How to get your business cited by ChatGPT and Google AI Overviews

To get your business cited by ChatGPT and Google AI Overviews, you need pages that are easy for an AI to quote: clear direct answers near the top, named sources, real statistics, and quotations from identifiable people. Peer-reviewed research from Princeton, IIT Delhi and the Allen Institute for AI found that adding quotations, statistics and cited sources can lift visibility in AI-generated answers by up to 40%, while keyword stuffing did almost nothing. You also need the basics in place, because brands ranking on page 1 of Google show a strong correlation with being mentioned by ChatGPT. This practice is called generative engine optimisation, or GEO, and for UK small and mid-sized businesses it is one of the few marketing channels where the playing field currently tilts towards smaller players.

"GEO is mostly SEO done properly, plus formatting an AI can quote. The page that reads like a source gets cited. The page that reads like an advert gets skipped."

Dean Cookson, founder, Operosus

What is generative engine optimisation, and how is it different from SEO?

Generative engine optimisation is the practice of making your content more likely to be used, quoted and linked when an AI assistant answers a question. Classic SEO competes for a ranked position on a results page. GEO competes for a citation inside a written answer, where the AI has already synthesised several sources and your reward is a mention, a link, or being the source of the recommendation itself.

The term comes from a 2024 academic paper, GEO: Generative Engine Optimization, presented at KDD 2024 by researchers from Princeton, IIT Delhi, the Allen Institute for AI and Georgia Tech. We have written a plain-English walkthrough of the Princeton study if you want the findings without the methodology. They built a benchmark of 10,000 real-world queries and tested nine different content changes to measure which ones made a page more visible in AI-generated answers.

The two disciplines overlap heavily. The pages AI assistants cite are overwhelmingly pages that already perform in search, so GEO is best understood as a layer on top of solid SEO, not a replacement for it. The difference is in what you optimise for: search engines reward relevance signals, while generative engines reward content that is quotable, attributable and verifiable.

Does my Google ranking still affect whether ChatGPT mentions me?

Yes, and the effect is large. Seer Interactive ran 10,000 industry questions through GPT-4o and measured which brands were mentioned in the answers. Brands ranking on page 1 of Google showed a correlation of roughly 0.65 with being mentioned by the LLM, which is strong for this kind of study. Bing rankings showed a slightly weaker but still meaningful relationship.

The surprise in the same study was backlinks. The Seer team wrote: "We expected backlinks to play a big role, but their impact was weak or even neutral." Link building, the activity UK agencies have sold hardest for two decades, did not predict whether a brand appeared in AI answers. Ranking for the question itself did.

The practical reading for an SMB:

• If you already rank on page 1 for your money queries, you are most of the way to AI visibility, and GEO tactics decide whether you get quoted.
• If you rank nowhere, no amount of GEO formatting will rescue an invisible page. Fix search visibility for a small set of specific questions first.
• Stop judging content investment purely by backlinks earned. Judge it by the questions it can answer outright.

What actually makes an AI engine cite a page?

The Princeton-led team tested nine tactics across their 10,000-query benchmark and measured the change in how much of each AI answer drew on the modified page. Three tactics clearly won, and they are all about evidence rather than style:

1. Add quotations. Quoting named people or organisations produced around a 40% relative improvement in position-adjusted visibility, the strongest result in the study.
2. Add statistics. Replacing vague claims with concrete numbers produced roughly a 30% improvement.
3. Cite sources. Linking claims to credible references also produced roughly a 30% improvement.

Just as useful is what failed. Keyword stuffing, the old trick of repeating the target phrase throughout the page, showed little to no improvement in AI answers. Generative engines are reading for substance they can reuse, not term frequency.

One finding matters more than any other for smaller UK businesses. When the researchers applied source citation to pages sitting fifth in the underlying search results, those pages saw a 115% increase in visibility within AI answers, while top-ranked pages actually lost share. The authors put it plainly: "GEO presents an opportunity for these small content creators to significantly improve their visibility." An AI answer does not have to respect the existing pecking order. It picks the page with the best evidence, which is rarely something only an enterprise budget can produce.

How do I apply these tactics to a normal business website?

Here is what each evidence tactic looks like on the kind of pages a UK SMB actually has:

None of this requires new technology. It requires editing the pages you already have so that every important claim is specific, attributed and checkable.

Does it matter how old my content is?

Yes. Ahrefs analysed roughly 17 million cited URLs across ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews and organic Google results. Pages cited by AI assistants averaged 1,064 days since publication, against 1,432 days for pages in organic results: AI citations were 25.7% fresher. ChatGPT showed the strongest freshness preference of all the platforms, citing pages roughly 458 days newer than what organic Google surfaced for the same topics.

For a small business this points to a maintenance habit rather than a publishing treadmill:

• Update your highest-value pages on a cycle, refreshing figures, dates, prices and examples.
• Show a visible and truthful "last updated" date. The same Ahrefs dataset found AI-cited pages were also more recently updated than organic results.
• Prioritise refreshing the handful of pages that answer commercial questions over publishing thin new posts. A genuinely updated page is fresh in the way that matters.

How do I find out whether AI tools already mention my business?

Measure before you optimise, and you can do it without buying a tool:

1. Ask the engines directly. Put your real customer questions to ChatGPT, Google (watch for AI Overviews), Perplexity and Copilot. Note who gets named and which pages get linked. Repeat monthly, because answers vary.
2. Check your analytics for AI referrals. In GA4, look for referral traffic from domains such as chatgpt.com and perplexity.ai. Volumes are usually small but they tell you citations are happening.
3. Watch which pages AI traffic lands on. Those are the pages the engines already trust. Strengthen them first.
4. Log competitor mentions. If a rival is consistently named for your core service, fetch their cited page and compare it against the evidence checklist above. The gap is usually visible within a minute of reading.

Is GEO worth it for a small UK business, or is it an enterprise game?

The evidence so far suggests the opposite of an enterprise game. The Princeton benchmark's clearest finding was that lower-ranked sites gained the most from evidence-based optimisation while incumbents lost share. The work involved is editorial, not technical: tightening answers, adding real numbers, naming sources, quoting people. That is an afternoon per page, well within reach of a business with no marketing department.

It also compounds with work you should be doing anyway. Every change that makes a page more citable by an AI also makes it more persuasive to the human who lands on it: specific claims, named evidence, a direct answer at the top. Unlike a Google algorithm update, there is no penalty waiting on the other side of this work, because the tactic is simply being more useful and more verifiable than the page next to you.

Be wary of anyone selling certainty here. AI answers are not stable, the platforms do not publish how citation selection works, and the research base is young: one peer-reviewed benchmark, a handful of large correlation studies, and a lot of vendors extrapolating. A current example of the hype outrunning the evidence is llms.txt, which we have tested and called as a dud. Anchor your decisions to the studies linked in this guide, not to screenshots of single answers.

Where to start

Pick your five most commercially important questions: the things customers actually ask before they buy from you. Then, for each one:

1. Check whether you rank anywhere on page 1 of Google for it. If not, that page needs SEO work before GEO formatting will pay off.
2. Rewrite the page to open with a direct, self-contained answer under a question-shaped heading.
3. Add one real statistic with a linked source, and one quotation from a named person.
4. Update anything stale on the page, and date the update honestly.
5. Ask ChatGPT and Google the question a month later and record whether you appear.

That loop, run across a handful of pages each month, is generative engine optimisation as the evidence currently supports it. The businesses that get cited will be the ones whose pages read like sources, not adverts. The flood of generic AI content makes this easier, not harder, for businesses with real evidence: we explain why in our piece on the AI content sludge problem. And if your website's foundations are the blocker, Idea3 is our rebuild service with search and AI visibility designed in from the start.