In AI search, hotel brand beats guest reviews
A study of 4,545 ChatGPT prompts found AI recommends big chains far beyond their room count — and a hotel's review score barely moves whether it appears
Driving the news. Lighthouse presented a study at its Luminate conference of 4,545 ChatGPT prompts across nine destinations and five traveler personas, watching the model name hotels 49,707 times (vendor research, ChatGPT only). The finding that should reach any property competing below brand scale: AI hotel discovery concentrates on the names the model already knows, and the guest-review score independents fight hardest for barely registers.
The concentration. A small set of chains and luxury collections took a share of mentions far above their share of actual rooms. In Los Angeles, the Dorchester Collection's two hotels pulled roughly 6% of every hotel mention in the market; Omni took 5% on a single property, Millennium 3% on one. Presence in the answer ran well ahead of presence on the ground. It runs flatter outside the US — Accor and Marriott near 11% each in Europe, Tokyo's top twenty names splitting 64% of mentions against three names taking 54% in the US — but the shape holds everywhere Lighthouse looked.
The catch. Review scores barely moved any of it. Lighthouse found the correlation between a hotel's Booking.com score and its ChatGPT visibility was weak. The metric a property builds one stay at a time, the thing it defends on every OTA, is close to invisible to the model deciding who gets named. Two hotels with identical guest satisfaction can land on opposite sides of the recommendation, separated by how often the wider web has named them.
The chains already know. Accor is treating it as strategy. Its loyalty and ecommerce chief, Mehdi Memici, told WIT Middle East that member experiences now feed discoverability — "differentiation used to be social media led. Now it's actually AI search led" (his framing, from a chain promoting its own loyalty play). The assets he means are concrete: Accor renewed its Paris Saint-Germain partnership through 2030 and runs member-only events like the Dream Tournament, the kind of talked-about, name-building moments the model picks up. That's the conversation the model rewards, and the biggest European group is building it on purpose.
Why it matters for hotels. AI visibility measures the conversation about a hotel, not the hotel. The model draws on what the open web has said about a name, repeatedly, in places it was trained to trust — the asset a global chain builds over decades and a four-room boutique with a 9.4 has almost no way to manufacture. The reviews still convert the traveler who finds the property; they have little say in whether the model names it first. A hotel reading AI visibility as a score to climb is watching a gauge wired to something it can barely touch.
Enjoying this analysis? Hospitality.today delivers daily insights on hotel distribution, AI trends, and travel commerce — straight to your inbox. Subscribe for free at Hospitality.today →