Home

About us

Projects

Blogs

Locations

SEO

How to Optimise Your Website for AI Search: AEO and GEO in Practice

A web designer editing a page with a highlighted answer paragraph, with a robots.txt file open on a second screen

Quick answer: To optimise a website for AI search, make each page easy to find, read and quote. Let the AI search crawlers in, put the answer to each question in plain text near the top, organise pages around the questions customers ask, and say things only your business can say. Google states that no special files or markup are needed for its AI features. AEO and GEO are names for this work, not a separate discipline from SEO.

This guide explains the terms, how AI search engines choose their sources, how to structure a page, which crawlers to allow, how to measure the result, and what is not worth your time.

What do AEO, GEO and AI SEO mean?

Term Stands for What people mean by it
SEO Search engine optimisation Getting pages to rank in search results
AEO Answer engine optimisation Getting pages used in direct answers, including AI answers
GEO Generative engine optimisation Getting pages cited by generative AI search; the term comes from a 2024 research paper
AI SEO A general label for either of the above

The names suggest three jobs. In practice it is one. Google's own guide puts it this way: "The best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems" (Google Search Central, updated July 2026).

How do AI search engines choose what to cite?

AI search engines search first and write second. An AI search tool takes the question, runs searches, reads the pages it finds, and writes an answer with links to some of them. Two things follow from that.

Your page has to be reachable by that tool's crawler. Google uses its normal index. ChatGPT search and Perplexity use their own crawlers. A page they cannot fetch cannot be cited.

Your page competes on a narrow question, not the broad one. Google describes a technique called query fan-out: "concurrent, related queries generated by the model to request more information". A long question is split into smaller searches. A page that answers one of those clearly can be used in the answer to the larger question. Our guide to Google AI Mode covers this in more detail.

What the research shows

The paper that named GEO, presented at KDD 2024, tested ways of rewriting content to see which were cited more often by generative search engines. Its headline finding was that the methods "can boost visibility by up to 40% in generative engine responses", with adding sources, quotations and statistics among the more effective ones (Aggarwal et al.). The authors also found that results varied by subject area.

Treat that as direction. It was measured on a research benchmark, not on Google or ChatGPT as they are today.

What Google says matters

Google's guide asks for content with "unique, expert or experienced takes that go beyond common knowledge and the ordinary". General advice that could appear on any site is the easiest content for an AI answer to replace. Your own prices, process, results and local knowledge are the hardest.

How to structure a page for AEO

Diagram of a web page structured for AI search: a title that states the topic, a direct answer, headings written as questions, a table, linked sources, questions and answers, and an author and date

A page that is easy to quote is also easy to skim. The parts:

  1. A title that states the topic in the words people search with.
  2. The answer first. Two or three sentences near the top that make sense on their own.
  3. Headings written as questions, each answered in the first sentence beneath it.
  4. A table wherever you compare options, prices or steps.
  5. Sources linked next to the claims they support.
  6. Questions and answers for the things customers ask.
  7. A date and an author, so a reader knows how current it is and who stands behind it.

The opening paragraph matters most. Compare two openings for a page about website speed:

Weak: "In today's fast-paced digital world, website performance has become increasingly important for businesses of all sizes."

Strong: "A slow website is usually caused by oversized images, too much JavaScript, a slow server or web fonts. Each can be fixed without a redesign."

The second can be lifted into an answer as it stands. The first says nothing.

One caution from Google: "There's no requirement to break your content into tiny pieces for AI to better understand it." Structure a page so a person can follow it. Do not chop it into fragments.

How do you optimise content for AI search engines?

In order of importance:

  1. Make the page crawlable and indexable. For Google's AI features, a page "must be indexed and eligible to be shown in Google Search with a snippet" (Google). Our technical SEO checklist covers the checks.
  2. Put the content in the HTML. Text that appears only after JavaScript runs is harder for Google and invisible to crawlers that do not run scripts. Our own pages were being redrawn by a site builder's JavaScript until we changed how they are served; the case study explains it.
  3. Answer first, explain second, on every page and under every heading.
  4. Cover the related questions. Because of query fan-out, a site that answers the questions around a topic has more chances to be used than one page aimed at one phrase. See content strategy for SEO.
  5. Add what only you know: your figures, your process, your examples.
  6. Link your sources and date your pages.
  7. Keep your business details consistent across your site, your Google Business Profile and directories.

Do you need llms.txt, special schema or AI files?

For Google, no. Its guide says: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search", and that Search ignores them. It also says "structured data isn't required for generative AI search".

Three practical points follow.

  • Structured data is still worth having for what it has always done: describing a page as an article, a product or an organisation. It is not a lever for AI answers.
  • FAQ rich results are gone. Google's documentation records that the FAQ rich result stopped appearing in Search from 7 May 2026 (Google). Questions and answers on a page are still useful to readers. Adding FAQ markup in the hope of a search feature is not.
  • llms.txt is a proposal, not a standard. It is a short markdown summary of a site for AI tools (llmstxt.org). We publish one on our own site because it costs nothing, and we have seen no evidence that it changes what gets cited. Our llms.txt guide explains the format and how we built ours, and our guide on how to rank in Google AI Overviews covers Google's requirements step by step. Do not put it ahead of anything in the list above.

Which AI crawlers should you allow?

Each company runs separate crawlers for search and for training its models. They can be allowed or blocked separately in robots.txt.

Crawler Company Used for
Googlebot Google Google Search, including AI Overviews and AI Mode
Google-Extended Google Controls use of your content for AI training and grounding in some of Google's other systems
OAI-SearchBot OpenAI Showing sites in ChatGPT's search features
GPTBot OpenAI Training OpenAI's models
ChatGPT-User OpenAI Visits made when a ChatGPT user asks for a page
PerplexityBot Perplexity Showing and linking sites in Perplexity's results
Perplexity-User Perplexity Visits made when a Perplexity user asks a question
Claude-SearchBot Anthropic Improving search results for Claude users
ClaudeBot Anthropic Collecting content that may be used in training
Claude-User Anthropic Visits made when a Claude user asks a question

Sources: OpenAI, Perplexity, Anthropic and Google.

If you want to be cited, allow the search crawlers: Googlebot, OAI-SearchBot, PerplexityBot and Claude-SearchBot. OpenAI states that sites which disallow OAI-SearchBot "will not be shown in ChatGPT search answers". Whether to allow the training crawlers is a separate decision about how your content is used, and blocking them does not block the search crawlers. The crawlers that act on a user's request do not always follow robots.txt: OpenAI and Perplexity both say so for theirs.

Check your robots.txt before anything else. A rule that blocks everything, left over from a staging site, blocks all of them.

How do you measure AI search visibility?

In Google. Search Console has a generative AI performance report. It shows impressions, meaning how often links to your site were shown in AI Overviews and AI Mode, by page, country, device and date. It does not show clicks, click-through rate or position (Search Console Help). The same impressions are also counted in the main Performance report.

For scale: in the first three weeks our own site was indexed, that report recorded three impressions. Early numbers are small. Look at the trend over months.

In other AI search tools. There is no equivalent report. Check by hand: ask the questions your customers ask and note whether your site is cited. In your analytics, look for visits referred from those tools.

In your business. Count enquiries, and how many people search for you by name. A Pew Research Center study of US browsing in March 2025 found people clicked a search result on 8% of visits when an AI summary was shown, against 15% when it was not (Pew). Fewer clicks does not have to mean fewer customers, so measure the customers.

What does not work

  • Writing separate pages for AI. Google's guide says to write for people, and its spam policies cover content produced at scale to manipulate rankings.
  • Stuffing keywords into headings and answers.
  • Guarantees. Nobody controls what an AI tool cites, so nobody can promise it.
  • Chasing a file or a tag while the pages themselves are thin.

Do you need an AI SEO service?

Often not. If your site is crawlable, fast and clearly written, most of this is already done. An AI SEO audit is worth it when you cannot answer these questions yourself:

  • Can the AI search crawlers reach every page that matters?
  • Is the main content in the HTML, or does it depend on JavaScript?
  • Does each service page state plainly what you do, for whom and where?
  • Which customer questions does the site not answer?
  • Is Search Console's generative AI report being watched?

Ask anyone offering AI SEO services what they will change and how they will measure it. For the business case, the results the data supports and what a service should include, see AI SEO for business. Our guide to choosing a local SEO agency has a longer list of questions.

We start with whether your pages can be crawled, read and quoted, then look at what they say. Get in touch and tell us which questions you want to be the answer to, or see how our AI SEO, AEO and GEO services work.

If you would rather have this handled for you, these pages explain what we do and where we work:

Frequently asked questions

What is answer engine optimisation (AEO)?

Answer engine optimisation is the practice of making a website easy for AI search tools, such as Google's AI Overviews and AI Mode, ChatGPT and Perplexity, to find, understand and cite. In practice it means crawlable pages, clear answers stated early, pages organised around real questions, and content that says something specific.

What is the difference between SEO, AEO and GEO?

SEO aims to rank a page in search results. AEO (answer engine optimisation) and GEO (generative engine optimisation) aim to have a page used and cited in an AI-written answer. The work overlaps almost completely. Google says its AI features are built on its core Search systems, so the same best practices apply.

Do I need llms.txt or special schema to appear in AI search?

Not for Google. Its documentation says you do not need machine readable files, AI text files or special markup to appear in its AI features, and that Search ignores such files. Structured data still helps describe a page, but it is not a requirement for AI search.

How do I get my website cited by ChatGPT?

Make sure OpenAI's search crawler, OAI-SearchBot, is allowed in your robots.txt, since OpenAI says sites that block it will not be shown in ChatGPT search answers. Beyond that, the same things apply as for any search engine: pages that load without JavaScript, clear answers, and content worth quoting. Nobody can guarantee a citation.

How long does AI search optimisation take to show results?

There is no published timeline. A page has to be crawled and indexed before it can be cited, so new pages and new sites take longer. Judge it over months, using Search Console's generative AI report for Google and your own enquiries and referral traffic for the rest.

Do I need an AI SEO service?

Not if your site is already well built and well written, because the work is mostly the same as good SEO. An AI SEO audit is worth paying for when you do not know whether AI crawlers can read your pages, or when competitors are cited for your services and you are not. Ask what will be changed and how the result will be measured.

← Back to all posts