--- title: "Which pages should we target to show up in AI recommendations?" description: "Ranked “best of” lists on other sites are the page type AI engines cite most when naming brands; your own site is a small share of citations." canonical: "https://underneath.agency/resources/best-of-lists-ai-recommendations" published: 2026-10-11 updated: 2026-10-11 publisher: "Underneath (https://underneath.agency/agent)" entity: "https://underneath.agency/.well-known/entity.json" --- Guide · AI search # Which pages should we target to show up in AI recommendations? Start with ranked “best of” lists on other people’s sites: they were the single most-cited content format in a study of 102 brands across five AI engines. A brand’s own website was a small share of citations, and AI search engines leaned heavily on third-party reviews and publishers. Being on the right list is not a guarantee of being named, but it puts you in the evidence the answer is built from. ## The short version 1. Ranked “best of” lists were 35.7% of the content pages AI engines cited across 102 tracked brands, more than any other format ([Kumar](https://arxiv.org/abs/2606.20065)). 2. Only 2.9% of citations in that study pointed at the brand’s own website; 75.2% pointed at other companies’ sites. 3. When a web-enabled GPT model was asked to rank cars in the US, 81.9% of the sources it used were independent publishers and review sites, not brand sites ([Chen and colleagues](https://arxiv.org/abs/2509.08919)). 4. In [our study of cited lists](https://underneath.agency/research/self-promoting-best-lists-study), answers that cited an independent list named its top pick 43.4% of the time; other answers to the same question named that pick 33.4% of the time. 5. In [our brand study](https://underneath.agency/research/brand-entity-ai-recommendations-study), coverage on independent sites was the strongest predictor of being recommended that we measured. ## Which kind of page do AI engines cite most when they recommend brands? Ranked “best of” lists, the roundups titled “best CRM tools” or “top running shoes”. The largest study we found tracked 102 brands across ChatGPT, Gemini, Perplexity, Claude and Grok from March to May 2026, collecting 149,912 source citations ([Kumar](https://arxiv.org/abs/2606.20065)). About 59% of cited pages were content, such as articles, guides and videos. The rest were homepages, product pages and other landing pages. Within content, the ranked list led. | Content format | Share of cited content | |---|---| | Ranked “best of” list | 35.7% | | General article | 31.0% | | How-to guide | 9.7% | | Comparison | 4.5% | Counted across all citations, lists were about 21% of everything cited. The author argues that one list including your brand can be reused across many different questions. That is his reading of the citation shares, not a tested effect. The author also co-founded the tracking platform the data comes from, so treat this as a vendor’s study. ## Why do other people’s pages matter more than your own site? Because AI engines mostly cite third parties when recommending. In the same study, only 2.9% of citations pointed at the tracked brand’s own website. Another 75.2% pointed at other companies in the same space, such as competitors, peers and vendors. Independent research shows the same tilt. University of Toronto researchers ran 1,000 ranking-style questions, such as “Rank the best smartphones from 1 to 10”, through Google and a web-enabled GPT model ([Chen and colleagues](https://arxiv.org/abs/2509.08919)). For US car questions, Google’s results were 45.1% independent publishers; the AI model’s sources were 81.9% independent publishers and review sites. Engines differ in how far they lean. For questions about well-known brands, 93.5% of ChatGPT’s sources were independent publishers, against 63.4% for Gemini. Independent publishers were the majority of sources for every engine in that study; [your own pages carry less weight](https://underneath.agency/resources/do-ai-engines-cite-your-own-website) in this kind of answer. ## Does being on a list get you named in the answer? It often does, but not always. In [our study of cited lists](https://underneath.agency/research/self-promoting-best-lists-study), we collected every page cited by six AI surfaces in the US on 26 September 2026 that was titled as a ranked list. There were 1,574 such pages. When an answer cited an independent list, it named that list’s top pick 43.4% of the time. Other answers to the same question that did not cite the list named the same pick 33.4% of the time. The gap fits lists shaping answers, but our study measured association, not cause. [Our brand study](https://underneath.agency/research/brand-entity-ai-recommendations-study) points the same way. Each tenfold increase in independent sites naming a brand in the pages an assistant cited went with 4.7 times the odds of that brand being recommended. It was the strongest predictor we measured, stronger than having a Wikipedia article. ## Should you publish your own “best of” list? The evidence does not show a payoff. In our study, 24.2% of cited numbered lists with an identifiable publisher ranked their own publisher first. So AI engines do cite such lists. But these lists were a small part of what answers cited: 1.1% of all citations. And answers named a self-ranking publisher at a rate we could not tell apart from an independent list’s top pick, 47.1% against 43.4%. Our study did not test whether readers trust such lists less. ## Which lists are worth pursuing? The ones AI assistants already go looking for. In [our study of the searches assistants run](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. When an assistant’s search named a source, such as NerdWallet or Avvo, the answer cited that source 44.0% of the time. In comparable answers that did not name it, the figure was 8.1%. That is an association, but it suggests the recognized rankings in your category are where to start. Size matters too. Kumar found that on their first tracking run, global household-name brands appeared in 73% of category answers that did not name them, against 11% for small brands. His advice for small brands is to build presence in mainstream press, Wikipedia and YouTube before fine-tuning for each engine. Whether paid media can stand in for that presence is covered in [whether ad spend helps AI recommendations](https://underneath.agency/resources/does-ad-spend-help-ai-brand-recommendations). ## What should you do about it? Treat inclusion in respected ranked lists as a core channel, alongside your own site. 1. Ask an AI assistant your category’s buying questions and note which lists it cites. Those lists, and the publishers behind them, are your target set. 2. Find the rankings and awards your category’s buyers trust. Assistants often search for these by name. 3. Give list editors what they need to judge you fairly: clear pricing, specifications, and access for independent testing. 4. Pursue several independent lists rather than one. Coverage across many independent sites was the strongest signal in our brand study. Once an AI agent reads about you directly, your own site matters more; see [off-site mentions versus a readable site](https://underneath.agency/resources/off-site-mentions-vs-site-readability-ai-agents). 5. Do not count on ranking yourself first on your own site. We found no evidence it helps. 6. Re-check every few weeks. Cited lists change between engines and over time. If you want help building that coverage and tracking it, see our [generative engine optimization service](https://underneath.agency/services/generative-engine-optimization). ## What does the research not tell us yet? No study has tested whether getting onto a list causes AI engines to recommend a brand more. - The main list figures come from one vendor’s tracking data on 102 brands, sampled for convenience and skewed toward software, fintech and consumer brands. - Our list study comes from one collection date in the US. Its repeated runs were minutes apart, not days. - Nobody has measured how long a list stays cited, or what happens when a list is updated and drops a brand. - The research does not show whether readers trust AI answers that cite vendor-written lists less. - Engines change their sources after updates, so today’s mix of cited page types may shift. ## Frequently asked questions ### What type of page do AI search engines cite most for brand recommendations? Ranked “best of” lists. In a study of 102 brands across five AI engines, they were 35.7% of cited content pages, ahead of general articles at 31.0%. ### Does my own website get cited when AI recommends my category? Rarely, in current data. Only 2.9% of citations pointed at the tracked brand’s own website in a study of 149,912 citations. ### Do AI engines cite brands’ own “best of” lists that rank themselves first? Yes, sometimes. In our study, 24.2% of cited numbered lists with an identifiable publisher ranked that publisher first, but they were only 1.1% of all citations. ### Is PR coverage worth it for AI search? The evidence leans that way, though it is observational. In our brand study, coverage on more independent sites went with 4.7 times the odds of being recommended for each tenfold increase. ## Sources - Kumar (2026), [Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines](https://arxiv.org/abs/2606.20065), arXiv:2606.20065. - Chen, Wang, Chen and Koudas (2025), [Generative Engine Optimization: How to Dominate AI Search](https://arxiv.org/abs/2509.08919), arXiv:2509.08919. - Underneath (2026), [How many “best of” lists cited by AI rank their own brand first?](https://underneath.agency/research/self-promoting-best-lists-study) - Underneath (2026), [Do Wikipedia and schema make AI assistants recommend a brand?](https://underneath.agency/research/brand-entity-ai-recommendations-study) - Underneath (2026), [The hidden searches AI assistants run before they answer](https://underneath.agency/research/ai-hidden-searches-study) --- This is the Markdown twin of https://underneath.agency/resources/best-of-lists-ai-recommendations. The HTML page is canonical. Publisher: Underneath, https://underneath.agency/agent. Site index: https://underneath.agency/llms.txt.