Generative Engine Optimisation (GEO) Explained: What Works and What Doesn't

Generative engine optimisation, or GEO, is the practice of making your business and content easy for AI assistants such as ChatGPT, Claude, Gemini and Perplexity to find, trust and cite. Most of it is good SEO plus stronger evidence. The parts that genuinely differ are earned third-party coverage and passages written to be quoted on their own.

What is generative engine optimisation?

Generative engine optimisation is improving how often, and how accurately, AI assistants mention and cite your business when they answer questions. It covers the content on your site, how clearly it states facts, how readable it is to AI crawlers, and what credible third parties say about you. The goal is to be the source an answer draws on.

The term comes from a 2023 research paper by Pranjal Aggarwal and colleagues, presented at the KDD conference in 2024. Other labels for the same idea include answer engine optimisation, AI SEO and LLM optimisation. The names differ; the aim is the same.

The shift is from rankings to references. Traditional search shows ten links and lets the reader choose. An AI assistant reads many sources and writes one answer, naming a few. Being one of the few is the prize.

Is GEO different from SEO?

Mostly no, partly yes. Google states that the best practices for SEO remain relevant for its AI features, with no special optimisation needed. But AI assistants weigh third-party evidence heavily and lift short, self-contained passages, so content and reputation work matter more than they did for classic rankings.

Google's position is clear. Its guidance for AI Overviews and AI Mode says: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." It adds that you do not need new machine-readable files, AI text files or special schema markup.

That is a useful corrective to a lot of GEO marketing. However, Google is one engine among several, and research on others shows real differences in what they favour. The practical view is that GEO is SEO done properly, with extra weight on two things: earned coverage and extractable answers.

What does the research say works?

The original GEO study found that adding citations, statistics and quotations from credible sources improved visibility in AI answers, by up to 40% in its tests, with effects varying by subject. In short, specific, sourced and quotable content gets cited more than vague claims.

The study, by Aggarwal and colleagues, tested how changes to content affected visibility in generative answers. According to its abstract, GEO methods raised visibility by up to 40%, and their efficacy varied across domains. Treat that as an upper bound from controlled tests, not a promise.

In practice, the patterns that follow from it are familiar to good editors:

• Answer first. Put the direct answer at the top of each section, in a passage that makes sense on its own.
• Be specific. Numbers with named sources beat general claims.
• Use real questions as headings. Headings that match how people ask make passages easier to match.
• Show your sources. Link to the original research, government data or documentation.
• Be honest about trade-offs. Balanced content reads as more trustworthy than promotion.

Why does earned media matter so much?

Because AI assistants lean heavily on what others say about you, not just what you say about yourself. A 2025 University of Toronto study found AI search engines show a systematic bias towards earned media, meaning third-party, authoritative sources, compared with Google's more even mix. Your own site corroborates; others' coverage persuades.

The study, by Chen, Wang, Chen and Koudas, compared AI search services with traditional search. Beyond the earned-media finding, it found engines differ in how fresh their sources are and how sensitive they are to the wording of a question. It also warned of a "big brand bias" that can disadvantage smaller players.

For a growing B2B company, earned media means reviews on credible platforms, listings in respected directories, mentions in industry articles, partner pages and podcasts. Each one is another independent source describing what you do. The guide to getting recommended by ChatGPT covers which sources to prioritise.

What does GEO not do?

GEO cannot buy a place in an AI answer, guarantee a recommendation, or work through a single file or trick. It is not a substitute for being genuinely good at what you do and clearly described by others. Claims that one tactic, such as an llms.txt file, will get you cited are not supported by the major engines' own guidance.

Three common overclaims to avoid:

Paying for placement. ChatGPT began testing labelled ads in 2026, shown separately from answers, and OpenAI says ads do not influence the answers themselves. Organic recommendation is not for sale.

Magic files. An llms.txt file can help some AI agents navigate a site, but Google says no special AI text files are needed for its AI features. The guide to llms.txt sets out what it does and does not do.

Instant results. AI assistants draw on what has been published and indexed over time. Changes to coverage take months to show.

How does a B2B company start with GEO?

Start by asking AI assistants the questions your buyers ask and recording who gets named and which sources are cited. Then fix your own content so it answers those questions clearly, make sure AI crawlers can read it, and build third-party coverage where the assistants already look. Measure monthly.

A practical sequence:

1. Test the prompts. List twenty questions buyers ask, run them across several assistants, record the results.
2. Check access. Make sure your robots file admits AI crawlers and key content does not depend on scripts.
3. Fix the answers. Rewrite key pages so each section opens with a clear, sourced, self-contained answer.
4. Earn coverage. Reviews, directories, partner pages and articles in the sources the assistants cited.
5. Measure. Re-run the same prompts monthly and track who is named.

The audience justifies the effort. According to TechCrunch, ChatGPT reached 900 million weekly active users in February 2026. Meanwhile, according to Pew Research Center, Google users clicked a traditional result in 8% of visits with an AI Overview, against 15% without.

Skein tracks how often the major assistants recommend a business as part of the outcomes it measures each month; the outcomes page shows the full set. That suits a B2B company whose buyers research online before contacting sales. A business that wins almost entirely through referrals may get less from it.

Frequently asked questions

What is generative engine optimisation?

Generative engine optimisation, or GEO, is the practice of making your business and content easy for AI assistants such as ChatGPT, Gemini and Perplexity to find, trust and cite in their answers. It combines clear, sourced content, crawler-readable pages and credible third-party coverage.

Is GEO the same as SEO?

Largely, but with different emphasis. Google says SEO best practices remain relevant for its AI features and no special optimisation is needed. AI assistants, however, weigh third-party evidence heavily and quote short, self-contained passages, so earned coverage and clear answers matter more than in classic SEO.

Does llms.txt help with GEO?

It may help some AI agents navigate a website, but it is not a ranking factor. Google states that you do not need new machine-readable files or AI text files to appear in AI Overviews or AI Mode. Clear content and third-party coverage matter far more.

How long does GEO take to work?

Expect months rather than weeks. AI assistants draw on content and coverage that has been published and indexed over time, so new reviews, articles and improved pages take time to influence answers. Track the same set of buyer questions monthly to see the trend.

What is the most important GEO tactic?

Earned third-party coverage is the most consistently important factor. A 2025 University of Toronto study found AI search engines show a strong bias towards earned media over brand-owned content. Clear, sourced answers on your own site matter too, mainly as corroboration of what others say.

Related reading