Want to Accelerate the Growth of Your Business with AI? Talk to Us
Why brand mentions in AI search beat your own website
· Updated

When an AI engine mentions a brand, the page it got that from is somebody else's 85% of the time.

That figure comes from AirOps' analysis of 21,311 brand mentions across ChatGPT, Claude and Perplexity, and the same dataset found brands were 6.5 times more likely to be mentioned through a third-party source than through their own content.

For a UK small business that has spent five years polishing its website, that is an uncomfortable number. It means the review page, the Checkatrade listing, the Reddit thread and the trade body register are doing more to decide whether ChatGPT names you than your homepage is.

This piece sets out the evidence, explains why the engines behave this way, and then gets specific about which mention sources feed which engine and what a UK SME should do about each one.

What the citation studies found

Every large study of AI citations published since mid-2025 puts user-generated and third-party platforms at the top of the list, with brand sites nowhere near it. The studies differ in method and sample, but they agree on the shape.

Profound analysed 680 million citations between August 2024 and June 2025. Wikipedia alone was 7.8% of all ChatGPT citations. On Perplexity the top source was Reddit at 6.6%, followed by YouTube at 2%. G2, the software review site, was 1.1% of ChatGPT citations, and Yelp was 0.8% of Perplexity's. These are single domains taking a bigger share than entire industries.

Semrush ran 230,000 prompts weekly from July to October 2025 and logged over 100 million citations. Reddit and LinkedIn made the top five on ChatGPT, Google AI Mode and Perplexity. On AI Mode, LinkedIn appeared in nearly 15% of responses. In early August, Reddit appeared in close to 60% of ChatGPT answers, falling to around 10% by mid-September when OpenAI changed something.

Peec AI looked at 30 million cited sources in March 2026 across five engines. Reddit and YouTube were in the top five on all of them. Yelp was in the top five for AI Mode, G2 for Perplexity, and the overall top ten contained no brand website at all.

The Ahrefs study is the one I would show a sceptical owner. Ahrefs measured 75,000 brands against their AI Overview visibility and found the strongest correlation was with branded web mentions, at 0.664. Backlinks managed 0.218. Louise Linehan, who wrote it up, was careful about causation but still concluded:

"Online brand mentions appear to be more impactful than link building." Louise Linehan, Ahrefs

None of these studies is UK-specific, and Profound's data put .uk domains at 2.16% of ChatGPT citations against 80.41% for .com. I will come back to the UK evidence, because there is now some.

Why engines corroborate across sources

AI engines prefer third-party pages because a brand's own site is the one source that cannot be trusted to be neutral about the brand. Everything in the retrieval pipeline pushes that way.

Start with how a query is handled. Google's own documentation says AI Overviews and AI Mode may use a fan-out method, "issuing multiple related searches across subtopics and data sources". A question like "recommend a trustworthy plumber in Northampton" becomes several searches: who is in Northampton, who is Gas Safe registered, who has good reviews, who do people on forums recommend. Your site can answer the first. Other people's pages answer the rest.

Then there is the grounding step. A model asked to recommend a business is, in effect, being asked to vouch for it. A page that says "we are reliable" is a claim. A Checkatrade profile with 140 reviews, a Gas Safe register entry and a Reddit thread where three locals name the same firm are evidence, and they come from parties with no reason to flatter you.

The engines also lean on the sources they have learned are rarely wrong in aggregate. Reddit is noisy, but across a city's worth of threads the same names recur. Wikipedia is edited by people who argue about every sentence. Trade registers are maintained by regulators. These are the corpora the engines trust, which is why they keep showing up in the citation studies.

The same AirOps research found a second effect worth knowing. Brands that earned both citations and mentions were 40% more likely to resurface in later answers than brands cited only once. Being mentioned across sources does not only get you into the answer. It keeps you there. The broader survey of 45 generative engine optimisation studies rates this evidence as suggestive rather than proven, because the controlled test has not been run.

What the UK evidence says about local firms

The only UK-specific citation data I have found is from Whito, a small independent research publisher, and it is worth reading because it contradicts the global studies in useful ways. Whito examined 159 Google AI Mode answers to "recommend a trustworthy [trade] in [town]" across eight sectors and twenty UK towns, collected in August 2026.

The business's own website was cited in 143 of the 159 answers, 90%. So for local recommendations on Google's engine, your site does still get cited. But Google reviews or Business Profiles appeared in 128 answers (81%), trade directories and review platforms in 53%, regulators and trade bodies in 45%, and Reddit or Facebook in 32%. In 156 of 159 answers, the website or the Google profile or both were present.

The sector split is the part that should change how you spend. Checkatrade appeared in 50 of the 60 answers for electricians, plumbers and roofers. Across 79 professional services answers (accountants, estate agents, dentists, vets) it appeared once. Estate agents had the highest Reddit and Facebook share at 55%.

Two more figures from the same study. Trustpilot appeared in 2 of 159 answers. Local news outlets appeared in none. That does not mean a Trustpilot profile or local press coverage is worthless, but for "recommend a local tradesperson" on Google AI Mode in 2026, they are not what the engine reaches for.

A smaller Whito log from July 2026 ran the same prompt 18 times each on ChatGPT and Gemini. Reddit was ChatGPT's most-cited source, in 10 of 18 answers, with business websites second at 9. Gemini showed no sources at all in 13 of its 18 answers, and when it did cite, it cited only trade directories and the Gas Safe register. Yell, Yelp and Facebook were cited zero times on either engine.

These are small samples on single runs, and the publisher sells a visibility check service. Treat the percentages as indicative. The pattern, though, matches what the big studies found: the engines go to Google reviews, directories, registers and forums first.

Which mention source feeds which engine

Mention sources are not interchangeable, and the table below sets out where each one has been shown to appear. "Strong" means the source was in the top five or appeared in a third or more of answers in at least one named study; "weak" means it was measured and barely registered.

Mention source Google AI Mode / AI Overviews ChatGPT Perplexity Gemini Evidence
Google Business Profile and reviews Strong (81% of UK local answers) Not measured directly Not measured directly Likely, same index Whito, 159 answers
Checkatrade, TrustATrader, MyBuilder Strong for trades (83%), weak for professions Appeared, not top Not measured Only type cited when Gemini cited at all Whito UK, Whito July log
Reddit Strong (top five, 24% of UK local answers) Strong (top two globally, 10 of 18 UK answers) Strongest single source (6.6%) Top five Peec, Profound, Whito
Trade bodies and registers (Gas Safe, NICEIC, SRA, GDC) Strong (45% of UK local answers) Not measured Not measured Cited (Gas Safe) Whito UK
LinkedIn Strong (about 15% of responses) Strong (top five) Top five Not in top five Semrush, Peec
Facebook Top five Zero in UK log Not in top five Zero in UK log Peec, Whito
Review platforms (G2, Yelp, Trustpilot) Yelp top five on AI Mode G2 1.1% of all citations G2 and Yelp in top ten Not in top five Profound; Trustpilot 2 of 159 UK local answers
Wikipedia Weak (0.6% of AI Overview citations) Strongest single source (7.8%) Weak (0.8%) Top five Profound, Semrush
YouTube Strong (top five, 1.9%) Not in top ten 2% Top five Profound, Peec
Local and trade press Zero in 159 UK local answers Forbes and Reuters in top ten for national topics Forbes top ten Forbes, PCMag top five Whito UK, Profound

Two readings of this table. If you are a trade, your Google profile and Checkatrade are most of the game on Google's engines, and Reddit covers ChatGPT. If you are a professional firm, directories barely count, so your regulator's register, Google reviews, LinkedIn and forum presence carry the load.

An action list for a UK SME

Work the sources in the order the engines use them, which for most UK firms means Google first, then the register, then the directories, then the forums.

  1. Fix the Google Business Profile before anything else. Correct category, services, opening hours, the same business name and address as your website, and a steady flow of reviews. Google's own guidance says you may ask customers to review you via a link or QR code, but offering anything in exchange is prohibited, and so is paying to remove negative ones. Ask every customer, reply to every review.

  2. Check your regulator or trade body listing. Gas Safe, NICEIC, the SRA, the GDC, ICAEW, Propertymark. Whito found registers were named 111 times across 159 answers. If your entry is out of date, has a trading name that differs from your website, or is missing a branch, the engine has a reason to doubt you.

  3. Pick the directory the engines read for your sector, and only that one. For trades it is Checkatrade, with TrustATrader and MyBuilder behind it. For professional services the data says directories are close to irrelevant, so skip the subscription and spend the money on the next two items. Yell appeared in 3% of the UK answers. Yelp in none.

  4. Build a Reddit and forum presence that would survive a moderator reading your post history. Reddit is the one source that was strong on every engine measured. The way to do it is for a named person from the firm to answer questions in the local subreddit and the trade subreddit without linking to the business, for months, until other people start naming you. Forum posts that read like adverts get removed and do you no good.

  5. Make your entity consistent everywhere. The engines resolve "Smith Plumbing", "Smith Plumbing Ltd" and "Smith's Plumbing & Heating" as three businesses unless every listing, your website footer, Companies House and your LinkedIn page use the same name, address and phone number. A short, factual "about" paragraph in the same words on each one helps the model match them up.

  6. Treat Trustpilot and local press as national or sector plays, not local ones. Both were near zero in the UK local study. They matter if you are an ecommerce brand selling nationally online, where review platforms do appear in ChatGPT and Perplexity's top ten, or when the question is about your field and not your postcode.

  7. Put the people on LinkedIn, not just the company. Semrush's data has LinkedIn in about 15% of AI Mode responses and in ChatGPT's top five. A founder or senior person with a complete profile who posts about the work gives the engines a second named entity that points back to the business.

The common failure is spending on all of these at once and thinly. A plumber with 40 Google reviews, a live Checkatrade profile and a Gas Safe entry that matches is better placed than one with a presence on twelve platforms and ten reviews on each.

How to check what the engines currently cite about you

Run your own recommendation prompts and write down every source the engine shows, because the citation mix changes and the studies above are already months old. Ask each engine to recommend a firm like yours in your town, five or six phrasings, on ChatGPT, Perplexity, Gemini and Google AI Mode. Note which sources appear and whether you are named.

Do it monthly. Semrush's data showed ChatGPT's Reddit share falling from 60% to 10% in six weeks. A plan built on one month's citation pattern can be wrong by the next quarter. Whatever source the engines are leaning on for your sector this month is the one to feed. If you are not named, look at which third-party pages the named competitors appear on, because those are the pages you are missing from.

One decision rule covers most cases. If a potential customer asked a neighbour about you, the pages that neighbour would find to back up their recommendation are the pages the engine is also reading. Make sure those pages exist, agree with each other, and say what you would want said.

If you would like us to map which third-party sources the engines are citing for your sector and town, and which ones you are absent from, our AI visibility service starts with exactly that audit, and our local SEO work covers the Business Profile and directory side. Get in touch and tell us your trade and your town.

Share this post
Vijay Rathee
Vijay Rathee
Co-Founder

Open your inbox right now. Count the unread emails. If you are like most people, there are dozens of marketing emails sitting there, untouched, slowly sliding toward the archive button. Not because email marketing is broken.

View Bio