What a 36.9% Appearance Rate Taught Me About the Future of Brand Trust
I have spent more than twenty years telling clients that visibility and reputation are not the same thing — a distinction that connects directly to how I think about named-principal reputation generally. Usually it is a distinction I have to argue for, patiently, against the instinct that any mention is a good mention. This month, for the first time, I have a number precise enough to prove it outright.
My firm just completed a study measuring how AI engines — ChatGPT, Claude, Gemini, and Perplexity — talk about wine brands. We put two dozen consumer-style questions to all four systems, collected 268 usable answers, and tracked two entirely separate things for every brand named: how often it came up, and, independently, how often the AI actually pointed back to that brand's own website as the source of what it had just told someone.
The results separated those two things more cleanly than anything I have encountered in twenty-some years of this business. Domaine de la Romanée-Conti was named in 36.9% of all answers returned — more than any other wine brand, by a wide and statistically unambiguous margin. And in every single one of those answers, the AI's supporting evidence originated from somewhere other than DRC's own site. Not "usually." Not "mostly." Zero, across all 99 qualifying instances. Read the full Wine AI Visibility Index 2026 for the complete data behind that number.
I want to sit with that number for a moment longer than is comfortable, because I believe it is the clearest single data point I have yet seen for something I have suspected was coming for a while now: brands are quietly losing the ability to serve as their own source.
Why has the link between telling your story and being believed broken?
For the entire history of modern marketing, a brand that told its own story well enough, consistently enough, and loudly enough could eventually become the trusted source of information about itself — this is the same territory I mapped out in Brand Positioning and Public Relations: The 2026 Strategic Read. That was true in print. It was true on the open web. It was, for the most part, true in search — if you ranked, you were simultaneously the answer and the source, because the search result and the underlying page were, by definition, one and the same object.
Generative AI is the first major discovery channel in my career where that link breaks apart entirely. A system can name a brand relentlessly and never once, not a single time, treat that brand as the authority on itself. The brand wins the mention and loses the microphone in the very same instant, and most executives I talk to have no idea it is happening, because nobody inside their organization is currently measuring the second half of that equation.
Who does get treated as the authority, in this particular case? Not a rival château. Not a competing producer trading on a similar prestige narrative. A pricing and comparison tool called Wine-Searcher, which posted the highest rate of being directly cited of anything we measured in the entire study — 19%, more than double the next closest name. Wine Enthusiast, Vinovest, VinePair, and Forbes rounded out the remainder of what I have started calling the evidence layer. None of those four publications is in the business of selling you a bottle of wine. All four are, structurally and measurably, more trusted by these systems than the producers whose reputations are actually at stake.
Why is this not, in the end, a story about wine?
I do not think this is fundamentally a wine story. I think wine is simply the category where the underlying pattern happened to be clean enough, and statistically distinct enough, to publish with confidence. We have been running this same kind of study across dozens of categories throughout the year, and the shape keeps repeating itself with almost unnerving consistency: appearance is cheap to win; citation is not. A brand can win the mention and lose the narrative in the very same breath, and the overwhelming majority of companies have no visibility into the fact that it is happening at all, because nobody has built the second metric yet — which is exactly the measurement gap I wrote about in PR ROI: How Public Relations Got Measurable.
There is an entirely independent number that corroborates this outside our own research, and I think it strengthens the case considerably. Muck Rack's Generative Pulse research — a separate organization, a separate methodology, no relationship to my firm whatsoever — put earned, third-party media at roughly 84% of all AI citations across the platforms it studied, with brand-owned content registering at a small fraction of a percent by comparison. Two entirely different research efforts, built on two entirely different methodologies, arrive at the identical structural conclusion: these systems are, by design and by training, built to trust someone other than you.
What does this mean for how we do the work?
If I am right that this is the direction brand communication is genuinely headed, then the job itself changes in a specific and identifiable way. For years, the pitch to a prospective client was some version of "let's control the story." I think the pitch now has to become something closer to "let's make certain someone credible is telling the story you would actually want told, because the AI is not going to quote you telling it yourself, no matter how well you tell it." That is a genuinely different discipline than maintaining a press release calendar or a content publishing schedule. It looks, in practice, much more like an evidence strategy — deciding, category by category and fact by fact, which independent voices need to be the ones saying the specific, true thing about your brand, because those are the voices the machine has actually learned to cite.
The wine industry did not ask for this shift, and frankly, neither did anyone else's. But the number is the number, and it does not move because an industry finds it inconvenient: 36.9% named, 0% cited. I would rather every client I work with understand that gap exists, and understand precisely how to close it, before a competitor figures out how to close it first.
— Ronn Torossian
