Airbnb and Hotels.com Now Write Hotel Descriptions, Not Hotels
Airbnb began auto-writing property descriptions on September 30; Hotels.com followed four weeks later with per-traveler AI comparisons. Hotels now answer to pitches they never wrote.

Airbnb started writing its own descriptions of listed properties on September 30, replacing host-authored copy in search results with AI summaries built from reviews and listing details. Four weeks later, Hotels.com began generating a different comparison pitch for every individual traveler. Hotels no longer control the sentence that sells their rooms.
On Airbnb, the company calls these AI-powered listing descriptions. They appear in search results beside the host's photos and title, picking out what Airbnb considers a property's key features. A second tool, Compare, lines up a traveler's saved places and gives each one a short summary beside the price and review highlights. Airbnb says the summaries draw on two inputs: reviews, and the details hosts provide.
The reviews had already moved up the page. In May, Airbnb began pulling recurring themes from a listing's entire review history and displaying them near the top. The platform has more than a billion reviews to draw from. The same release added boutique hotels in 20 cities.
What Airbnb has not said is whether the new descriptions run on hotel listings, or whether a hotel or host can see the line before a guest does, let alone change it. The rollout started in the US.
What is Hotels.com doing differently?
Hotels.com goes further. Its AI Compare, announced in late October, produces what the company calls "personalized, plain language comparisons based on expressed interests" — location, amenities, price and policies — and the traveler can keep asking follow-up questions. A second tool, Property Expert, answers questions about a single hotel on Expedia and Hotels.com, working from property details, guest reviews and the traveler's own reservation. Both run in the US for now.
Google's AI Mode does the same job in search. The traveler describes the trip, and it returns hotels with guest reviews and the factors worth comparing.
The consequence is structural. A guest who mentioned her car gets the hotel compared on the garage; a couple wanting a quiet weekend gets it compared on the rooms at the back. Same hotel, two pitches, and the hotel wrote neither and will read neither. A positioning statement used to be one sentence, chosen with care and repeated everywhere. It is now as many sentences as there are travelers asking.
Why does the feedback loop matter?
A guest judges a stay against what she expected, and she expected what she was told — increasingly the platform's line. Her review measures that promise, not the hotel's own. If the comparison sold quiet and her room faced the tram stop, the review says so.
Then the review feeds back in. Airbnb names reviews as an input to its summaries; Expedia grounds Property Expert in them. The next traveler's pitch is built partly from what the last guest wrote about the last pitch. A hotel that renovated, replaced its restaurant or repositioned toward couples can rewrite every line on its own website, while the summary keeps reading guests who stayed before the change.
Independent hotels feel this most. A chain's name tells the traveler the tier, the room and the rate before any summary is written. An independent is known by what its guests wrote, so the summary is the whole introduction.
Is Booking.com following?
The largest hotel platform is holding back. Booking.com announced review summaries in October 2024 as a test, designed to give travelers key points without scrolling. It has not confirmed them for hotel pages since, though it shipped AI review summaries for car rental companies this year. On Booking.com, a guest still reads the hotel's own description and raw reviews.
That gap gives operators time to audit their position. Operators can run the comparison themselves — on Hotels.com for US bookings, in ChatGPT or Google's AI Mode otherwise — naming their hotel against two competitors, for a couple, a family and a business traveler. Where the AI answer matches the sentence the hotel would write, the record works. Where it does not, the gap closes not through the website but through what the next hundred guests write, which depends on what they find at the door.
More from Marcus Bennett
Show full bio
Market editor covering technology and product news in Restaurant and hotel operations at The Pass Brief.
798 articles


