Last quarter, a luxury hospitality group asked us to run a standard AI visibility audit. They expected to show up well. Strong Google rankings, a polished website, consistent press coverage for two decades. When we tested 50 prompts across ChatGPT, Perplexity, Claude, and Gemini, their brand appeared in seven. A boutique competitor with half their properties appeared in nineteen.

The disconnect wasn't a fluke. It revealed something most marketing teams haven't fully internalized yet: AI platforms have already formed an opinion about your brand, and that opinion was built from signals your team probably isn't tracking.

The signals aren't what you think

Traditional brand management assumes you control the narrative through your website, your advertising, your PR. AI platforms don't work that way. They build brand understanding from a mix of training data, real-time web retrieval, and third-party content you may have never seen.

Onely's research into ChatGPT's recommendation engine found that authoritative list mentions account for 41% of what drives commercial brand recommendations. Awards and accreditations contribute 18%. Online reviews add 16%. Your own website content? It barely registers as a direct signal. The hierarchy inverts everything traditional marketing teams have prioritized. The industry ranking you dismissed as a vanity metric carries more weight in AI answers than the landing page you spent six months perfecting.

Ahrefs' study of 75,000 brands found that YouTube mentions showed the strongest correlation with AI visibility at roughly 0.737, outperforming every other factor. Branded web mentions followed closely. Meanwhile, content volume (number of site pages) showed almost no relationship at roughly 0.194. Publishing more doesn't help. Being mentioned more does.

Different platforms, different perceptions

Here's what makes AI brand perception particularly volatile: each platform can see your brand differently. ChatGPT might position you favorably based on strong Wikipedia presence and press coverage. Claude might reflect skepticism drawn from Reddit threads in its training data. Perplexity, with its real-time retrieval, might surface a recent negative review that other models miss entirely.

BrightEdge's research across thousands of prompts in four industries found that ChatGPT and Google AI Mode disagree on brand recommendations primarily when queries are action-driven. ChatGPT acts as what they describe as a "trusted coach," directly suggesting tools and products. Google AI Mode functions more as a research assistant, linking to informational content. In healthcare, the divergence hit 62%.

This means your brand doesn't have one AI perception. It has several, and they may contradict each other. A financial services firm we audited appeared as a top recommendation in Perplexity for retirement planning, showed up with a lukewarm qualifier ("suitable for some investors") in ChatGPT, and didn't appear at all in Gemini for the same query.

The training data problem

AI models carry forward perceptions from their training data, which means outdated characterizations can persist indefinitely. A product issue from 2023 that was resolved within weeks might still influence how ChatGPT frames your brand in 2026. A competitor's aggressive content campaign from two years ago might have permanently shifted the association patterns models use to decide who gets recommended.

Semrush's analysis of one million non-branded queries found that AI models include brand mentions in 26% to 39% of responses. That means for roughly a third of relevant conversations, AI is actively naming brands. The question is whether yours is among them, and if so, how it's being framed. A mention with the qualifier "affordable but limited" creates a very different buyer perception than "industry-leading" or "trusted by enterprises."

BrightEdge found that ChatGPT mentions brands 3.2 times more often than it provides clickable citations. The brand name itself becomes the primary visibility mechanism, not a link to your website. That makes the framing language around your mention far more consequential than it would be in a traditional search result, where the user clicks through and forms their own impression.

What you can do about it

The first step is measurement. Run 30 to 50 prompts that match how your buyers research your category. Test across ChatGPT, Claude, Perplexity, and Gemini. For each response, document whether your brand appears, what language surrounds the mention, and which competitors show up instead.

Then look at the gap between how AI frames your brand and how you want to be positioned. If ChatGPT calls you "a good option for small teams" but you serve enterprise clients, that's a perception misalignment rooted in the signals AI has absorbed. The fix isn't tweaking your website copy. It's shifting the external signals: earning mentions in enterprise-focused publications, generating reviews from enterprise customers on G2 or Gartner Peer Insights, and ensuring your Wikipedia presence reflects your current positioning.

The brands treating this as an ongoing discipline rather than a one-time audit are the ones pulling ahead. AI perception isn't static. Models update, training data refreshes, real-time retrieval shifts with every new piece of content published about your brand. The companies that monitor and shape these signals will own the narrative. Everyone else will discover what AI thinks of them only when a prospect tells them.