The rise of GEO

What GEO is

Generative engine optimisation, or GEO, is the work of making a website easy for AI assistants to read, trust and cite. The name comes from a 2023 research paper, later presented at KDD 2024, which tested what makes a source more visible inside answers written by AI.^[4]

Search engine optimisation (SEO) aims at a link in a list of results. GEO aims at being one of the sources an answer is built from, in ChatGPT, Claude, Gemini, Perplexity, or the AI Overviews above Google’s results. The two share their foundations: pages crawlers can reach, a clear statement of what each page is about, structured data that describes it, and content that answers the question.

AI assistants are now mainstream

AI assistants have gone from a novelty to something most people use. These are the most recent figures each company has published itself.

The measures differ, weekly users for one and monthly users or queries for others, so the figures are not added together or ranked. ChatGPT alone grew from more than 900 million weekly users in February 2026 to more than a billion by August.^[5]

In the UK, 54% of adults now use AI tools such as ChatGPT, Copilot or Gemini, rising to 79% of 16 to 24 year olds.^[8] ChatGPT had 1.8 billion UK visits in the first eight months of 2025, against 368 million in the same months of 2024.^[9]

Search has not gone away

AI did not replace search. Google handles more than 5 trillion searches a year.^[10] It had 89.94% of worldwide search in September 2026, and 91.39% in the UK.^[11] Four in five UK adults (82%) use Google Search, which handles about 3 billion UK searches a month.^[9]

In February 2024 Gartner predicted that traditional search engine volume would drop 25% by 2026.^[12] Google’s own figures do not show that fall. What changed is what happens after the search: more of the answer now appears on the results page itself.

Fewer clicks reach websites

The clearest evidence comes from Pew Research Center, which studied the browsing of 900 US adults in March 2025. When Google showed an AI summary, people clicked a normal search result in 8% of visits, against 15% when it did not. They clicked a link inside the summary in just 1% of visits, and ended their browsing on 26% of pages with a summary, against 16% without.^[3]

Studies of clicks on Google point the same way:

• Ahrefs found that an AI Overview goes with a 58% lower average click-through rate for the top-ranking page, across 300,000 keywords.^[13]
• Seer Interactive found organic click-through on searches with an AI Overview fell from 1.76% to 0.61% between June 2024 and September 2025.^[14]
• SparkToro found 68.01% of US Google searches ended without a click in the first four months of 2026, against 60.45% in 2024.^[15]

In the UK, about 30% of searches now show an AI Overview, and 53% of adults say they see these summaries often.^[9]

These studies show correlation, not cause, and most of them measure the US. Semrush found that AI Overviews do not automatically increase searches that end without a click,^[16] and Google says clicks from results pages with AI Overviews are of higher quality.^[17] But they agree on the direction: fewer clicks to go round, and more of them going to the sources an answer cites. Seer found that brands cited in an AI Overview got 35% more organic clicks than those that were not.^[14]

AI visitors are few, and worth more

Visits from AI assistants are still a small part of most sites’ traffic: 0.25% of a typical site’s visits in Ahrefs’ study of 81,947 sites, though that share grew 9.7 times in a year.^[18] Similarweb counted more than 1.1 billion referral visits from AI platforms in June 2025, up 357% on a year earlier.^[19]

What stands out is how those visitors behave. Adobe, measuring US retail sites, found AI traffic grew 393% in the first quarter of 2026, and that shoppers arriving from AI stayed 48% longer and viewed 13% more pages.^[20] Over the 2025 holiday season, AI tools drove a 693.4% increase in traffic to retail sites.^[21]

Semrush estimates the average visitor from AI search is worth 4.4 times a visitor from traditional organic search, by conversion rate, and forecasts that for the topics it studied, AI search may send more visitors than traditional search by early 2028.^[22]

McKinsey found half of US consumers already use AI-powered search, with $750 billion of US revenue expected to flow through it by 2028, and projects that unprepared brands could lose 20 to 50% of their traffic from traditional search.^[23]

What gets a website cited

The GEO paper tested different ways of rewriting a source on a benchmark of 10,000 queries, and measured how much of each AI answer drew on it.^[4] The methods that added substance won:

• Citing sources, adding quotations and adding statistics improved visibility by 30 to 40%; clearer, more fluent writing by 15 to 30%. Overall, the methods raised visibility by up to 40%.^[4]
• Keyword stuffing did not help: it scored below the unchanged source, and did 10% worse than it on Perplexity. A more persuasive or authoritative tone made no significant difference.^[4]
• Lower-ranked sites gained most. Citing sources raised the visibility of the fifth-ranked result by 115.1%, while the top-ranked result lost 30.3%.^[4]
• The methods held up on a live engine, Perplexity, with gains of up to 37%. The best pair, fluency plus statistics, beat any single method by more than 5.5%.^[4]

McKinsey adds a warning: a brand’s own website is often only 5 to 10% of the sources AI search draws on, and only 16% of brands track how they appear in it.^[23]

AI crawlers are the gate

An AI assistant can only cite a page its crawler is allowed to fetch. Those crawlers take far more than they send back. In July 2025, Cloudflare counted these pages crawled for every visit referred:^[24]

80% of AI crawling over that year was for training models, not for answering a question.^[24] But crawling triggered by someone asking an AI a question grew more than 15-fold in 2025, and AI crawlers such as GPTBot, ClaudeBot and CCBot are the ones websites most often block outright.^[25]

The choice matters. Blocking training crawlers is a reasonable decision; blocking the crawlers that fetch a page when someone asks about it keeps a business out of the answer. Our audit scores the second, not the first.

Some proposed fixes have not shown an effect. Ahrefs found 28% of the domains it measured publish an llms.txt file, but 97% of those files received no requests at all in May 2026.^[26] Google says there are no extra requirements or special optimisations needed to appear in AI Overviews or AI Mode.^[17]

The UK picture

• 54% of UK adults use AI tools, and 79% of 16 to 24 year olds.^[8]
• 82% of UK adults use Google Search, and about 30% of UK searches show an AI Overview.^[9]
• 78% of UK businesses have a website, up from 68% in 2023 to 2024; 76% of sole traders do.^[27]
• The most common reason UK businesses use AI is to research information, often in place of a search engine such as Google (28%).^[27]

People now look for a UK business in two places: a search engine, and an assistant that answers for them. The second reads the same website as the first.

What it means for a business website

The evidence points to the same few things, and none of them is a trick:

1. Let the right crawlers in. Allow the crawlers AI assistants use to answer questions, and make sure your content is in the HTML before any JavaScript runs. (V27)
2. Answer first. Open each section with a direct answer to its heading’s question. (V10)
3. Back what you say. Specific figures, sources and quotations made the biggest difference in the research. (V1)
4. Describe yourself in structured data that matches what the page says, and keep your business’s identity consistent everywhere. (V6, V8)
5. Keep the search basics right. Titles, descriptions, canonicals and speed still decide whether Google shows you at all. (V9, V13, V19)
6. Measure each assistant on its own, and judge by what happens, not what is promised.

That is the work Martar Search Solutions does: see how it works.

How we put this together

We read every figure here on the page of the organisation that produced it, on 23 October 2026, and left out figures we could find only in news reports or could not open at the source. The one exception, Perplexity’s query count, is labelled as a report of its chief executive’s statement.

Most studies of clicks and traffic are from US data and from companies that sell search tools. They show what goes together, not what causes what. Where a figure is a forecast, we say so.