Generative engine optimisation (GEO) is the work of making your brand the source that AI answers quote, cite and recommend. When someone asks ChatGPT, Google's AI Mode or Perplexity which agency, product or supplier to trust, the answer is assembled from pages the engine can reach and believes. The short answer to how you win there: keep your SEO fundamentals sound, write pages that contain quotable facts with named sources, and get your brand mentioned on the sites AI engines already lean on.
This guide explains what the research actually shows, what Google, OpenAI, Microsoft and others say officially, and gives you a seven step plan and a checklist you can run this month. It goes deeper on AI citations than our YouTube SEO guide and builds on our report on AI in marketing in 2026.
The short version
- Facts beat keywords. In the original GEO study, adding quotations, statistics and cited sources lifted a page’s visibility in AI answers by up to 40%. Keyword stuffing made it worse.
- Google says there is no special trick. Its official guidance is that AI Overviews and AI Mode need no extra optimisation beyond solid SEO.
- Mentions matter more than links. Across 75,000 brands, web mentions correlated with AI Overview visibility far more strongly than backlinks (Ahrefs, vendor data).
- AI engines cite outside the top 10. Only 37.9% of URLs cited in AI Overviews ranked in Google’s top 10 for that query in early 2026.
- You can now measure it. Search Console and Bing Webmaster Tools both report how your pages appear in AI answers.
What is generative engine optimisation?
The term comes from a 2023 research paper by a team from Princeton, Georgia Tech, the Allen Institute for AI and IIT Delhi, later presented at the KDD 2024 conference. They defined generative engines as systems that search the web and then write a single answer with citations, and they built a benchmark of 10,000 queries to test which page edits made a source more visible inside those answers (Aggarwal et al., arXiv).
The difference from classic SEO is the unit of success. SEO wins a ranked link. GEO wins a sentence: your brand named, your figure quoted or your page cited inside an answer that many people never click past. You still need the page to be crawlable and trusted, so GEO sits on top of SEO rather than replacing it.
Which page edits made sources more visible in AI answers
The pattern is clear and useful. The edits that helped most all added evidence: a quotation from a credible person, a hard number, a named source. Making the text read more fluently also helped. Stuffing in the query terms scored below the untouched page. On Perplexity specifically, the researchers measured gains of up to 37%.
What the platforms say officially
Before spending on GEO tools, read what the engines themselves publish. Google’s Search Central page on AI features is blunt:
“There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” Google Search Central, AI features and your website
Google says the same foundational SEO practices apply, that no special AI text files or extra schema are needed, and that AI features use “query fan out”, running multiple related searches across subtopics to build one answer (Google Search Central). That fan out is why pages that rank on page two can still be cited. In a March 2026 study of 863,000 keyword results, Ahrefs found only 37.9% of AI Overview citations ranked in the top 10 for the same query, down from about 76% in July 2025 (Ahrefs, vendor data).
The other engines run their own crawlers, and each one can be allowed or blocked separately in your robots.txt file.
| Engine | Crawler that powers search answers | Separate training crawler | What blocking the search crawler does |
|---|---|---|---|
| ChatGPT | OAI SearchBot | GPTBot | Your pages will not be shown in ChatGPT search answers |
| Perplexity | PerplexityBot | None (PerplexityBot is not used for model training) | Your pages are not surfaced or linked in Perplexity results |
| Claude | Claude SearchBot | ClaudeBot | May reduce your visibility in user search results |
| Google AI Overviews and AI Mode | Googlebot | Google Extended (does not affect Search) | Controlled with snippet rules or the Search Console AI toggle |
Summary of each company’s crawler documentation, checked 15 October 2026. Copy the exact user agent names from the official pages linked in the sources, as several contain punctuation.
The practical point: you can block a company’s training crawler and still allow its search crawler. A brand that blocks every AI bot also removes itself from AI search answers. In June 2026 Google also added a Search Console toggle to opt out of its generative AI features, tested first with UK site owners and rolled out worldwide by 31 August 2026, and it says opted out sites receive no traffic or impressions from those features (Google). For most brands that toggle should stay on.
How to get your brand cited by AI: a 7 step plan
This is the process we use. Steps one and seven are about access and measurement; the middle five are about becoming worth citing.
1. Open the door to AI crawlers
Check your robots.txt and any firewall or bot protection rules. Make sure Googlebot, Bingbot and the search crawlers in the table above can reach your key pages. Check that important content is in the HTML rather than loaded only by scripts, and that snippet controls such as nosnippet are not set on pages you want quoted.
2. List the questions buyers ask AI
Build a prompt list of 20 to 50 real questions: the ones your sales team hears, the ones in Google autocomplete, and comparison questions such as “best X for Y”. Write them as full sentences, the way people type into a chatbot. This list becomes your test set for step seven.
3. Write answer first pages with evidence
For each priority question, publish one page that answers it in the first two sentences, then backs the answer up. Use the edits the research rewards: specific statistics with dates, short quotations from named experts, and links to primary sources. Add clear headings, a table where comparisons exist, and an FAQ written in your buyers’ words. Microsoft’s own guidance for Bing and Copilot asks for the same things: clear headings, tables, FAQs and claims backed by data.
4. Earn mentions where AI already looks
What correlates with brand visibility in Google AI Overviews
AI engines repeat what the web says about you, so the most important GEO work often happens off your own site. Brands in the top quarter for web mentions averaged 169 AI Overview mentions in the Ahrefs study, more than ten times the next group. Citation sources also differ by engine: Profound’s analysis of about 680 million citations found Wikipedia was 7.8% of all ChatGPT citations, while Reddit led Perplexity at 6.6% and Google AI Overviews at 2.2% (Profound, vendor data). Those mixes shift month to month, so chase the principle rather than one site: digital PR in trade press, expert commentary, honest participation in communities, and reviews on the platforms your category uses. Never post fake reviews or undisclosed promotion; that breaches UK rules and our ASA guidance applies just as much here.
5. Publish video AI can read
YouTube is among the most cited domains in both Google AI Overviews (1.9% of citations) and Perplexity (2.0%) in Profound’s data. Engines read the words, not the pictures, so give every brand video a strong title, a full description, accurate captions and chapters. Our YouTube SEO guide covers the metadata in detail. Video also builds the mentions in step four: our Formula E work has passed 7M+ views across YouTube, Instagram and TikTok, and that kind of reach creates conversation that other sites write about.
6. Keep the SEO basics strong
Fast pages, clean internal linking, accurate structured data, an author with real credentials, and content that is updated when facts change. Google says the same foundational SEO practices apply to its AI features, so a technically weak site will struggle everywhere. Date your statistics and refresh them when newer data appears.
7. Measure AI visibility every month
Three free sources now exist. Google Search Console reports AI feature traffic within the Web search type and has added impression reporting for AI responses. Bing Webmaster Tools launched an AI Performance report in February 2026 showing total citations and the grounding queries behind them, and added Citation Share in June 2026 (Microsoft Bing). In GA4, track referrals from chatgpt.com, perplexity.ai and similar sources, knowing some AI visits arrive as direct traffic. Then run your prompt list monthly and log whether you are named, cited and described accurately. This is the AI version of our share of search method.
Is generative engine optimisation worth it for brands?
Growth in AI referred traffic, holiday season 2025 vs 2024
The volume is still small but the direction and quality are not. Ahrefs’ 2025 study of about 35,000 websites found AI assistants sent just 0.1% of total traffic, against 43.8% from search (Ahrefs, vendor data). Yet AI referred visitors on US retail sites converted 42% better than other traffic by March 2026 and stayed 48% longer (Adobe). In the UK, Deloitte’s 2026 Digital Consumer Trends survey found 64% of 16 to 75 year olds had come across AI generated search summaries in the past year, rising to 73% of Gen Z (Deloitte UK).
Our view: GEO is worth doing now because most of it is good marketing anyway. Evidence led content, earned media and well packaged video help you in Google, on social and with buyers directly. Be wary of anyone selling GEO as a secret formula. Ignore it for two years, though, and competitors will become the names AI repeats when your buyers ask.
Generative engine optimisation checklist
| Check | What good looks like | Done |
|---|---|---|
| Crawler access | Search crawlers allowed in robots.txt; training crawlers decided separately | ☐ |
| Prompt list | 20 to 50 real buyer questions written as full sentences | ☐ |
| Answer first pages | One page per priority question, answer in the first two sentences | ☐ |
| Evidence | Dated statistics, named quotes and primary source links on every key page | ☐ |
| Structure | Clear headings, comparison tables and an FAQ in buyer language | ☐ |
| Mentions | Monthly PR, expert comment and review activity in your category | ☐ |
| Video | Captions, chapters and full descriptions on every brand video | ☐ |
| Measurement | Search Console, Bing AI Performance, GA4 AI referrals and a monthly prompt audit | ☐ |
Advertain guidance.
Frequently asked questions
What is generative engine optimisation in simple terms?
It is making your brand easy for AI tools such as ChatGPT, Perplexity and Google AI Overviews to find, trust and quote, so you are named or cited when people ask questions about your category.
Is generative engine optimisation different from SEO?
It builds on SEO rather than replacing it. SEO aims for a ranked link; GEO aims for a mention or citation inside an AI answer. Google says the same SEO practices apply to its AI features, while off site mentions and evidence rich writing matter more for GEO.
Does llms.txt help with generative engine optimisation?
There is no evidence it does today. In June 2025 Google’s John Mueller said no AI system currently uses llms.txt, and Google’s own guidance says no special AI text files are needed. Spend the time on content and mentions instead.
How do you measure generative engine optimisation?
Use Google Search Console’s AI reporting, the AI Performance report in Bing Webmaster Tools, AI referral sources in GA4, and a monthly audit of your prompt list that records whether your brand is named, cited and described correctly.
How long does generative engine optimisation take?
Technical fixes such as crawler access can take effect within days; OpenAI says robots.txt changes are picked up in about 24 hours. Earning mentions and becoming a trusted source usually takes months, which is why it pays to start early. If you want help, talk to us or see our work.
