GEO (generative engine optimisation) is the work of getting your pages used and cited in AI-generated answers, and it builds on SEO rather than replacing it. Google says there are no extra requirements to appear in AI Overviews or AI Mode beyond being indexed and eligible to show a snippet, so the groundwork is the same.
What does change is how answers are put together, how much traffic they pass on, and how you measure success.
Below, I separate what Google actually documents from what I've learned doing this work on live eCommerce sites. A lot of GEO advice online blurs the two.
You'll get a side-by-side comparison of SEO and GEO, the mistakes I see most often, and the order I work in when a site needs to start showing up in AI answers.
Key takeaways
- Google documents no special requirements, files or schema for AI Overviews and AI Mode. Crawlable, indexed, snippet-eligible pages are the entry ticket.
- AI answers are built by retrieving pages and running several related searches at once (query fan-out), so you compete on sub-questions, not one keyword.
- Fewer people click through when an AI summary appears, so citations and brand mentions matter more as a measure of visibility.
- Unique, first-hand content is the lever Google itself calls most influential. llms.txt files, chopped-up pages and manufactured mentions are not.
- Measure with Search Console's Generative AI performance report, Bing's AI Performance report and a sampled citation tool, and read the numbers as trends.
What GEO actually means
GEO is a label for one goal: making your content easy for AI systems to find, trust and cite when they write an answer. You'll also see AEO (answer engine optimisation) and AIO (AI optimisation). They describe the same territory with different emphasis, and I don't think the label matters much.
The mechanism matters more. Google's guidance on optimising for generative AI features describes two techniques. The first is retrieval-augmented generation, also called grounding: Search pulls current pages from its index, the model writes a response from them, and links to supporting pages appear alongside. The second is query fan-out, where the model runs a set of related searches at the same time to gather more information.
Other assistants work on similar lines, but with their own crawlers and indexes. OpenAI, for example, recommends allowing OAI-SearchBot in robots.txt if you want your site to appear in ChatGPT's search results. So "AI search" isn't one system. It's several retrieval systems, each deciding which pages to read and quote.
That's why I treat AI SEO and GEO work as an extension of technical SEO and content strategy, not a separate channel with its own tricks.
What Google documents, and what it doesn't
Google has been unusually direct about AI search. Its documentation confirms that AI Overviews and AI Mode have no additional requirements: a page must be indexed and eligible to show in Search with a snippet, and nothing more. The table below sets out the main documented points next to how I apply them.
| Topic | What Google documents | What I do with it (practitioner view) |
|---|---|---|
| Eligibility | No extra requirements beyond being indexed and snippet-eligible. Inclusion is never guaranteed. | Technical SEO comes first. A page that can't show a snippet can't be cited. |
| Special files and markup | No special schema.org markup, AI text files or Markdown versions are needed. Google Search ignores llms.txt. | I don't spend a sprint on llms.txt. That time goes into content and crawlability. |
| Chunking content | There's no requirement to break content into tiny pieces for AI to understand it. | I write clear sections for readers. That happens to make passages easier to lift as well. |
| Mentions | Chasing inauthentic mentions across the web isn't as helpful as it might seem. | I aim for real coverage and reviews, never paid "best of" placements. |
| What matters most | Unique, non-commodity content is likely the most influential factor over time. | First-hand data, prices, photos and decisions beat rewritten summaries. |
| Reporting | Clicks from AI features count in the Performance report under the Web search type. A separate report shows AI feature impressions. | I track AI impressions by page monthly and compare them with organic clicks. |
| Opting out | Settings, then Search generative AI, lets you exclude a site. It isn't a ranking signal for the rest of Search. | Most businesses should leave it on Include. |
Everything in the right-hand column is my experience, not Google's documentation. The exclusion setting is described in Search Console's help page on the Search generative AI control, which covers AI Overviews, AI Mode and generative AI features in Discover.
One more documented point is easy to miss. The Google-Extended robots token controls whether your content is used for Gemini model training and grounding in Gemini Apps and Vertex AI, and Google states it doesn't affect inclusion or ranking in Google Search. Blocking it won't remove you from AI Overviews, and allowing it won't get you in.
What stays the same
Most of the work that earns rankings also earns citations. In Google's case that's by design, because its AI features draw on the same index and the same ranking and quality systems as ordinary results.
- Crawling and indexing. Google's best practices list starts with allowing crawling in robots.txt and in any CDN or hosting layer, then making content findable through internal links.
- Content as text. Important information needs to be in the HTML as text, not locked inside images, PDFs or scripts that never render.
- Search intent. A page still has to match what the person wants: a definition, a comparison, a price, a local provider.
- Authority. Links and mentions from reputable sites still signal that you're worth trusting.
- Page experience. Fast, stable, readable pages help both the reader who clicks and the system deciding what to show.
If your technical base is shaky, work through a technical SEO audit checklist before touching anything labelled AI. I've yet to see a site with indexing problems fix them by adding AI-specific content.
What actually changes
The changes are real, but they're about how answers are assembled and consumed, not about a new set of ranking rules.
You compete on sub-questions, not one keyword
Query fan-out means one question becomes many searches behind the scenes. Google's own example: a search for how to fix a lawn full of weeds can fan out into "best herbicides for lawns" and "remove weeds without chemicals".
For a business, that means your page might be retrieved for a sub-question you never targeted. A search for the best meat delivery for a large family could fan out into delivery areas, price per kilo, freezer space and bulk versus weekly boxes. If your category page or FAQ answers those clearly, you have more ways in.
The trap is to build a page for every fan-out query. Google's guidance says creating pages for every variation primarily to manipulate results falls under its scaled content abuse policy. Cover sub-questions on the page where they naturally belong.
Fewer clicks per search
When an AI summary appears, people click less. Pew Research Center found that users clicked a traditional result in 8% of visits with an AI summary, against 15% without, based on March 2025 browsing data from 900 US adults.
Ahrefs measured a similar effect on click-through rate. Its re-run on December 2025 data found that the presence of an AI Overview correlated with a 58% lower average click-through rate for the top-ranking page, up from 34.5% in its first study. Both are correlations, not controlled experiments, but they point the same way.
My response is practical. I report AI impressions and citations as visibility, and I make sure the pages that do get clicked convert well, with clear prices, delivery details and a next step.
Being named matters as much as being linked
AI answers often name brands without linking them. A mention in an answer about the best bulk meat suppliers is visibility even if nobody clicks.
Those mentions come from what the web says about you: your own pages, reviews, directories, press and forums. If your prices, delivery areas or opening hours differ between your site, Google Business Profile and Merchant Center, you give AI systems conflicting facts. Google's guidance specifically mentions keeping Merchant Center and Business Profile information current for shopping and local visibility.
There's more than one engine
Google AI Overviews and AI Mode, ChatGPT, Perplexity, Microsoft Copilot and Gemini each retrieve differently. Since February 2026, Bing Webmaster Tools has offered an AI Performance report covering Copilot citations and Bing's AI-generated summaries. Ignoring Bing is a common blind spot.
Measurement becomes sampled
You can't see every AI answer. Search Console now shows AI feature impressions by page, but not by query. Third-party tools run large prompt sets and report what they find, which makes them samples, not complete counts.
SEO vs GEO side by side
Here's how I explain the difference to a business owner. Read the right-hand column as additions to SEO, not replacements.
| Dimension | Traditional SEO | GEO |
|---|---|---|
| Goal | Rank a page and win the click | Be retrieved, cited or named inside an answer, then win the click that follows |
| Unit of competition | A page against other results for a query | A passage or fact against many sources across fan-out queries |
| Main inputs | Relevance, intent match, links, technical health | The same inputs, plus self-contained answers, first-hand detail and consistent business facts |
| Primary metrics | Rankings, clicks, click-through rate, conversions | AI impressions, citations, mentions, share of voice, plus referral clicks and conversions |
| Main tools | Search Console, GA4, rank trackers, site crawlers | Search Console's Generative AI report, Bing AI Performance, citation trackers, manual prompt checks |
| Engines | Mostly Google, some Bing | Google AI features, ChatGPT, Perplexity, Copilot, Gemini and others |
| Volatility | Rankings shift over days and weeks | Answers can change with wording, location and time of day |
What I've seen on real sites
I'll be careful here, because it's easy to overclaim. I can't prove which change caused which citation, and no tool sees every AI answer.
On Cost Cut Meats, an Australian online meat retailer, 1,115 AI answers cite the site according to Ahrefs AI responses data from 6 Nov 2026, and 536 of those come from Google AI Overviews. None of that came from AI-specific tricks. The work was ordinary SEO done thoroughly: Ahrefs Site Health went from 79 to 100, orphan pages went from 45 to 0, and 163 live pieces of content (110 blog posts and 53 recipes) were each written from a brief.
On Carnivore Society's bulk meat store, Ahrefs Site Health went from 60 to 99, and AI tools cite the site about 1,400 times.
The pattern I take from both is simple. Pages that answer a specific question directly, on a technically clean site, get picked up. Every site has its own challenges, so the fixes differed, but neither needed a separate GEO hack.
Where to start: the order I work in
Every business starts from a different place, so treat this as a priority order rather than a fixed recipe. If a step is already solid on your site, move to the next one.
1. Confirm AI systems can reach you
Check robots.txt allows Googlebot, Bingbot and OAI-SearchBot. Then check your CDN or security plugin, because bot-management rules can block crawlers without anyone noticing. In Search Console, open Settings, then Search generative AI, and confirm the site is set to Include.
2. Fix indexing on the pages that matter
Use the Page indexing report and URL Inspection on your top category, product, service and guide pages. Also look for snippet controls such as nosnippet or a very low max-snippet value, because these limit what can be shown in AI features as well.
3. Rewrite key pages to answer first
Put the direct answer in the opening lines, then the detail. I cover the patterns, with before-and-after examples, in writing content AI tools can quote.
4. Make your business facts consistent
Line up your name, address, phone, prices, shipping and returns across your site, Google Business Profile, Merchant Center and social profiles. Then describe them with structured data for AI search, such as Organization, LocalBusiness and Product markup that matches what's on the page.
5. Earn real mentions
AI answers lean on what others say about you. Original data, expert commentary and useful tools give journalists and bloggers a reason to cite you, and digital PR for small brands is how I go after that without a big budget.
6. Set up measurement before you change much
Record a baseline first, or you won't know what moved. My monthly routine for tracking AI mentions of your brand combines Search Console, Bing, a citation tool and a fixed set of real customer questions.
GEO mistakes I see most often
- Treating llms.txt as a priority. Google Search doesn't use it. Other services might, but it won't fix weak content.
- Chopping pages into fragments. Google says you don't need to. Short, scattered pages often answer less than one strong page.
- Publishing a page for every question variation. That drifts into scaled content abuse, and it thins out your site.
- Buying mentions. Paid "top 10" placements are exactly the inauthentic mentions Google says aren't as helpful as they seem.
- Rewriting everything into robotic Q&A. Write for the reader. Google's systems understand synonyms and meaning without special phrasing.
- Judging from one screenshot. AI answers vary by wording, location and time. One result proves nothing either way.
- Blocking crawlers by accident. A firewall rule or a copied robots.txt line can quietly remove you from an engine.
How to tell whether it's working
Pick a baseline month, then compare like with like. These are the signals I watch, roughly in order of reliability.
- Search Console Generative AI performance report. Impressions in AI Overviews and AI Mode by page, country, device and date. It doesn't show queries or clicks.
- Search Console Performance report. Clicks from AI features are included in the Web search type totals, alongside ordinary results.
- Bing Webmaster Tools AI Performance. Citations, cited pages and the grounding queries Bing used to retrieve your content.
- A citation tool such as Ahrefs Brand Radar. Mentions, citations and share of voice across several AI platforms, from sampled prompts.
- Manual prompt checks. The same customer questions each month, logged consistently.
If AI impressions rise but clicks don't, look at whether the cited pages give people a reason to visit. If a competitor is cited where you aren't, compare the two pages section by section.
The bottom line
GEO is SEO done with more care for clarity, evidence and consistency. The fundamentals haven't moved. The way answers are assembled and measured has.
Start with access and indexing, then make your best pages the clearest answer available for the questions your customers actually ask.
If you'd like a second pair of eyes on where your site stands in AI search, I can help through AI search optimisation (AIO and GEO).
FAQs
1. Is GEO replacing SEO?
No. Google's own position is that optimising for generative AI search is still SEO, because its AI features rely on the same index and ranking systems. GEO adds emphasis on clear answers, first-hand content and consistent facts, but the foundations don't change.
2. Do I need an llms.txt file?
Not for Google. Google Search ignores llms.txt, so it neither helps nor harms your visibility there. Some other services may read it, but I treat it as a low priority behind indexing and content quality.
3. Does structured data help you appear in AI Overviews?
Google says structured data isn't required for its generative AI features and there's no special markup to add. It's still worth using to describe your business and products accurately and to stay eligible for rich results. It doesn't guarantee a citation.
4. How do I know if my site appears in AI Overviews?
Open the Generative AI performance report in Search Console, under the Performance reports. It shows how often links to your site appeared in AI Overviews and AI Mode, broken down by page, country, device and date.
5. Can I stop Google using my content in AI Overviews?
Yes. In Search Console, Settings, then Search generative AI, lets you exclude your site, and snippet controls such as nosnippet also apply. Excluding means losing the impressions and traffic those features send, so I rarely recommend it.
6. How long does GEO take to show results?
There's no fixed timeline. It depends on how quickly pages are recrawled, how competitive the topic is and how much needs fixing. I judge progress on trends across two or three monthly reports, never on a single week.