Someone asks an AI tool where to buy a new sofa, a furniture brand shows up in the answer and the brand’s marketing team assumes its website must have helped it get there – not necessarily though, AI platforms can mention a company without citing its website among the sources behind the answer. Basically the brand gets invited to the party but doesn’t make the guest list of sources.
According to Semrush’s AI Visibility Index 2026, which covered 22 industries in the UK and US, fewer than half of the top 100 mentioned brands also appeared among the top 100 cited domains, on average. Being mentioned and being cited are two different kinds of visibility and a brand can have loads of one without much of the other.
A Mention Is Not The Same As A Citation
A mention means a brand appears in an AI-generated answer, while a citation identifies a source the platform uses to support that answer. Leigh McKenzie, Director of Online Visibility at Semrush, explains the difference: “A mention tells us whether an AI platform is recommending or discussing your brand in response to a user’s question. A citation tells us which sources the platform is using to support its answer.”
A mention also doesn’t automatically mean an endorsement. A brand might be discussed without being recommended, so appearing in an answer doesn’t exactly tell the whole story.
The report’s Tesco and Asda figures show just how far apart mention and citation rankings can be. Both reached the top 100 for mentions on all four platforms in every month measured. Yet in July, Tesco ranked 25th among ChatGPT’s cited domains while Asda ranked 606th. On Gemini, they were much closer, ranking 29th and 60th.
More from Artificial Intelligence
- AI Could Move From The Cloud To The Phone – What Does That Mean For App Developers?
- What Happens When A Startup’s Entire Codebase Is Written By AI?
- The Most Important AI Design Decision Is Where The Model Stops
- OpenAI’s Astra Could Change How Startups Are Built, Says OpenAI’s Tricia Troth
- AI Influencers Vs. Real Humans: What Actually Works In Marketing?
- The Rise Of AI-Powered Start-Ups
- Trump’s Calling It “Superintelligence” Now – But The AI Industry Might Have Bigger Problems Than Its Name
- Are AI Agents Secretly Building Surveillance Files On Your Friends And Family?
How Often Does A Brand’s Own Website Go Missing?
A figure from McKenzie’s research found that only 13% of branded AI answers in the research linked back to the brand’s official website. This measures answers, not the percentage of brands receiving citations, but it shows how rarely a brand’s own site appeared as a source in those answers.
In the US, Chevrolet, Disney, Ford, Honda, Marvel, Nintendo and Toyota all appeared in the top 100 for mentions on every platform while remaining outside the top 500 for citations in every month measured. Being well known enough to appear in an answer didn’t guarantee a place among its cited sources.
Other Websites Are Doing Some Of The Talking
AI platforms don’t rely exclusively on a company’s own website when generating answers. As McKenzie explains, they draw on “a much broader ecosystem of information, including publishers, review sites, communities, and other third-party sources.”
A company’s website still helps establish who it is and what it offers, but independent coverage also influences how AI platforms understand and describe the brand. A retailer might publish detailed product information on its own site while reviewers and customers discuss how those products perform in practice. Those external sources can provide information that the company itself doesn’t.
McKenzie says the gap between a company’s own content and its wider online presence can affect which sources AI platforms cite. If the two tell different stories, models may cite external channels instead. That doesn’t make the company’s website irrelevant, but it does mean polishing its pages alone might not address every citation gap.
A Favourite On One Platform Isn’t Automatically A Favourite On Another
The research covered ChatGPT, Gemini, Google AI Overviews and Google AI Mode. In the UK, any two platforms shared only 63 to 70 of their top 100 mentioned brands on average. There isn’t one universal list of brands that every platform brings into its answers.
The platforms also differ in how they produce results. McKenzie notes that Gemini mentions more distinct brands per response, while ChatGPT cites a higher volume of external sources. Google’s AI experiences also draw on different information systems, which can influence the brands and websites that appear.
As McKenzie puts it, “Success on one engine doesn’t guarantee presence on another, so multi-platform measurement is essential.”
The Tesco and Asda comparison makes that difference tangible. ChatGPT placed hundreds of positions between their citation rankings, while Gemini ranked them much closer together. Checking only one platform could give a business a very different picture of its visibility.
Where Businesses Can Start Looking
McKenzie’s advice starts with a company’s own website. AI models need clear, structured content that answers questions directly, and technical or layout issues can make it harder for them to extract information from a page.
Businesses also need to look beyond their own domains. Reviews, news coverage and community discussions can all contribute to the information AI platforms use, so checking how a brand is described elsewhere can reveal gaps that a website audit alone won’t catch.
McKenzie also recommends looking at the questions people actually ask AI tools. Traditional SEO reports might track a short phrase such as “best accounting software”, while someone using AI might ask for the best accounting tool for a mid-sized agency handling multi-currency invoicing in Europe.
Someone with that specific brief isn’t looking for just any accounting tool; they’re looking for one that fits a specific business, handles a particular type of invoicing and works across a specific region. A report tracking “best accounting software” won’t tell you whether AI is getting those details right or not.
Prompt data can also show whether AI platforms describe a brand as reliable, report its pricing accurately or repeat outdated information. Businesses can then investigate what is shaping those answers instead of guessing which parts of their messaging need attention.
Who Is Telling AI About Your Brand?
Back to that sofa shopper. If an AI tool recommends the brand but cites a review site instead of its own website, the marketing team has a lead to follow. It can read the review, see what information the AI might be drawing on and check whether its own product pages cover the same details.
That still won’t explain every mention or citation, but it gives the team something concrete to investigate: the sources appearing alongside its name, not just how often the name appears.
