--- title: "How do brands build authority that AI search recognizes?" description: "By being written about by sources AI engines already trust, in the outlets each engine reads, and by keeping a site AI agents can read." canonical: "https://underneath.agency/resources/how-brands-build-authority-for-ai-search" published: 2026-10-07 updated: 2026-10-07 publisher: "Underneath (https://underneath.agency/agent)" entity: "https://underneath.agency/.well-known/entity.json" --- Guide · AI search # How do brands build authority that AI search recognizes? Brands build AI search authority mostly by being covered by independent sources the engines already trust, in the outlets each engine actually reads. Your own site still matters, but as the place where facts are checked, not as the voice that vouches for you. Measure it by where you appear in AI answers, not by backlinks alone. ## The short version 1. In tests of niche brands, 95.1% of the sources ChatGPT drew on were earned media, such as reviews and independent publications ([Chen and colleagues](https://arxiv.org/abs/2509.08919)). 2. In [our brand entity study](https://underneath.agency/research/brand-entity-ai-recommendations-study), each tenfold increase in independent sites naming a brand went with 4.7 times the odds of being recommended. 3. Products AI assistants recognized by name 99.4% of the time surfaced in only 3.32% of open “best tools” questions on ChatGPT ([Sharma](https://arxiv.org/abs/2601.00912), a small study). 4. In lab tests, fake authority claims worked, but when all rivals used them, the famous brand’s recommendation rate recovered to 93.8% ([Chu and Hou](https://arxiv.org/abs/2606.17443)). ## What does “authority” mean to an AI search engine? To an AI engine, authority is mostly what credible others publish about you, not what you say about yourself. [Chen and colleagues](https://arxiv.org/abs/2509.08919) compared AI search engines with Google across product categories. For niche brands, ChatGPT drew 95.1% of its sources from earned media and only 4.9% from brand sites. Perplexity was more mixed, with 17.5% from social platforms such as YouTube and Reddit. Google’s results kept a far larger share of brand-owned and social pages. A health study shows what kind of authority counts. [Jacques and colleagues](https://arxiv.org/abs/2601.17109) coded 615 sources ChatGPT cited for 100 consumer health questions in January 2026. Institutional sources such as hospitals, government sites and encyclopedias made up 75.7%. Yet 64.7% of the cited pages had no author attribution at all. The authority was the institution’s, not the byline’s. ## Why does authority matter differently in generative search? Because an AI answer names only a few brands, and a short list of trusted sources decides which ones. A ranked page of results lets a reader browse past the leaders. An AI answer does the browsing and hands back a verdict. In an analysis of a public dataset of 602 prompts, official, news and specialist sources made up 79.12% to 87.52% of citations on each platform ([Zhang, He and Yao](https://arxiv.org/abs/2604.25707)). Being one of the recognized sources is the entry ticket. Google rankings carry only part of that weight. In [our AI citations and Google rankings study](https://underneath.agency/research/ai-citations-google-rankings-study), only 8.3% of the pages ChatGPT cited ranked in Google’s top 10 for the question. Across six AI search engines and 55,936 queries, 37% of domains cited were unique to AI search and never appeared in Google or Bing results ([Zhang and colleagues](https://arxiv.org/abs/2512.09483)). For more on rankings, see [is traditional SEO still important for AI search?](https://underneath.agency/resources/is-seo-still-important-for-ai-search) ## Will authority matter more than keyword rankings? For being recommended, the evidence points that way: being known by name is not the same as being recommended. In [Sharma’s](https://arxiv.org/abs/2601.00912) test of 112 Product Hunt startups, AI assistants recognized products asked about by name 99.4% of the time on ChatGPT. Asked open questions such as “What are the best AI tools launched this year?”, ChatGPT surfaced them in only 3.32% of cases. Referring domains, a classic sign of others linking to you, predicted visibility on Perplexity. This is a small, single-author study. A vendor’s tracking data shows the same ladder. Household-name brands appeared in 73% of relevant AI answers on their first tracking run, established mid-market brands in 44% and small brands in 11% ([Kumar](https://arxiv.org/abs/2606.20065), 2026). Those tiers reflect fame built over years, which no keyword ranking supplies on its own. ## Should companies invest more in third-party coverage? Yes; independent coverage is the strongest signal of being recommended that we have measured. In [our brand entity study](https://underneath.agency/research/brand-entity-ai-recommendations-study), each tenfold increase in the number of independent sites naming a brand in the cited pages went with 4.7 times the odds of being recommended. Having a Wikipedia article mattered mainly because it went with being named in the evidence. Brands with an article were named in the cited pages 74.3% of the time, against 59.5% without. Once named there, they were recommended at about the same rate. So coverage works by getting you into the pages AI engines read before they choose. The live guides on [small brands](https://underneath.agency/resources/how-small-brands-get-recommended-by-ai) and [ad spend](https://underneath.agency/resources/does-ad-spend-help-ai-brand-recommendations) cover who benefits and what does not substitute for it. ## Is PR becoming part of search strategy in the age of AI search? Yes, because some AI assistants now search specifically for publications, rankings and awards before answering. In [our hidden searches study](https://underneath.agency/research/ai-hidden-searches-study), ChatGPT ran a search aimed at a named publication, ranking or award in 43.8% of its answers. Gemini did so in 26.2% and Claude in 1.2%. Coverage in those outlets is what such searches find. PR coverage gets you selected, but other pages may shape the words. In the public-dataset analysis, encyclopedia pages scored 0.2144 on the authors’ 0-to-1 measure of influence on the answer text, against 0.0726 for news pages. News gets you into the pool; clear explanatory pages supply more of what is said. In one vendor’s data, ranked “best of” lists were the most cited format, at about 21% of all citations ([Kumar](https://arxiv.org/abs/2606.20065)). ## Does AI search change the value of third-party reviews and publications? It raises their value, and concentrates it in a small number of outlets per engine. In [our reputation study](https://underneath.agency/research/is-it-legit-ai-reputation-study), answers about whether brands were legitimate cited 484 domains. The evidence was concentrated enough to behave like 30.5 equally used sources. A few review sites and publications carry most of the weight, so knowing which ones your category’s answers cite matters more than total coverage volume. Engines also read different outlets. In [our brand agreement study](https://underneath.agency/research/ai-assistants-brand-agreement-study), 66.3% of the options recommended for a question came from one assistant only. Coverage that one engine reads may be invisible to another. The guide [do AI engines cite your own website or third-party reviews?](https://underneath.agency/resources/do-ai-engines-cite-your-own-website) covers the split by engine. ## Can a brand manufacture its own authority? Not reliably, and the attempts that work in tests carry real risks. Self-ranked lists are common but small. In [our best-lists study](https://underneath.agency/research/self-promoting-best-lists-study), 24.2% of cited numbered “best X” lists ranked their own publisher first. Such lists were 1.1% of all citations, and we found no detectable difference in how often their top pick was named. Invented authority is another matter. In lab tests by [Chu and Hou](https://arxiv.org/abs/2606.17443), claims that mimic evidence, including made-up clinical trials, beat a famous brand 50–73% of the time. But when all nine rival brands used the same language, the famous brand’s recommendation rate recovered to 93.8%. The authors call fabricated claims potential false advertising. The advantage disappears once everyone copies it. ## Should you optimize your entire digital presence, not just your website? Yes; AI answers draw on peers, reviews and publications, and your site must still be readable when agents check it. In one vendor’s data, own and third-party company pages together made up about 78% of 149,912 citations, often competitors’ pages in “alternatives” answers ([Kumar](https://arxiv.org/abs/2606.20065)). Your presence on comparison, review and partner pages is part of your authority. Your own site still decides the final check. In a vendor study that held third-party mentions equal, AI agents clearly recommended businesses with agent-readable sites 20% of the time, against 11% for sites agents struggled to read ([Finder and colleagues](https://arxiv.org/abs/2609.34951)). See [where AI agents get their answer when they can’t read your site](https://underneath.agency/resources/when-ai-agents-cant-read-your-site). ## What should you do about it? Treat authority as a coverage program aimed at the specific sources AI engines read, and measure it there. 1. **Map the cited sources in your category.** Ask each engine your buyers’ questions and list the outlets, reviews and lists they cite. 2. **Pitch those outlets first.** Prioritize publications, rankings and awards that the assistants search for, over volume of placements. 3. **Earn real, checkable proof.** Use genuine certifications, studies and expert reviews; avoid invented claims. 4. **Keep review profiles healthy.** Review platforms are a major source for reputation answers. 5. **Keep one clear reference page about the company.** Explanatory, factual pages shape what answers say. 6. **Measure authority in AI answers.** Track, per engine, how often you are named, which sources name you, and how you are described. Ownership across SEO, PR and content is covered in [who should own AI search visibility](https://underneath.agency/resources/who-should-own-ai-search-visibility). For outside help, see our [generative engine optimization service](https://underneath.agency/services/generative-engine-optimization). ## What does the research not tell us yet? The research links coverage to recommendations, but no study has yet proved that new coverage causes them. - The coverage findings, ours included, compare brands at one moment; none tracked a brand before and after a PR push. - Several key figures come from vendors who sell AI visibility tools or readiness scores. - The startup study is small and single-authored, and tested older assistant versions. - The fabricated-authority tests used invented products, mostly in skincare, in a lab setup. - How long earned coverage takes to show up in AI answers is unmeasured. - Results vary by engine, language and industry, so one category’s cited outlets may not transfer. ## Frequently asked questions ### Is PR becoming part of SEO in the AI search era? In practice, yes. ChatGPT ran a search aimed at a named publication, ranking or award in 43.8% of its answers in our study, so press coverage is now direct input to AI answers. ### Should CMOs measure earned-media authority differently for AI search? Yes. Count how often each engine names you and which outlets it cites, because only 8.3% of the pages ChatGPT cited ranked in Google’s top 10 for the question. ### Do third-party reviews matter more for AI search than for Google? They appear to. For niche brands, ChatGPT drew 95.1% of its sources from earned media such as reviews, while Google showed far more brand-owned pages. ### Does a Wikipedia page give my brand authority in AI search? Mostly as a sign of wider coverage. In our study, the Wikipedia advantage largely disappeared once brand prominence and independent coverage were taken into account. ## Sources - Chen and colleagues (2025), [Generative Engine Optimization: How to Dominate AI Search](https://arxiv.org/abs/2509.08919), arXiv:2509.08919. - Jacques and colleagues (2026), [Authority Signals in AI Cited Health Sources: A Framework for Evaluating Source Credibility in ChatGPT Responses](https://arxiv.org/abs/2601.17109), arXiv:2601.17109. - Zhang, He and Yao (2026), [From Citation Selection to Citation Absorption: A Measurement Framework for Generative Engine Optimization Across AI Search Platforms](https://arxiv.org/abs/2604.25707), arXiv:2604.25707. - Zhang and colleagues (2025), [Source Coverage and Citation Bias in LLM-based vs. Traditional Search Engines](https://arxiv.org/abs/2512.09483), arXiv:2512.09483. - Sharma (2025), [The Discovery Gap: How Product Hunt Startups Vanish in LLM Organic Discovery Queries](https://arxiv.org/abs/2601.00912), arXiv:2601.00912. - Kumar (2026), [Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines](https://arxiv.org/abs/2606.20065), arXiv:2606.20065. - Chu and Hou (2026), [Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems](https://arxiv.org/abs/2606.17443), arXiv:2606.17443. - Finder, Elovic, Shalev and Yosef (2026), [AX is the New AEO](https://arxiv.org/abs/2609.34951), arXiv:2609.34951. - Underneath (2026), [Do Wikipedia and schema make AI assistants recommend a brand?](https://underneath.agency/research/brand-entity-ai-recommendations-study) - Underneath (2026), [Do ChatGPT, Gemini, Perplexity and Claude cite pages that rank?](https://underneath.agency/research/ai-citations-google-rankings-study) - Underneath (2026), [The hidden searches AI assistants run before they answer](https://underneath.agency/research/ai-hidden-searches-study) - Underneath (2026), [“Is this brand legit?” How AI assistants build a reputation](https://underneath.agency/research/is-it-legit-ai-reputation-study) - Underneath (2026), [Do ChatGPT, Gemini, Perplexity and Claude agree on brands?](https://underneath.agency/research/ai-assistants-brand-agreement-study) - Underneath (2026), [How many “best of” lists cited by AI rank their own brand first?](https://underneath.agency/research/self-promoting-best-lists-study) --- This is the Markdown twin of https://underneath.agency/resources/how-brands-build-authority-for-ai-search. The HTML page is canonical. Publisher: Underneath, https://underneath.agency/agent. Site index: https://underneath.agency/llms.txt.