Definition
Third-party mentions are references to a brand on websites and platforms the brand does not control: news coverage, review sites, comparison articles, listicles, forums, podcasts and social posts. In AI search they work as independent evidence. When a model researches a category, most of the pages it reads are written by someone other than the brands it ends up recommending.
The best public numbers on this come from two studies. AirOps (October 2025) classified 21,311 brand mentions from GPT-5, Claude Sonnet 4.5 and Perplexity Sonar across more than 500 commercial discovery queries and found 85% were attributed to third-party domains, 13.2% to the brand's own domain and about 2% to no source at all. Nearly 90% of those third-party mentions came from listicles, comparisons or reviews. Ahrefs (May 2025) studied 75,000 brands and found branded web mentions had the strongest correlation with Google AI Overview visibility (Spearman 0.664), well ahead of backlinks (0.218). A follow-up Ahrefs analysis of the 50 most-mentioned sites in June 2025 found the correlation much weaker for ChatGPT (0.15) and Perplexity (0.30), so the effect is real but uneven across engines, and correlation is not proof of cause.
Mentions reach answers in two ways. Through retrieval, a model runs a query fan-out, pulls up the top listicles and reviews for each sub-query, and lifts the brands those pages name. Through training, repeated mentions across the web shape the model's parametric knowledge of which brands belong in a category. Agreement across many independent sources is what builds entity consensus. Note that a mention is not the same as a citation: a mention is your brand named in the answer text, a citation is a URL attached as a source, and a third-party mention often produces a mention of you alongside a citation of someone else's page.
Not every mention carries the same weight. The page has to be one that models actually retrieve for the prompts you care about, and the source mix differs per engine and changes over time. Citation data from early October 2026 showed social platforms making up 0.46% of ChatGPT citations, against 12.2% in Google AI Overviews, so a Reddit-heavy plan works very differently on the two surfaces. See UGC citations for that subset.
To act on this, start from the pages models already cite rather than from a generic PR target list. Collect the third-party URLs cited for your tracked prompts, check which ones mention competitors but not you, and pitch those authors with a concrete reason to include you. Promptwatch's Offsite Mentions view runs this check by scanning cited pages for each tracked brand. Then measure the effect over weeks with mention rate, AI share of voice and citation share. Our guide on how to get your brand mentioned by AI covers outreach tactics in more detail.
Examples of Third-Party Mentions
- A CRM vendor finds that the five listicles ChatGPT cites most for 'best CRM for startups' name four competitors and leave it out. It sends updated product details to two of the authors, and once the pages are re-analyzed its name appears on both.
- A fintech brand's own blog is rarely cited, but a comparison article on a personal finance publication that mentions it shows up as a source in a large share of its tracked Perplexity answers, which explains most of its visibility there.
- After a product recall, news coverage of the recall becomes the most cited third-party source for a brand's category prompts, and AI answers start naming the recall next to the brand.
- A B2B team that counted Reddit threads as its main ChatGPT lever sees almost no Reddit citations in ChatGPT in October 2026 and moves the effort to review sites and industry publications.
