GEO in one paragraph
GEO stands for generative engine optimisation (most people search for the US spelling, generative engine optimization). It is the practice of improving how often, and how accurately, AI answer tools mention and cite your brand. Those tools build each answer from sources they find on the web, so GEO works on two fronts: your own site, which has to be clear, crawlable and worth quoting, and what other trusted sites say about you, which is where most citations come from. In practice that means earned coverage in the publications AI reads, pages that answer real customer questions, the same facts about your brand everywhere they appear, an honest presence on Reddit and review sites, and no technical barriers for AI crawlers.
GEO in numbers
- 58% of consumers have replaced traditional search engines with generative AI tools for product and service recommendations. (Capgemini, 2025)
- 84% of AI citations come from earned media, and about 27% are journalism. (Muck Rack, May 2026)
- Brand mentions across the web correlate with visibility in Google’s AI Overviews about three times more strongly than backlinks (0.664 against 0.218). (Ahrefs, 2025)
- 62% of the sources Google’s AI Overviews cite are not in the top 10 organic results. (Ahrefs, March 2026)
- 85% of the 1,094 categories Semrush studied in ChatGPT have no clear brand owner yet. (Semrush, July 2026)
- Adding citations, quotations and statistics lifted a source’s visibility in AI answers by up to 40% in the original GEO study. (Aggarwal et al., KDD 2024)
From JBH’s work: “Clifton is now surfaced and recommended across AI search for high-intent bridging queries, with supporting source visibility across up to 21 cited sites per query.” (Clifton case study)
What is GEO?
A generative engine is any tool that answers a question by writing a response, rather than handing you a list of links. ChatGPT, Gemini, Perplexity, Copilot, Claude, Google AI Overviews and Google AI Mode all work this way. They read a set of sources, summarise them and, more and more often, name the brands they think fit the question.
GEO is the work of influencing what goes into those answers. For a consumer brand, success looks like this:
- Your brand appears when someone asks for the best option in your category.
- The AI describes you correctly: what you sell, who it is for, what it costs, why people choose you.
- Your site, or coverage about you, is one of the cited sources, so curious readers can click through.
That matters because AI tools now sit at the start of a lot of buying research. If the answer only names three brands, fourth place gets you nothing.
Where the term came from
The name comes from a research paper. In November 2023, a team led by researchers at Princeton University and IIT Delhi published “GEO” on arXiv, and it was accepted to KDD 2024, one of the main academic conferences on data mining. (Aggarwal et al.)
The team built a benchmark of 10,000 queries and tested how small changes to a web page affected its visibility in AI-written answers. Three changes worked best: citing credible sources, adding quotations from relevant experts and adding statistics in place of vague claims. Those lifted visibility by up to 40% in their tests, and by up to 37% when they tried the same methods on Perplexity. Keyword stuffing, the old SEO habit, offered “little to no improvement”.
Two findings are worth keeping in mind. Pages with real evidence on them got used more. And lower-ranked sites gained most, because the AI worked from the content on the page rather than from signals such as backlinks.
Read those numbers with care: they come from a lab benchmark, and AI tools have changed a great deal since 2023.
GEO, AEO, LLM SEO and AI SEO
You will see several names for the same work. AEO (answer engine optimisation), LLM SEO, LLMO and AI SEO all describe improving how a brand shows up in AI answers. The differences are mostly about who coined the term and which tools they had in mind.
We use GEO because it is the most widely searched term and has a clear origin. For a side by side view of every term, and where each overlaps with traditional search, see SEO vs GEO vs AEO.
How do AI tools choose which brands to mention?
AI answers draw on two kinds of knowledge. One is what the model learned in training, which is broad but dated. The other is a live web search the tool runs at the moment you ask, which is how it picks up recent products, prices and coverage. Questions about what to buy usually need the live search, because the answer depends on current products and prices.
Google explains its own version plainly. Its AI features use its core ranking systems to retrieve relevant pages from the Search index, then generate an answer with links to the pages that support it. The model also runs “fan-out” queries: related searches it generates to gather more material. A question about weedy lawns might fan out into searches for herbicides, chemical-free removal and prevention. (Google Search Central, 2026)
That has a practical consequence. Your brand can be found through searches the customer never typed, as long as you are well represented across the topics around your category.
What the research says gets cited
| Signal | What the research found | Source |
|---|---|---|
| Earned coverage | About 84% of AI citations come from earned media and about 27% are journalism. | Muck Rack, 2026 |
| Brand mentions | Web mentions correlate with AI Overview visibility at 0.664, against 0.218 for backlinks. | Ahrefs, 2025 |
| Organic rankings | Page-one Google rankings were the strongest predictor of brand mentions in AI answers, with a correlation of about 0.65 across 10,000 GPT-4o answers. | Seer Interactive, 2025 |
| Beyond the top ten | Only 37.9% of AI Overview citations also appear in the top 10 organic results. | Ahrefs, 2026 |
| Freshness | AI assistants cite pages that are on average 25.7% fresher than the pages in Google’s organic results. | Ahrefs, 2025 |
| Community sites | Wikipedia, YouTube and Reddit together supply 15% of the sources in Google’s AI summaries. | Pew Research Center, 2025 |
Most of these studies are correlations published by tool vendors, so none of them proves cause. Read together, they point the same way. Ranking well still helps, but AI tools also weigh what independent sites say about you, and they prefer recent material.
The mix also differs by platform. ChatGPT leans heavily on news publishers. (Ahrefs, 2025) Google’s AI summaries cite news sites less often: Pew found 5% of them linked to a news website. (Pew) A plan that only targets one tool will miss the others.
What does GEO work involve?
Most GEO work will look familiar to anyone who has run SEO or PR. What changes is where the effort goes. For consumer brands, five areas do most of the lifting.
Coverage in the publications AI reads
If most citations come from earned media, the most direct route into AI answers is to be written about by the titles AI already trusts in your category. This is digital PR: data-led campaigns, expert commentary on breaking stories and products placed in the roundups shoppers read.
The target list changes, though. Muck Rack found the overlap between the journalists a brand pitches most and the journalists AI cites for that brand averages just 2%. (Muck Rack, December 2025) So GEO starts by asking ChatGPT, Gemini and Google the questions your customers ask, recording which publications they cite and making those titles a priority.
Each placement should also say plainly what your brand is and what it is known for. AI repeats what it reads. An article that names you as “a Manchester skincare brand known for fragrance-free moisturisers” gives it something accurate to repeat.
Our guide to digital PR for AI search covers the method in detail.
Clear pages that answer the question
Your own site is still a source. AI tools cite category pages, product pages, guides and FAQs when they answer the question better than anything else they can find.
- Answer the question near the top of the page, in a sentence that could be quoted on its own.
- Put the facts a shopper compares in plain text: prices, sizes, ingredients, delivery times, guarantees.
- Add evidence: your own data, named experts, first-hand testing. The original GEO study found this kind of material earned the biggest lift.
- Keep key pages up to date, and show the date.
Google is clear about what you can skip. You do not need to break pages into tiny “chunks” for AI, rewrite content in a special style or add schema written for AI. It recommends “non-commodity” content: first-hand experience and expert views that go beyond what everyone else has already said. (Google Search Central)
The same facts about your brand everywhere
AI tools build a picture of your brand, often called an entity, from every source they read. When those sources disagree about your name, what you sell, where you are based or who founded you, the answer gets vague or wrong.
Entity work is unglamorous and worth doing early:
- Use one brand name and one short description of what you do, and repeat them on your site, social profiles and press materials.
- Keep your About page factual: founding date, founders, locations, what you sell and who for.
- Fix out-of-date listings on directories, marketplaces and comparison sites.
- Keep your Google Business Profile and Merchant Center product data complete. Google says both can help products appear in AI responses. (Google)
Reddit and reviews
When people ask which product to buy, AI tools look for what real customers say. Google notes its AI features can show what is being said about products “across the web, including in blogs, videos, and forum discussions”. (Google) Reddit is one of the most cited community sites, and review platforms carry the opinions AI summarises when it explains why one brand is better than another.
The same Google guidance warns that “seeking inauthentic ‘mentions’ across the web isn’t as helpful as it might seem”. Fake accounts, planted threads and bought reviews are a reputational risk, and the systems that filter spam from search also feed the AI answers. The honest version of this work is:
- Find the threads and review pages AI already cites for your category, and read what they say about you and your competitors.
- Fix the real problems customers raise, then answer openly as the brand where community rules allow.
- Ask happy customers for reviews on the platforms that matter in your category.
- Feed the recurring questions and misconceptions back into your site content and PR angles.
Our community PR service covers audits, monitoring and transparent participation on Reddit, YouTube, forums and review sites.
Technical access for AI crawlers
An AI tool can only cite a page it is allowed to read. A blanket robots.txt rule or a bot-blocking setting at your CDN can shut AI crawlers out without anyone noticing. OpenAI, Anthropic and Perplexity each run a search crawler that is separate from any training crawler, so you can allow one and block the other.
| Crawler | Company | What it does |
|---|---|---|
| OAI-SearchBot | OpenAI | Surfaces sites in ChatGPT search. Sites that block it will not be shown in ChatGPT search answers. (OpenAI) |
| GPTBot | OpenAI | Collects content for training OpenAI’s models. Blocking it does not remove you from ChatGPT search. (OpenAI) |
| PerplexityBot | Perplexity | Surfaces and links sites in Perplexity’s results. Perplexity says it is not used to train AI models. (Perplexity) |
| Claude-SearchBot | Anthropic | Indexes content for Claude’s search results. Blocking it may reduce your visibility there. (Anthropic) |
| Google-Extended | A control token for Gemini training and grounding. It does not affect inclusion or ranking in Google Search. (Google) |
For Google’s AI Overviews and AI Mode, the rule is simpler: a page has to be indexed and eligible to show in Google Search with a snippet. (Google) Ordinary technical SEO covers that.
And llms.txt? Google says creating one “will neither harm nor help” your visibility in Google Search, because Google Search ignores it. (Google) It is harmless to add, but it is not where the gains are.
How does GEO relate to SEO?
Google’s position, published in May 2026, is that optimising for its generative AI features is optimising for search, “and thus still SEO”. (Google Search Central) For Google’s own AI surfaces, that is fair. AI Overviews and AI Mode are built on Google’s ranking systems, so a site with weak SEO will struggle to be cited there.
Where GEO adds something is emphasis and measurement:
- Off-site weighs more. In SEO, other sites matter mainly through links. In AI answers, what those sites say about you in words, and how often, carries real weight.
- More than one engine. ChatGPT, Perplexity and Claude run their own searches with their own crawlers and their own preferences.
- Different results to track. Mentions, citations and the way you are described, alongside rankings and clicks.
The safest way to think about it: good SEO is the foundation, and GEO is the work of making sure the wider web gives AI the right things to say about you. Most brands need both. See our SEO service for the search side.
The full comparison, including AEO and LLM SEO, is in SEO vs GEO vs AEO. For how Google’s own AI surfaces work, see Google AI Mode and AI Overviews.
What do honest GEO results look like?
Nobody can guarantee that ChatGPT or Google will mention your brand. Anyone promising a fixed position in AI answers is promising something they do not control. Google itself warns against third-party tools that “promise ranking success”, and says no outside tool has access to its internal ranking or AI systems. (Google)
AI answers also vary. Ask the same question twice, or phrase it slightly differently, and the brands named can change, so a single screenshot proves little. Useful measurement runs a fixed set of customer questions across several tools at regular intervals and tracks the trend.
What can be measured:
- How often your brand is mentioned for a set of target questions, and how that compares with competitors.
- How often your site, or coverage about you, is cited as a source.
- Whether the description is accurate.
- Branded search and direct visits, which often rise when AI recommends you even if no click is recorded.
- Your performance in Google’s AI features, using the Generative AI performance report in Search Console.
That fourth point matters. Similarweb found people who see a brand recommended by AI are 2.5 times more likely to visit its site within seven days, mostly through branded search. (Similarweb, 2026) And on Google, brands cited inside an AI Overview get a 35% higher organic click-through rate than brands that are not. (Seer Interactive, 2025)
What results look like for JBH clients
Nature’s Best: 83 placements, 56 of them linking to category pages. AI citations rose from 784 to 1,307 in six months. (Case study)
Spacegoods: 92 placements in titles AI reads. Share of search rose from 15% to 30% between March and August 2025, according to MyTelescope data. (Case study)
Clifton Private Finance: JBH mapped the 21 sites AI cites for bridging finance, and five high-intent queries now surface Clifton in AI search. (Case study)
The timescale in each case is months. Coverage has to be published, crawled and picked up before it shows in answers, and the gains build as coverage accumulates. Our guide to measuring AI visibility covers the tracking in detail.
How to start with GEO
Most brands can make progress with these steps, in this order.
- Find out where you stand. Our free AI visibility check asks ChatGPT 10 real customer questions and shows where you appear, and where you are left out. For a full review across more questions and engines, see the AI visibility audit.
- Write down the questions that matter. List the questions customers ask before they buy, in their words. These become the prompts you track.
- Check crawler access. Make sure robots.txt and your CDN allow OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot, and that key pages are indexed.
- Fix the facts. Make your About page, product data and listings consistent, and correct anything AI gets wrong about you.
- Map the sources. Record which publications, Reddit threads and review sites AI cites for your questions. That is your target list.
- Earn coverage there. Plan digital PR around those titles, with a steady flow of articles rather than one big campaign a year.
- Measure monthly. Rerun the same questions, track mentions and citations, and watch branded search.
If you would like help with any of it, our GEO service brings digital PR, SEO and community PR together for consumer brands.
Sources:
- Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, the original GEO paper (arXiv 2023, KDD 2024)
- Google Search Central, Optimizing your website for generative AI features on Google Search (2026)
- Google, common crawlers: Google-Extended
- OpenAI, overview of OpenAI crawlers
- Perplexity, Perplexity crawlers
- Anthropic, how site owners can manage Anthropic’s crawlers
- Muck Rack, What Is AI Reading? (May 2026)
- Muck Rack, earned media and AI citations (December 2025)
- Ahrefs, brand mentions and AI Overview visibility (2025)
- Ahrefs, AI Overview citations and the top 10 (2026)
- Ahrefs, the most cited domains in AI assistants (2025)
- Ahrefs, AI assistants and fresh content (2025)
- Seer Interactive, what drives brand mentions in AI answers (2025)
- Seer Interactive, AI Overview impact on CTR (2025)
- Semrush, ChatGPT topic authority study (2026)
- Capgemini Research Institute, consumer trends (2025)
- Pew Research Center, AI summaries and clicks (2025)
- Similarweb, AI visibility and site visits (2026)



