Where AI Learns About Your Brand — Your Website Isn't on the List

According to Boris Djingarov, CEO of ESBO ltd, when a shopper asks an AI assistant to recommend a brand like yours, the answer comes back in seconds — and your website has almost nothing to do with it. Models form their picture of a brand from what other people publish about you, so the real question is where they're "reading" that from. It's a completely different map from the one SEO trained us to think in, where every road led back to your own domain — this new map runs through territory you don't control.

The first and most important source is earned media. Muck Rack analyzed more than 25 million links cited by ChatGPT, Claude and Gemini and found that earned media accounts for 84% of all AI citations — with journalism alone making up 27% — while paid placements account for a mere 0.3%. The practical takeaway is simple: respond quickly to reporter queries and put out original, quotable data about your brand — the coverage you can't buy is exactly the coverage the machines trust.

The second source is video and podcast transcripts. In an Ahrefs study covering 75,000 brands, being mentioned on YouTube turned out to be a stronger citation signal than any traditional metric — because transcripts exist as text, and text is exactly what models read at scale. You don't need your own channel for this: get booked as a guest on the niche industry shows your customers already follow, and say your company's name out loud a few times.

The third source is review sites. According to SE Ranking research, the vast majority of review links cited in AI answers point back to a handful of leading platforms — typically G2 or Capterra for B2B software, and Trustpilot and similar sites for consumer brands. Claim your profile on these platforms, ask customers for reviews consistently, and respond to the ones that come in.

The fourth source is forums and communities, including Reddit and Quora, which now feed data directly to AI companies through licensing deals; when marketing copy reads as too polished, models lean more heavily on candid, authentic discussion instead. The fifth source is reference material — Wikipedia, industry wikis, directories and rating sites — which are often compiled from the same press coverage, looping the list back to the first item. The author closes with a suggestion: write down five questions a customer might ask an AI, test them through two different AI tools, and compare the sources cited against last quarter's marketing budget — at most companies, the two lists barely overlap.