most of it yourself, and if you have the time you should start with the checklist above. Search for “AEO near me” and you will find agencies, but nothing about this work depends on being local to you.
Where it is worth paying someone is when you do not have the hours, or when the site itself is the problem: slow, unindexed, or built on a theme that buries every answer under a slider. Our SEO Foundation plan (£299 a month) covers technical SEO, schema and AI-readability set-up. Answer engine work proper, the question research, rewriting and citation tracking, is in SEO Growth at £599 a month, with four articles a month and fortnightly reporting. SEO Authority, from £1,200 a month, adds a full answer-engine programme with competitor tracking. All three are cancellable any time, and none of them comes with a citation guarantee, for the reasons above.
We will check your robots.txt, Bing indexing and how your key pages open, and tell you honestly whether you need us or just an afternoon with the checklist.
Get a free website review
See SEO Growth
Get web & SEO tips that actually move the needle
Short, occasional emails for business owners: what's changing in Google, how to win more enquiries, and the fixes worth your time. No spam, no jargon.