Quick answer: The likeliest reason is that the third-party content AI reads when it answers, meaning comparisons, roundups and reviews, does not mention you, or mentions you too rarely. In one study, at least 92.7% of the brands ChatGPT recommended appeared in the content it cited, and brands in six or more cited sources were six times more likely to be recommended than brands in one. The fix is to appear in as many credible sources as you can.
Last updated: October 2026
AI recommendations are the new word of mouth. As we explored in our comparison of Google and ChatGPT, a search engine hands you a list of links, while an AI assistant tells you whom to call. Ask one to name the best tools in your category and it will answer in seconds, with a confident shortlist. The uncomfortable question for every founder is not whether that list exists. It is why your name is missing from it.
The stakes are no longer hypothetical. In a March 2026 survey of 1,076 B2B software buyers, G2 found that 51% now begin their research in an AI chatbot more often than in Google, up from 29% the year before. Roughly a third of respondents bought from a vendor they had not known before. The shortlist is being written before your sales team ever hears the buyer’s name.
The shift is not confined to software. Among consumers, the share who used an AI tool to find a local business in the previous 30 days rose from 9% in 2025 to 52% in 2026, according to SOCi data summarized by Metavert. Whether you sell enterprise software or book appointments, the first recommendation increasingly comes from a machine.
AI is where the search now starts
B2B software buyers who start research in an AI chatbot more often than in Google
Consumers who used an AI tool to find a local business in the previous 30 days
B2B: G2 survey of 1,076 software buyers, March 2026. Consumers: SOCi data, as summarized by Metavert.
Explanations for why a business is left off that shortlist are plentiful and mostly unexamined. So this guide does what our series on AI search engines has always done: it separates what the evidence shows from what is merely assumed and traces each absence to a pattern you can actually address. One caveat governs everything below. “Proven” here means consistently observed in large studies, not causally established. A digest of 2026 research notes that no reviewed optimization technique has yet shown a stable, cross-platform causal effect. Anyone promising you a guaranteed lever is selling certainty the science does not yet possess.
Mentioned, cited, recommended: Three different prizes
Before strategy comes vocabulary, because most confusion about AI visibility is a confusion of terms. They describe three distinct outcomes:
Cited: the AI links to your page as a source for its answer.
Mentioned: the AI names your business in its answer, whether or not it links to you.
Recommended: the AI endorses you for the specific need the user described.
You can be cited without being named, and named without being recommended. AuthorityTech’s testing guide treats these as separate events to record, and so should you. If a chatbot is not recommending you, start by asking which of the first two prizes you are missing, because the path to the third runs through them.
Proven: The sources AI reads do not mention you
The most rigorous evidence on recommendations comes from Contender’s analysis of ChatGPT’s answers to 46 B2B software prompts between October 2025 and January 2026, covering 2,654 cited URLs. Three findings matter most. They come from software recommendations, the most closely studied category, but the reviews and mentions research below spans industries from home services to travel.
ChatGPT now searches before it recommends: across more than 86,000 software-recommendation questions in Contender’s database, ChatGPT relied on web search 95.5% of the time, up from 29% in October 2025.
The recommendation follows the sources: at least 92.7% of the brands ChatGPT recommended appeared in the content it cited.
Staying cited keeps you recommended: brands already recommended were six times more likely to keep that status if they stayed in cited content (75% against 13%).
ChatGPT: brands that stayed cited mostly stayed recommended
Share of already-recommended brands that kept ChatGPT's recommendation, depending on whether they stayed in the content it cited. ChatGPT answers to 46 B2B software prompts, October 2025 to January 2026, covering 2,654 cited URLs. Source: Contender / Visible
The implication is quietly radical. When a buyer asks ChatGPT for a recommendation, your own website is only one voice in the room. The decisive testimony comes from the comparison articles, listicles and reviews the AI reads in that moment. If those sources leave you out, the answer usually will too. Most citations, according to research summarized by Metavert, point to sites the brand does not own: 85.7% in one study of 128 brands across 12 markets.
Four engines, four reading lists, one shortlist
No comparable study yet traces Claude’s, Gemini’s or Perplexity’s recommendations back to their sources in this way. The closest thing to a four-engine view comes from Slate, which put the same 1,000 B2B SaaS questions to ChatGPT, Perplexity, Claude and Gemini, each with web search switched on, in a single capture on 3 August 2026. The first finding is that each engine reads a strikingly different amount of the web. Perplexity showed an average of 19.08 source links per answer and Claude 12.49. ChatGPT showed 3.10, and Gemini 0.07, with a link in only 16 of 1,000 answers.
Four engines, four very different reading lists
Average number of source links shown per answer, across the same 1,000 B2B SaaS questions put to each engine with web search on, captured 3 August 2026. Source: Slate
The second finding is the more important one. The engines almost never cite the same pages: on a typical question, ChatGPT and Claude overlapped on just 0.5% of their sources, and the closest pair, Perplexity and Claude, on 13.6%. Yet any two engines agreed on 35.2% to 42.0% of the vendors they named, Gemini included, even though it cites almost nothing. Slate’s reading is that the shortlist reflects what the web as a whole says about a category, absorbed in training and topped up by each day’s retrieval, rather than any single page.
Different sources, similar shortlists
ChatGPT vs Perplexity
Perplexity vs Claude
ChatGPT vs Claude
How often two engines agreed on the vendors they named, compared with the pages they cited, across the same 1,000 B2B SaaS questions, captured 3 August 2026. Gemini is not shown because it displayed links in only 16 answers, so its page agreement could not be measured; its vendor agreement with the other engines ranged from 35.8% to 38.6%. Source: Slate
For a business that is not being recommended, the lesson cuts both ways. There is no single page whose citation unlocks every engine, because each one reads its own library. But the shortlists converge, so the remedy is the same everywhere: be present, and present well, across enough of the web that every engine, whatever it happens to read, keeps encountering your name.
Proven: You appear in too few sources, or too far down
A single mention opens the door but rarely carries you through it. In the same Contender study, 96% of brands that newly appeared in a cited URL still did not get recommended. Two factors separated the winners:
Quantity: brands appearing in six or more cited URLs were six times more likely to be recommended than brands appearing in one.
Position: brands listed nearer the top of cited articles were recommended more often, and a brand’s single best placement predicted recommendation better than its average one.
Treat “six” as a comparison, not a magic threshold. AuthorityTech is explicit that no qualifying number of articles has been established. The lesson is directional: appear in many credible sources, and lead in at least one.
Proven, by correlation: Without reviews, AI has reason to doubt you
In a study carried out on behalf of Trustpilot, Seer Interactive analyzed 804,491 AI responses across 1,926 brands on ChatGPT, Google AI Mode, Gemini and Perplexity. Brands without a Trustpilot profile appeared alongside a Trustpilot citation in a median of just 1% of responses. Brands with a minimal profile of as few as one to thirteen reviews rose to 53.5%. Because Trustpilot commissioned the research, its findings about Trustpilot itself deserve scrutiny. The broader pattern, that some review presence beats none, is the more transferable lesson.
Two details make this especially relevant to smaller companies. First, Seer observed ChatGPT explicitly flagging the absence of reviews as a reason for doubt, so silence is not neutral. Second, brands with low domain authority and a minimal profile were cited at nearly the same rate as mid-authority brands. Where Trustpilot was absent, the AI sought validation from other review platforms such as G2 and Capterra. Seer acknowledges that factors such as marketing spend may also play a part.
AI responses citing a brand alongside Trustpilot, by review profile
Median share of AI responses in which a brand appeared alongside a Trustpilot citation, across 804,491 responses on ChatGPT, Google AI Mode, Gemini and Perplexity. Study carried out by Seer Interactive on behalf of Trustpilot. Source: Seer Interactive
Proven, by correlation: Not enough of the web talks about you
In Ahrefs’ study of 75,000 brands, the signals most closely associated with appearing in ChatGPT, AI Mode and AI Overviews were not backlinks or domain authority. They were mentions. YouTube mentions correlated most strongly, at about 0.737, followed by branded web mentions at 0.66 to 0.71.
Read this with care. As one analysis of the study points out, large brands have more of everything, from mentions to backlinks to search volume, so correlation cannot isolate cause. Still, the direction is consistent with the Contender findings: AI systems appear to learn who matters from how widely the world talks about them.
Necessary, not sufficient: Schema markup
Structured data is not a citation switch, and it is not obsolete. Both claims circulate, and the evidence supports neither. The defensible position sits between them. Schema markup is how you tell machines, in their own language, who you are and what a page contains.
Microsoft has been explicit. Fabrice Canel of Bing confirmed on stage at SMX Munich in March 2025 that schema markup helps Microsoft’s LLMs understand content, and those systems power Copilot. Google takes a more measured line. Its documentation says there is no special schema needed to appear in AI Overviews or AI Mode, while its Article guidance notes that markup lets you tell Google more explicitly what a page is, who wrote it and what it is titled.
What schema does not do is guarantee citations. Ahrefs tracked 1,885 already well-cited pages that added JSON-LD, compared them with 4,000 matched pages, and found no major uplift on any platform. The study did not test pages that AI has never surfaced. Read together, the evidence supports a clear role for schema: a foundation of clarity, not a lever for citations.
The minimal starting point we recommend:
Organization, on your homepage: your business name, website, logo and official profiles. Google says this markup can help it understand your organization’s details and tell it apart from others in search results.
Article or BlogPosting, on every post: the headline, author, publication date and image, so each page states plainly what it is and who stands behind it.
One rule governs both. Google requires that structured data match what visitors can actually see on the page. Markup that claims more than the page shows does not help, and it can work against you.
Assumed: You must rank in Google’s top ten
The old logic said AI would only cite the winners of traditional search. The evidence no longer supports it. According to Ahrefs data summarized by Metavert, only 37.9% of URLs cited in Google’s AI Overviews rank in the organic top ten. Nearly a third rank beyond position 100.
Where Google's AI Overview citations rank in organic search
Source: Ahrefs (March 2026), as summarized in Metavert, State of AI Search Citations: Q4 2026
That figure has moved fast, and the studies measure overlapping but not identical things, using different datasets and methods. Ahrefs measured about 76% in July 2025, while BrightEdge puts the overlap nearer 17%. Ahrefs attributes the decline to query fan-out: Google splits a question into related sub-queries and cites the pages that appear most often across them. Organic visibility still matters. It simply counts across a wider set of searches than the one the user typed.
We have seen this on our own site. Between 27 September and 26 October 2026, our article “What Search Engine Does ChatGPT Use?” averaged position 14.7 in Google, on the second page, where few people click. Yet in that same period, 16 of its 66 Google impressions, or 24%, came from appearances inside AI Overviews and AI Mode, according to Google Search Console’s Generative AI report. The algorithm that ranks pages and the systems that cite them are reading the same web through different eyes.
Ranking still matters, but perhaps not in the way most people assume. Early results from our ongoing study of ChatGPT’s citations suggest that ranking in Google’s top ten makes AI more likely to cite your domain, but not necessarily the page that ranks. A larger dataset points the same way. In Ahrefs’ study of roughly 3,000 short-tail queries, only 10% of the URLs ChatGPT cited appeared in Google’s top ten, while 31.8% of the domains it cited did. Search rankings, in other words, appear to vouch for the site more than for any single page on it.
Our own evidence, and its limits
Our own figures are illustrative, not evidence on the scale of the studies above.
Over the same period, Bing Webmaster Tools recorded 19 citations of our content in Microsoft Copilot and Bing’s AI summaries. Seven went to our ChatGPT article, seven to our article on what search engine Claude uses, and five to our guide on checking whether AI can access your site. Most of our other posts received none.
The cited posts share a pattern. Each is titled as a specific question people actually ask, answers it in its opening lines, and separates confirmed facts from assumptions with sources for every claim. We offer this as an observation, not a proof. Three posts cannot establish causation. They are, however, consistent with the broader research: specific, evidence-led answers get cited, while the broad and generic does not.
One observation surprised us. Our Claude article earned as many Copilot citations as our ChatGPT article, yet its human readers left after an average of fourteen seconds. AI systems and human readers can judge the same page very differently, which is why measuring only traffic will mislead you.
If you serve local customers, the rules shift
Local search is the exception to much of the above. Yext found that for local queries, businesses’ own websites supply 44% of AI citations and listings a further 42%, according to Metavert’s summary. Your site and your business listings carry more weight here than third-party roundups do.
Visibility is far from guaranteed, even for strong performers. Zarla tested 1,851 businesses already ranking in Google’s top 20 for their trade and city, and fewer than half were confirmed named by ChatGPT or Google’s AI Mode. One caution in reading that figure: the businesses were chosen for their Google rankings, and the study did not examine whether any of them had prepared their sites for AI. What it shows is that ranking alone does not carry a business into the answer. It did find one telling difference. The businesses that were named had a median of 444 Google reviews against 192 for those that were not, while their ratings were nearly identical, at 4.9 and 4.8 stars. For a local business, accurate listings, a complete website and a steady stream of reviews are the foundation. For a step-by-step local checklist, including your Google Business Profile, see our guide How Do I Get Gemini to Recommend My Business?
How to measure whether AI recommends you
Two free tools now show part of the picture directly:
Google Search Console, Generative AI report: shows how often your pages appear in AI Overviews and AI Mode. Google launched it in June 2026. It reports impressions only, not clicks or queries.
Bing Webmaster Tools, AI Performance: counts how often Copilot and Bing’s AI summaries cite your pages and shows the queries behind those citations. It does not cover ChatGPT.
Neither tool tells you whether ChatGPT recommends you. For that, you have to ask it, repeatedly. Research cited by Metavert found that roughly 65% of the sources an AI engine cites for a query change from one day to the next. A single check is a snapshot, not a verdict.
The Danu Agency AI Recommendation Check
Ask the buyer's question: type the exact question your customer would ask into ChatGPT, Claude, Gemini and Perplexity. Note who is recommended and which pages are cited. Expect Gemini to show few or no links; there, being named is the only result you can check.
Find the missing prize: for each engine, note whether you were cited, mentioned or recommended. If you were none of these, the problem lies in what the web says about you. If you were cited but not recommended, it lies in how you are presented.
Read the cited sources: open the articles the AI relied on. Are you listed? Where?
Check your reviews: search for your business on the review sites your customers use, such as Trustpilot, G2, Capterra or Google. If you are absent, that is the lowest-effort gap to close.
Open your Generative AI report: in Google Search Console, go to Performance, then Generative AI.
Open AI Performance: in Bing Webmaster Tools, note your citations and the grounding queries behind them.
What this means for your company
Assembled from the evidence above, the remedy for being left off the shortlist is less mysterious than the industry suggests, and more demanding:
Make sure AI can read your site: check that your robots.txt does not block the AI search crawlers. OpenAI, for example, states that sites blocking its OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. Our guide The Bouncer at the Door shows how to check.
Earn your place in the sources AI reads: the comparisons and roundups in your category shape most recommendations. Aim to appear in many, and to lead in at least one.
Don’t optimize for a single engine: ChatGPT, Claude, Gemini and Perplexity read almost entirely different sources yet arrive at overlapping shortlists. Broad presence across the web serves all four; chasing one engine’s favorite pages serves only that one.
Establish a review presence: even a minimal review profile is associated with far more AI citations, and AI has been observed flagging missing reviews as a reason for doubt. For local businesses, depth matters too: in Zarla’s study, the businesses AI named had more than twice as many Google reviews as those it passed over.
Answer specific questions on your own site: title each page as the question buyers ask, answer it first, and support every claim.
Be discussed beyond your website: mentions across the web and on YouTube are the signals most closely associated with AI visibility at this time.
Use YouTube deliberately: according to Otterly data summarized by Metavert, long-form videos account for 94% of YouTube citations, and views and subscriber counts showed no relationship with being cited. Gumshoe found that how closely a video’s title and description match the question mattered more than channel metadata. Title videos as the exact question your buyers ask, and you don’t need a large audience to start.
Keep it current: Seer Interactive found that 75% of cited pages had been updated within the previous year, according to Metavert’s summary. Refresh key pages, date your updates, and keep your facts consistent across your site and listings. Our guide The Free Five covers the first steps, none of which cost anything.
Measure monthly, not once: citations shift constantly, so track the trend rather than any single result.
None of this is a guarantee. All of it is grounded in evidence rather than folklore, and that distinction is the whole of the advantage.
Does AI know your business? Our free AI Awareness Check asks ChatGPT, Gemini and Perplexity what your business does and who it serves and shows you each answer and the sources behind it. Or explore our AEO services.
Frequently asked questions
Why is my business not being recommended by chatbots?
Most likely because the pages chatbots read when they answer do not mention you. When ChatGPT recommends software, it overwhelmingly names brands that appear in the pages it cites: at least 92.7% in Contender's study of B2B software prompts. The practical goal is to be listed, and listed early, in the comparisons, roundups and review sites your buyers' questions bring up.
Do ChatGPT, Claude, Gemini and Perplexity recommend the same businesses?
Partly. In Slate's study of 1,000 B2B SaaS questions, any two of the four engines agreed on 35.2% to 42.0% of the vendors they named, even though they almost never cited the same pages. Each engine reads its own sources, but their shortlists overlap.
Does schema markup help AI recommend my business?
Schema helps AI systems understand who you are and what a page contains. Microsoft has confirmed it helps its LLMs understand content. It is not a citation switch, though: Ahrefs found no major citation uplift when already well-cited pages added it, and Google says no special schema is needed for its AI features. Start with Organization markup on your homepage and Article markup on every post.
Do I need to rank first on Google to be cited by AI?
No. Ahrefs data shows only 37.9% of URLs cited in Google's AI Overviews rank in the organic top ten. Our own article on ChatGPT's search engine ranks on Google's second page, yet appears in AI Overviews and AI Mode.
Do reviews affect whether AI recommends a business?
They appear to. In a study commissioned by Trustpilot, Seer Interactive analyzed 804,491 AI responses and found brands with even a minimal Trustpilot profile were cited alongside it far more often than brands with none, and ChatGPT sometimes flagged missing reviews as a reason for doubt.
How can I tell if ChatGPT is recommending my business?
Ask it the questions your buyers ask, and repeat the check over time, because cited sources change frequently. Google Search Console and Bing Webmaster Tools now report AI visibility for Google's and Microsoft's AI features, but neither covers ChatGPT.
Is being mentioned by AI the same as being recommended?
No. A mention names your business, a citation links to your page as a source, and a recommendation endorses you for the user's need. Each is a separate outcome, and each should be measured separately.
Sources
G2, The Answer Economy (2026)
https://company.g2.com/news/g2-research-the-answer-economyTechstrong.ai, coverage of G2’s buyer survey (2026)
https://techstrong.ai/?p=47678Contender / Visible, how much can you influence which brands ChatGPT recommends? (February 2026)
https://visible.beehiiv.com/p/how-much-can-you-influence-which-brands-chatgpt-recommendsSlate, How to Get Cited by ChatGPT, Perplexity, Claude and Gemini: A Detailed Study (August 2026)
https://slatehq.com/data-studies/how-to-get-cited-by-chatgpt-perplexity-claude-and-geminiSeer Interactive, Study of 800K AI Responses (May 2026)
https://www.seerinteractive.com/insights/study-of-800k-ai-responses-how-reviews-shape-brand-presence-in-ai-searchSeer Interactive, Trustpilot partners with Seer Interactive
https://www.seerinteractive.com/news/trustpilot-seer-study-how-review-profiles-impact-brand-ai-visibilityAhrefs, Top Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews
https://ahrefs.com/blog/ai-brand-visibility-correlationsHartzer, Brand mentions in AI search
https://hartzer.it.com/brand-mentions-in-ai-search/Search Engine Land, Microsoft Bing/Copilot use schema for its LLMs (2025)
https://searchengineland.com/microsoft-bing-copilot-use-schema-for-its-llms-453455Google Search Central, AI features and your website
https://developers.google.com/search/docs/appearance/ai-featuresGoogle Search Central, Article structured data
https://developers.google.com/search/docs/data-types/articleGoogle Search Central, Organization structured data
https://developers.google.com/search/docs/data-types/logoAhrefs, We Tracked 1,885 Pages Adding Schema (May 2026)
https://ahrefs.com/blog/schema-ai-citations/Ahrefs, How ChatGPT citations compare with Google results
https://ahrefs.com/blog/chatgpt-google-citationsMetavert, State of AI Search Citations: Q4 2026
https://metavert.io/state-of-ai-search-citationsAuthorityTech, B2B Data Analytics Citations in ChatGPT and Perplexity (2026)
https://authoritytech.io/blog/b2b-data-analytics-chatgpt-perplexity-citations-2026.mdOpenAI, Overview of OpenAI crawlers
https://developers.openai.com/docs/gptbotGoogle Search Central, Search Generative AI performance reports (June 2026)
https://developers.google.com/search/blog/2026/06/gen-ai-performance-reportsAdobe Experience League, Bing Webmaster Tools AI Performance Report
https://experienceleague.adobe.com/en/docs/brand-visibility-learn/tutorials/measurement/bing-webmaster-tools-ai-performance-reportZarla, AI Visibility Index (August 2026)
https://www.zarla.com/guides/ai-visibility-indexDanu Agency, Google Search Console and Bing Webmaster Tools data (2026)
