What kinds of content get a brand mentioned in AI answers?

AI answers name a brand for two reasons: the model already knows it, or a page the engine finds in its search names it. So two kinds of content earn mentions. One is being written about widely across the web, which goes with stronger model recognition. The other is pages that match the questions buyers ask: your own pages and, especially for smaller brands, third-party pages that name you. FAQ blocks and schema markup show no measurable benefit once you compare pages within the same website.

This answer draws on published studies we reviewed in October 2026; we ran no tests of our own. The studies find associations, not proof that a type of content causes more mentions.

Why does an AI answer name one brand and not another?

An AI answer names a brand either from memory or from search. A 2026 study by Chen and Koudas tracked 158,099 answers from ChatGPT, Gemini and Perplexity. Its summary: "Pages are thus cited when searched; brands are also named when remembered."

A mention is when the answer names your brand in its text. A citation is a linked source the answer shows. The two often differ, as our guide to generative engine optimization (GEO) metrics explains.

The memory route is large. Almost half of ChatGPT's mentions of brands the question didn't name came with no cited page naming that brand. And when cited pages backed two brands equally, the brand the model already knew was far more likely to be named: at 5.8 times the odds.

So the question "what content earns mentions?" splits in two: what builds memory, and what wins the search.

What makes a model "know" a brand?

The best available evidence points to being written about across the web. Ahrefs studied 75,000 brands in Google AI Overviews, the AI summaries at the top of Google results. Mentions of a brand on other websites tracked its AI Overview mentions much more closely than backlinks did.

On a correlation scale where 1 means perfectly in step, web mentions scored 0.66 and backlinks 0.22. Ahrefs stresses that correlation is not causation, and its data covers only AI Overviews in 2025.

Memory builds slowly, though, and it doesn't cover every brand. For the rest, the search route decides.

Which pages does the search route pick up?

Search favours pages whose wording matches the question. In the Chen and Koudas study, every page ChatGPT cited had been found by its own search. Among pages on the same website, whether and how high search found a page explained 32% of the differences in citations. The pages' own features explained under 3%.

Pages whose titles matched the question more closely got about 14% more citations. A closer match in a page's best paragraph went with about 15% more.

Forum pages stood out, with 78% more citations than articles on the same site. List pages had a smaller edge, about 13% more. Comparison, how-to, pricing and product pages did not differ clearly from articles.

Lists come with a catch, though. Lists and comparisons were often found by the search, yet cited less often than other pages that ranked as high in the engine's search results.

Most of what engines cite is written by others. Muck Rack, a PR software company, analysed more than 25 million links from ChatGPT, Claude and Gemini in May 2026. It found that 84% of citations went to earned media: coverage the brand doesn't own or pay for, of which journalism is only part.

What this means: your own pages should answer your buyers' actual questions in their titles and paragraphs. But for smaller brands, your own site is rarely enough.

Why do third-party pages matter more for smaller brands?

Smaller brands have little memory to fall back on, so pages written by others carry more weight. Chen and Koudas repeated their analysis in a business-to-business market: proposal software for architecture, engineering and construction firms. There, the model behind ChatGPT named the small vendors from memory for at most 16% of queries. Large providers were named for over 60%.

Third-party pages filled that gap. When a cited third-party page named a small vendor, ChatGPT became far more likely to name it too: roughly 40 times the odds. In the consumer market, with better-known brands, the same kind of page raised ChatGPT's odds only about 4 times.

If placement and match matter this much, do formatting tricks add anything?

Do FAQ blocks, schema and summary boxes help?

Not within a website. Across websites, pages with FAQ blocks got about 18% more citations. Compare pages on the same site, though, and the advantage disappears. FAQ blocks and schema markup then showed no gain. Summary boxes showed a small one, too small to rule out chance. The authors' reading is that these features tend to appear on websites that are cited more for other reasons.

So formatting on its own is unlikely to move mentions. The bigger factors, being written about and being found, depend on which sources AI answers trust in your category.

How do you find which content works in your category?

Look at what AI answers already cite for your buyers' questions, engine by engine, because the mix varies. Muck Rack found that industry-trend questions cite journalism at more than twice the rate of how-to questions. It also found different favourite sources per engine: Wikipedia for ChatGPT, PubMed Central for Claude, Reddit for Gemini.

In practice, list the pages cited for your category's questions on each engine, and note where competitors are named and you are not. GEO tools' source reports can produce this list; check what the sources report is counting before relying on it.

Taken together, the evidence points to two jobs. Memory is built off-site, over time, by being written about. Search is won by pages that match buyers' questions, and for smaller brands those pages are often someone else's. Since none of this proves cause and effect, treat cited pages as a map of where to work, and watch whether your mentions move.