When a guest asks ChatGPT what your hotel costs, whose price do they hear?
A test of 99 ChatGPT hotel price queries found 95 returned a price but only 18 specified the date, with OTAs cited far more often than hotel direct sites.
Photo by Lumina
On 15 November I could give you two good answers to the question “What does a room at Ashford Castle cost?” The castle’s own booking engine offered a Corrib for two at €527 on Tuesday 9 February and €1,250 on Tuesday 15 June. Both dates are in 2027. Both prices were right when we checked. Imagine being the person at reservations who picks up the phone after a guest has heard just the first number.
We check hotel rates for the guide we publish at Lumina. In October a reader sent us the answer ChatGPT had given him about the Venetian in Las Vegas. It had used our guide. The part I kept was its warning about our figures: “historical quotes, not current offers”. To compare them fairly, it wanted prices for the same suite on the same night, the cancellation terms and an answer to a very Las Vegas question. Was the resort fee in?
Who gets to quote the room?
That exchange prompted us to put our own hotel price questions to a ChatGPT model on 31 October, with web search on and a US location. We saved its answers and the links behind them. These were questions we wrote for a test. The reader’s Venetian question came from his own conversation.
When we asked what a night cost, 95 of 99 answers gave us a price. Only 18 said which night it bought. For a revenue manager, the missing date matters more than another decimal place. If you run Ashford Castle, which of those two Tuesdays would you like a guest to have in mind?
In those same 99 answers, 80 cited a booking site or Google Hotels. We could identify the hotel’s or its group’s site in 11. An answer could cite both. The distribution shifts when the guest asks whether to book direct: we found an official hotel or group site in 61 of 100 answers to that question. We published the wording and results in our study of hotel prices in ChatGPT.
What can the hotel say that a booking site can’t?
At Ashford Castle, the useful detail on that February Tuesday was that the winter offer cost €527, while the same Corrib on the castle’s bed-and-breakfast rate cost €585. The cheaper offer included breakfast and a welcome drink. By Friday the winter offer no longer started that night. A visitor who knew only the room name and the first price would have little chance of working that out.
Reading the official pages ChatGPT cited elsewhere, I kept finding the pages that reservations teams could have written: Marina Bay Sands’ booking terms, the Broadmoor’s policies, an offer at Resorts World Las Vegas that explicitly did away with the resort fee. They answered questions a guest might otherwise have to ring up and ask.
What I’d ask revenue and marketing to publish together
I’d start with the offer reservations has to explain most often. I’d ask revenue to pick a real room and two dates that show why its price changes. Marketing can put both on the offer page, alongside the total for two, what breakfast costs, when the guest can cancel and the date the figures were checked. Leave a link to live availability for the person ready to book.
Then I’d give that page to whoever answers the phone on Friday. If they still have to explain why Tuesday’s rate has disappeared, that’s the sentence I’d ask them to write.
Marco Di Cesare is Founder & CEO of Lumina Frontier, a Zurich company building an AI marketing team for hotels.
Comments
Comments for this content
0 comments available