Unlinked brand mentions are the new backlinks for AI search

For AI search, how often the rest of the web mentions your brand matters more than how many sites link to you. In Ahrefs' May 2025 study of 75,000 brands, branded web mentions correlated with appearing in Google's AI Overviews at 0.664, while the number of backlinks correlated at 0.218. A December 2025 follow-up across ChatGPT, AI Mode, and AI Overviews found the same pattern on every engine. The likely reason is simple: language models learn which brands belong in an answer from how often and in what context those brands are written about, and a mention teaches that whether or not it carries a link.

What the data shows

The concentration is as striking as the correlations. In the May study, brands in the top quarter for web mentions averaged 169 mentions in AI Overviews, against 14 for the next quarter down, more than ten times as many. Branded search volume, at 0.392, also outranked every link metric. Over a quarter of the brands studied were never mentioned at all. AI visibility behaves like a winner-takes-most market, and mentions are where the winners separate.

Ahrefs is careful to say that "correlation isn't causation." Big, well-known brands have more mentions and more AI visibility for the same underlying reason: they are big and well known. The numbers do not prove that earning one more mention moves your visibility. What they do show is where links stop explaining the picture, and that is enough to rebalance a budget that is still mostly spent on link building.

This idea is older than AI search

Search engines have considered mentions for a long time. A Google patent granted in March 2014, "Ranking search results" (US 8,682,892), describes counting "implied links" alongside ordinary links, and defines one as "a reference to a target resource, e.g., a citation to the target resource, which is included in a source resource but is not an express link." A patent is not proof of what Google ranks on today. It does show that the idea of treating a citation without a link as a vote predates large language models by a decade. AI engines have made the idea central rather than marginal.

Not all mentions are equal

Three qualities separate mentions that plausibly help from mentions that do not:

• Context. A mention in an article comparing providers in your category teaches an engine what you are and who you compete with. A mention in a sponsor list teaches it almost nothing.
• Source. Engines draw heavily on a small set of sources per topic: industry publications, review platforms, community forums, and video. Our glossary calls this the citation oligarchy. A mention in one of those carries further than ten mentions on sites nobody cites.
• Consistency. If every mention describes you the same way, same name, same category, same location, the engine can resolve you as one entity. Inconsistent descriptions split you into several weak ones.

Local businesses see the same effect. In Whitespark's 2026 survey of local search experts, three of the top five factors for AI search visibility were about third-party mentions: presence on expert-curated "best of" lists, prominence on key industry domains, and the authority of unstructured citations. Its earlier study of local AI Overviews found that 60 percent of citations for Houston plumbing queries went to third-party sites rather than to the businesses themselves.

A practical program for earning mentions

1. Find the sources that already shape answers in your category. Ask ChatGPT, Perplexity, and Google the questions your buyers ask, and list every third-party domain they cite. Those domains are your target list.
2. Publish something worth citing. Original data, a clear point of view, or a genuinely useful tool earns mentions; a press release announcing a new hire does not. One credible piece of original research can be mentioned for years.
3. Make your expertise available. Answer journalist requests in your field, appear on industry podcasts, and speak at the events your buyers attend. Each appearance produces a page, a transcript, or show notes that describe you in context.
4. Show up where the community talks. Contribute genuinely in the forums and subreddits where your buyers ask questions. Promotion gets removed; useful answers get quoted.
5. Put your brand into video. YouTube mentions had the strongest correlation of any factor in the December study. Titles, descriptions, and spoken transcripts all count.
6. Standardize how you are described. Write one plain sentence that says what you are and use it everywhere: your site, your profiles, your bios, and the descriptions you send to anyone writing about you.
7. Reclaim links where it is natural, but do not make it the goal. A friendly note asking a writer to link a mention is fine; the mention was the valuable part.

Audit your mention footprint in an afternoon

You can get a useful baseline without buying anything. The aim is to see how the web describes you today, and how that compares with the two or three competitors that keep appearing in AI answers.

1. Search Google for your exact brand name in quotes, excluding your own domain with the minus-site operator. Note the first 50 results: who mentions you, in what context, and whether the description is accurate.
2. Run the same search for each competitor and count the mentions on the publications and communities your buyers read.
3. Search YouTube and the main forums in your field for the same names.
4. Write down every description of your company you find. If they disagree on what you do, where you are, or what you are called, fixing that is your first job.
5. Set up free alerts for your brand and your founders' names so new mentions reach you while there is still time to respond or ask for a correction.

What to avoid

Paid mention schemes and fake community accounts are the link farms of AI search. Forums remove them, review platforms penalize them, and an engine trained on a thread where your brand is called out for astroturfing has learned exactly the wrong thing. Earned mentions compound; manufactured ones are a liability that stays on the web.

How SEO for AI Agents measures this

We do not sell a mentions count, because a raw count without context says little. What we measure is what mentions are for. The source of citation check collects the actual URLs each AI engine cites when it answers questions in your category and groups them by domain, so you can see which publications, directories, and communities are feeding the answers, and whether you appear in them. That list is the target list from step one, built from receipts rather than guesswork.

The entity graph check covers the consistency half: it reads your Organization structured data and the identity profiles it links to, which is how engines connect scattered mentions back to one brand. For context on the wider strategy, read the glossary entries on digital PR and Wikidata, and our explanation of how AI engines pick sources.