In our research at Nobori, we analyzed more than 200,000 AI prompt runs tied to commercial discovery and found that 85% of AI visibility does not come from the brand’s own website.
Nobori is an AI search optimization platform for B2B companies. We help teams track whether their brand appears in buyer-intent prompts across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews, see which pages and domains those systems cite, and turn that data into clear actions to improve visibility.
In our research at Nobori, we analyzed more than 200,000 AI prompt runs tied to commercial discovery and found something most companies still underestimate: 85% of AI visibility does not come from the brand’s own website. It comes from third-party pages, comparison articles, review platforms, listicles, directories, community discussions, and other external sources that AI systems rely on when they decide which brands to surface.
This is the part many B2B teams still miss. They treat AI visibility as a website problem. They assume that if they publish enough content, improve technical SEO, and clean up their service pages, AI systems will naturally start recommending them. That logic is incomplete.
Your website still matters. It matters a lot. But in AI search, especially for commercial prompts, your website is rarely the whole story. In many categories, it is not even the dominant one.
This is not a theoretical shift. Buyer behavior is already moving in this direction.
In Wynter’s 2026 B2B SaaS buyer journey report, 84% of CMOs said they now use AI assistants such as ChatGPT, Claude, and Perplexity for vendor discovery. In G2’s 2025 Buyer Behavior Report, generative AI chatbots became the number one source influencing vendor shortlists, ahead of vendor sites and software review sites. In 6sense’s 2025 B2B Buyer Experience Report, buyers filled 3.6 shortlist positions on day one and ended up choosing from that day-one shortlist 95% of the time.
Put those three findings together and the implication is obvious. If AI helps shape the shortlist, and the shortlist is largely formed before a buyer talks to sales, then AI visibility is not just a search metric. It is a pipeline metric. It affects whether your company is even considered.
The headline from our own dataset is simple: most AI visibility in commercial discovery does not come from your own domain.
That explains a lot of what B2B marketers are seeing right now. A company can have a decent website, publish content consistently, even rank reasonably well in traditional search, and still fail to appear when a buyer asks AI a commercial question like “best ERP for manufacturers,” “top nearshore development partner for DACH startups,” or “best B2B demand generation agency for SaaS.”
Why? Because AI systems do not only ask, “What does this company say about itself?” They also ask, implicitly, “What does the rest of the web say about this company, and is that story repeated often enough to trust?”
That is why off-site presence matters so much more than most teams expect.
Our finding is not isolated. It lines up closely with what other researchers are seeing.
In AirOps’ research on off-site signals in AI search, 85% of brand mentions in commercial discovery came from external domains rather than owned websites. The same study found that brands were 6.5 times more likely to be cited through third-party sources than through their own domains, and that nearly 90% of those third-party mentions came from listicles, comparison pages, and review-style content.
That matters because it tells us this is not random noise. There are recurring source patterns behind AI recommendations. Commercial AI search is not shaped by the whole web equally. It is shaped disproportionately by a relatively small set of formats that are easy for models to retrieve, compare, and summarize.
Another mistake many teams make is treating all visibility as equal.
It is not enough for AI to mention your brand somewhere. What matters is whether AI recommends you when the query has buying intent.
That distinction shows up clearly in Semrush’s AI visibility study. Semrush found that only 6% to 27% of the most-mentioned brands also became top cited sources, depending on industry and platform. In other words, being present in the conversation and being trusted enough to shape the answer are two different outcomes.
This is exactly why many “AI brand monitoring” dashboards feel unsatisfying. They tell you that your brand was mentioned. They do not tell you whether you were shortlisted. For B2B companies, that difference is massive.
A brand mention can make you feel visible. A shortlist inclusion can generate revenue.
None of this means your website is irrelevant.
Your site still provides the official version of who you are, what you do, who you serve, and why you are credible. It gives AI systems the facts, proof points, case studies, category language, and use-case specificity they need when the buyer moves from discovery into validation. AirOps’ off-site signals research specifically notes that brand-owned pages become more important when intent shifts from broad exploration to verification.
But that is exactly the point. Owned content becomes more influential later in the decision process. Earlier in the process, when the buyer is asking broad commercial questions and AI is assembling a shortlist, external validation often carries more weight.
So the strategic mistake is not underinvesting in your website. The mistake is assuming your website is the whole battlefield.
When we look at commercial AI search, the pattern is more structured than it first appears.
The sources that tend to shape AI recommendations are usually some combination of category listicles, “best X” roundups, comparison pages, review platforms, directories, community threads, YouTube explainers, and clear product or service pages that explicitly state category fit. That is also the pattern highlighted in AirOps’ off-site signals research, where listicles, comparisons, and reviews drove nearly 90% of third-party mentions.
This has two important consequences.
First, generic advice like “publish more content” is too vague to be useful. AI systems do not reward content volume on its own. They reward clear, repeated, corroborated positioning across the places they already use to form answers.
Second, AI visibility is won cluster by cluster, not keyword by keyword in the old SEO sense. A company does not “win AI search” in general. It wins specific commercial buying moments.
For one company, that cluster might be “best CRM for agencies.” For another, it might be “top cybersecurity consultant for fintech.” For another, it might be “nearshore software development company for German startups.” Each of those clusters has its own shortlist logic, its own cited sources, and its own competitive gaps.
The practical takeaway is not to abandon SEO. It is to widen the model.
You still need strong owned content. Your site should clearly state who you serve, what problem you solve, what proof you have, and why you are a credible option. Ambiguous copy loses in AI search because it gives the model nothing precise to quote or validate.
But you also need a deliberate off-site visibility strategy. That means identifying the external sources that already shape answers in your category, then systematically improving your presence there. In many B2B markets, that means review platforms, comparison pages, category listicles, industry articles, founder-led content, community discussions, and selected video content.
Just as importantly, you need to measure the right thing. Stop asking only, “Did AI mention us?” Start asking, “When someone asks AI who to consider in our category, are we actually in the answer?”
That is the more useful question. It is closer to revenue, closer to buyer behavior, and closer to how AI discovery actually works.
The old search model was simple. Rank your page, win the click, enter the funnel.
The new model is different. AI systems synthesize what the web says, compress early-stage research, and often shape the shortlist before the buyer ever visits a vendor website. That means your website is still part of the system, but not the center of it in every commercial query.
That is the real meaning of our finding.
When we say that 85% of AI visibility does not come from your website, we are not saying websites stopped mattering. We are saying that AI recommendation is increasingly a function of broader web evidence, not just owned content. And for B2B companies, that changes both the strategy and the measurement model.
The winners in AI search will not be the brands that simply publish more on their own domain. They will be the brands that build enough clear, repeated, trusted evidence across the wider web that AI systems can confidently include them when buyers ask who to consider.
