Travelers aren't using AI to find your hotel. They're using it to be sure about it.
Two-thirds of pre-booking queries on Trip.com's AI are about conditions, not price — and unclear terms lose the booking
The assumption. For three years the industry has treated AI as the new front door — the discovery layer that decides which hotel a traveler finds. The companies actually running these assistants describe something else. Trip.com, presenting three years of its own usage data at WiT Japan, says the main job its AI does is not finding hotels. By the time the tool reaches your property, the traveler's question has moved. It is no longer "which hotel." It is "am I sure about this one."
What the data shows. Read the figures as one OTA's, not the market's — this is Trip.com's own product. Two-thirds of the questions travelers ask before booking a flight are about conditions: baggage, cancellation, lounge access. Not price. Real-time use during the trip itself is climbing fast — menu translation, local navigation, what to do when a plan falls through. The assistant has moved off the planning desk and into the journey, answering doubt in the moment it arrives. Across three years the pattern holds. People reach for AI to understand what they are actually buying, and to decide with less doubt. The hotel version of that question is familiar to any front desk: can I cancel, is breakfast included, will the late arrival hold, is the room actually quiet.
Compression, not discovery. Trip.com's hotel-comparison tool cut the average browsing session from several sittings to under two. The value was not more information. It was a confident yes, reached faster. Discovery widens the set of options. This narrows the set to a decision. Those are different jobs, and they happen at different moments in the trip.
What it means for hotels. Getting recommended puts you on the shortlist. It does not close the booking. At the decision point, the AI answers the traveler's last questions for them — what's included, what the cancellation terms are, whether the room fits the trip — using whatever data it can reach about you. If your terms sit in a PDF, or read one way on your own site and another on an OTA, the AI cannot settle the doubt. Picture the same question two ways. One hotel states, in plain text a machine can read: free cancellation until 48 hours out, breakfast included, quiet rooms at the back on request. The next offers "flexible rates" and a link to a policy document. Asked "can I cancel this?", the AI answers the first cleanly and hedges on the second. It moves the traveler to a property it can speak for. The clear policy wins the booking the vague one loses. Same room; one of them the machine could vouch for.
The booking no longer turns only on whether AI found you. It turns on whether AI can be sure about you.
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 →