Buyers now ask ChatGPT, Perplexity and Google's AI Overviews which agency, tool or approach to pick, and they often act on the answer without visiting ten blue links. Generative engine optimisation (GEO) is the work of making your pages the sources those answers quote. The good news: most of it is disciplined SEO. The bad news: the disciplined part is exactly what most sites skip.
What is generative engine optimisation (GEO)?
GEO is the practice of making your content easy for AI answer engines to find, trust and quote. Where classic SEO aims for a ranking position, GEO aims for a citation: your brand named, or your page linked, inside the generated answer. You will also see it called AEO (answer engine optimisation) or AI SEO. They describe the same work.
Google's own position is blunt. Its Search Central guide to generative AI features says optimising for AI search is optimising for the search experience, and therefore still SEO. AI Overviews and AI Mode are grounded in the same index and ranking systems as regular results. If a page is not indexed and eligible for a snippet, it cannot be a source.
How do AI answer engines choose which pages to cite?
The engines work in broadly the same way. They take the question, fan it out into several related searches, retrieve pages from an index, and write an answer grounded in the passages they retrieved. The pages that get cited are the ones that rank for those fan-out searches and contain a passage that answers the sub-question cleanly.
The index differs by engine. Google AI Overviews and AI Mode use Google's index. ChatGPT search and Microsoft Copilot lean heavily on Bing. Perplexity runs its own crawler alongside other sources. That is why a site can appear in one engine's answers and be invisible in another's, and why Bing Webmaster Tools matters again.
- Google AI Overviews and AI Mode: Google's index, same quality systems as Search
- ChatGPT search and Copilot: largely Bing's index, plus OpenAI's own crawler
- Perplexity: its own crawler plus third-party search results
- All of them: reward pages that answer a specific sub-question directly
The GEO myths you can safely ignore
A lot of GEO advice is invented. Google's guide lists several tactics it says you do not need: an llms.txt file, special AI markup, rewriting content into tiny chunks, stuffing pages with long-tail phrasings, and chasing inauthentic mentions on forums and videos. None of them are ranking inputs for AI Overviews.
We still keep llms.txt on client sites because it costs nothing and some smaller AI tools read it. But we never sell it as the fix. When a site is not being cited, the cause is almost always one of the fundamentals below.
1. Make sure AI crawlers can actually reach you
Start with access. Check robots.txt, your CDN and any bot-protection rules for GPTBot and OAI-SearchBot (OpenAI), PerplexityBot, ClaudeBot, Bingbot and Googlebot. We regularly find firewalls silently blocking AI crawlers that the marketing team assumed were allowed.
Then check rendering. If your key content only appears after JavaScript runs, some AI crawlers will see an empty page. Server-rendered or statically generated HTML is the safest foundation.
- robots.txt allows the search crawlers you want citing you
- WAF and CDN bot rules do not block them with a 403 or a challenge page
- Primary content is in the initial HTML, not injected later
- Your site is verified in both Google Search Console and Bing Webmaster Tools
2. Answer the question in the first two sentences
AI engines quote passages, not pages. A section that opens with a direct, self-contained answer is far easier to lift than one that warms up for three paragraphs. We rewrite key sections so the first 40 to 60 words answer the heading's question on their own, and the rest adds evidence and nuance.
Question-style headings help for the same reason: they mirror the fan-out searches the engine runs. 'How much does CRM automation cost?' is a better H2 than 'Pricing considerations'.
3. Publish what nobody else can
Google's guidance draws a clear line between commodity content and first-hand experience. A generic '7 tips' article competes with thousands of near-identical pages. A write-up of what happened when you rebuilt a client's pipeline, with the numbers, has no substitute.
For a service business, your unfair advantage is your delivery data: response times you cut, conversion rates you moved, the objections buyers actually raise. Put those numbers on your case studies and blog posts, attributed and dated. Specific, verifiable facts are what answer engines prefer to cite.
4. Make your brand an unambiguous entity
An engine can only recommend you if it is confident who you are. Use the same company name, description, address and social profiles everywhere: your site, Google Business Profile, LinkedIn, Clutch, directories. Back it with Organization structured data that links those profiles together.
Named people help too. Articles with a real author, a short bio and a link to their LinkedIn give both readers and engines a reason to trust the claims.
5. Earn mentions where buyers already ask
Answer engines lean on third-party sources when they compare vendors: review sites, industry roundups, Reddit threads, YouTube walkthroughs and LinkedIn posts. Being mentioned there, by people who are not you, is often what moves a brand into the shortlist an AI gives.
This is PR and community work, not link building. Genuine client reviews on Google and Clutch, guest appearances on niche podcasts and helpful answers in the communities your buyers use all compound.
6. Keep your important pages fresh
Answer engines favour current information, especially for anything involving tools, prices or platforms that change quickly. A service page last touched two years ago is a weak candidate against a competitor's page updated last month.
Put your top ten commercial pages and best-performing articles on a quarterly review. Update the facts, add what you learned since, and show a visible 'last updated' date so readers and engines can see it is current.
7. Measure citations, not just rankings
You cannot improve what you do not track. Pick 20 to 30 questions your buyers really ask, run them through ChatGPT, Perplexity and Google each month, and log which brands and URLs are cited. It is manual at first, and it is the most honest GEO report you will get.
In analytics, segment referral traffic from chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com. Watch which landing pages those visitors arrive on; they show you which content the engines already trust.
- A fixed list of buyer questions, checked monthly in each engine
- Share of answers that mention your brand, and which URL they cite
- AI referral sessions and conversions by landing page
- Search Console and Bing Webmaster data for the pages that get cited
Where to start this week
If you only do three things: confirm AI and search crawlers can reach and render your key pages, rewrite the opening of your five most important pages so they answer the question immediately, and add real numbers to your case studies. Those three changes fix the problems we find on most sites we audit.
GEO is not a separate channel with its own tricks. It is the reward for doing SEO properly, publishing things only you know, and being talked about in the places your buyers look.