Generative Engine Optimisation (GEO): A Practical Guide for 2026
Ask ChatGPT, Perplexity or Google's AI Mode to recommend a supplier in your field, and one of three things happens. Your business is named and linked as a source, it is named without a link, or it does not appear at all while two competitors do. Generative engine optimisation is the work of moving from the third outcome to the first, and most of it is less mysterious than the name suggests.
This guide explains what generative engine optimisation (GEO) is, how AI answers pick their sources, what the research says actually helps, which changes to make on your site and beyond it, and how to measure whether any of it is working. It replaces our earlier articles on AI optimisation and Google's AI search features, which were written before much of the current guidance existed.
What is generative engine optimisation (GEO)?
Generative engine optimisation is the practice of making a business easy for AI tools to find, understand and cite when they generate answers. It covers AI Overviews and AI Mode in Google, and assistants such as ChatGPT and Perplexity. It is also called AI SEO or LLM SEO, and it is built on classic SEO rather than replacing it.
The term comes from research. A paper titled GEO: Generative Engine Optimization, by researchers from Princeton University and IIT Delhi and published at the ACM SIGKDD conference in 2024, introduced the name and tested how content changes affect visibility in AI-generated answers. Its headline finding was that the right changes can boost a source's visibility by up to 40% in generative engine responses.
The key difference from traditional SEO is the goal. Classic SEO aims for a high position in a list of links. Generative engine optimisation aims for your content to be used and credited inside the answer itself, which is often read instead of the links.
How do AI answers choose which sources to cite?
For Google, the answer is surprisingly conventional. Google's documentation on AI features and your website says the usual SEO best practices still apply, and that there are no additional requirements or special optimisations needed to appear in AI Overviews or AI Mode. To be shown as a supporting link, a page simply needs to be indexed and eligible to appear in Search with a snippet.
Google also says sites do not need new machine-readable files, AI-specific text files or special schema markup to appear in these features. That rules out a lot of the shortcuts sold under the GEO label. What Google does recommend is foundational work:
Allow crawling in robots.txt and through any CDN or hosting firewall.
Make content findable through internal links.
Provide a good page experience.
Keep important content available as text.
Support text with relevant images and video.
Make sure structured data matches the visible page text.
Keep Merchant Center and Business Profile information current.
Other AI assistants work in broadly similar ways when they search the web: they retrieve pages they can access, then select passages that answer the question clearly. The practical lesson is the same across platforms. If a page cannot be crawled, or its answer is buried, it is unlikely to be used.
What does the research say actually works?
The KDD 2024 GEO paper tested nine content changes across a large benchmark of queries. Adding quotations from relevant sources performed best, improving visibility by about 41% over the baseline on the paper's main measure. Adding statistics improved it by about 31%, while fluency improvements and citing sources each added around 27 to 28%.
The same study found that keyword stuffing, a traditional SEO tactic, performed worse than doing nothing on that measure, at about 8% below the baseline. The authors also noted that the effect of each method varied by subject area, so there is no single formula that works for every industry.
Read the findings sensibly. The study tested controlled changes in a research setting, not live websites, and AI products have changed since. But the direction is consistent with Google's own advice: content that is clear, specific, well sourced and backed by real data is easier to use and safer to cite than vague copy optimised for keywords.
What should you change on your website?
Start with the pages that describe what you do and who you serve, then work outwards. These are the changes that matter most for generative engine optimisation.
1. Answer each question early
Put the direct answer at the top of each section, under a heading that matches the question people ask. AI tools extract passages, so a section that answers in its first two sentences is far easier to use than one that builds up to its point. Each section should make sense on its own, without the paragraphs above it.
2. State facts, prices and limits clearly
Vague claims give an AI tool nothing to cite. Specific ones do: where you operate, which services you offer, what they include, typical timelines, starting prices where you publish them, and who the service is not right for. If a detail is important to buyers, write it down on the page in plain text.
3. Back claims with sources and data
The research favours content with statistics, quotations and citations. Use data from primary sources, name the publisher and the year, and link to it. Your own first-party data, such as results from client work, is even more valuable, because nobody else can publish it.
4. Keep your business details consistent everywhere
AI tools build their picture of your business from many places. Align your business name, description and service list across your website, Google Business Profile, directories and social profiles. Inconsistent descriptions make it harder for a model to be confident about who you are and what you do.
5. Use structured data that matches the page
Organisation, service and FAQ schema can help machines parse your pages, but Google is clear that no special markup is required for its AI features, and that structured data must match the visible text. Build schema only from facts that are on the page, and treat it as support for good content, not a substitute.
6. Check crawler access
Make sure robots.txt and your CDN or firewall settings let the crawlers you want reach your key pages. Some security settings block AI crawlers such as GPTBot and PerplexityBot by default, which can keep you out of those tools' answers regardless of how good the content is. Our technical SEO audits include this check.
Why do third-party mentions matter so much?
Because AI answers often draw on what other sites say about you, not only what you say about yourself. When someone asks for the best provider of a service, the answer frequently reflects directories, review platforms, comparison articles and press coverage. A business that appears consistently in those places is easier for a model to recommend.
Earning those mentions is classic digital PR and outreach: getting listed in relevant directories, collecting genuine reviews, contributing expert commentary and being included in reputable comparisons. Quality matters far more than volume. Our link building services focus on earned editorial coverage for the same reason.
Can you control whether Google uses your content in AI answers?
Yes, within limits. According to Google's AI features documentation, the same preview controls that apply to normal search results apply to AI Overviews and AI Mode. You can use nosnippet, data-nosnippet, max-snippet or noindex to limit what Google shows from a page.
Google-Extended is a separate control. It lets you limit the use of your content for training and grounding in other Google AI systems, but it does not govern how Googlebot crawls for Search. Before restricting anything, weigh the trade-off: blocking previews also reduces the chance of being cited and linked.
Do you need an llms.txt file?
Probably not, at least not as a ranking tactic. llms.txt is a proposed standard for giving AI tools a summary of a site, but there is little public evidence that major AI tools use it, and Google says sites do not need AI-specific text files to appear in its AI features. Adding one is cheap and does no harm, which is why Invisio Solutions includes it in some projects, but we do not sell it as a ranking factor and you should be wary of anyone who does.
How do you measure AI citations?
Measure with a fixed set of prompts, run regularly, and track trends rather than single results. AI answers vary from one run to the next, so one check tells you very little. A monthly prompt set gives you a baseline and shows whether visibility is moving. This is the method we use as an AI SEO agency:
Agree a prompt set that reflects how buyers ask, for example "best [service] in [town]" or "[your product] vs [competitor]".
Record a baseline in the first month across the tools that matter to you, such as ChatGPT, Perplexity and Google AI Overviews.
Score each prompt as mentioned, cited (named and linked as a source) or absent, and note which competitors and sources appear instead.
Run the same prompts monthly and report the trend over several months.
Connect it to business results by tracking Search Console clicks and enquiries alongside AI visibility.
For Google, there is a useful detail in the documentation: traffic from AI Overviews and AI Mode is included in your overall Search traffic in Search Console, in the Performance report under the "Web" search type. It is not broken out separately, which is another reason to run your own prompt tracking.
Does GEO replace SEO?
No. Google states that standard SEO best practices are what make a page eligible for its AI features, and most other assistants rely on pages they can find and crawl. A site with indexing problems, thin content or poor internal linking will struggle in AI answers for the same reasons it struggles in search results.
Think of generative engine optimisation as an extra layer: the same foundations, with more attention to clear answers, verifiable facts, consistent entity information and third-party mentions. If your site still has basic ranking or indexing problems, fix those first through core SEO services, then build the GEO layer on top.
How long does generative engine optimisation take?
It depends on the change. Crawler access fixes are picked up the next time AI tools crawl the site. Page and listing changes usually take weeks to months to show up in answers. Third-party mentions build over months, because they depend on other people publishing about you.
Be sceptical of anyone who promises faster or guaranteed results. Nobody can guarantee that a specific AI tool will mention your business, and the answers themselves change as the products evolve.
What are the most common GEO mistakes?
Buying "AI optimisation" add-ons, special files or markup that Google says are not required.
Blocking AI crawlers through default CDN or firewall settings without realising it.
Writing vague, keyword-heavy copy instead of specific answers. The research suggests keyword stuffing can reduce visibility.
Ignoring third-party mentions and relying only on your own site.
Adding schema that does not match the visible page.
Judging progress from a single prompt run instead of a monthly trend.
Treating GEO as separate from SEO, when it depends on the same foundations.
Find out how AI tools see your business
Generative engine optimisation starts with knowing where you stand. Invisio Solutions can review your site, your business details across the web and your crawler settings, and give you a written AI search plan within two working days showing which areas are weakest.
Contact Invisio Solutions to request your free audit, or see how we work as an AI SEO agency.
Frequently asked questions
- Generative engine optimisation (GEO) is the practice of making a business easy for AI tools such as ChatGPT, Perplexity and Google AI Overviews to find, understand and cite in their answers. It builds on classic SEO with clearer answers, verifiable facts, consistent business details and third-party mentions.
