AI agents are reshaping online travel. Who wins and who loses?

AI agents are exerting pressure on conventional online travel agencies (OTAs) such as Booking Holdings and Expedia, potentially reshaping the landscape of digital travel. According to Bernstein analysts, while the immediate financial consequences remain modest, the long-term outlook for traditional OTAs is less favorable. AI agents present three primary risks: disintermediation, pressure on commission rates, and heightened price competition.

These agents can independently search for hotels and other travel suppliers, potentially diminishing the advantage that OTAs derive from aggregating inventory. Moreover, they prioritize customers' preferences over inventory, which may generate higher commissions. This is particularly significant for Booking, as directing customers towards higher-commission inventory could boost revenue by an estimated 10%.

Price discovery represents another threat. Currently, Booking and Expedia brands secure the cheapest hotel rate in U.S. searches only about 15% of the time. Conversely, smaller OTAs achieve this feat 77% of the time. This creates an opportunity for AI agents capable of searching across more providers. Nevertheless, existing technology still affords some protection to Booking and Expedia.

Their chat-based agents struggle with complex hotel searches involving maps, amenities, and multiple filters. Additionally, large OTAs possess extensive inventory originating from a single source.

However, Airbnb appears to be better positioned. Approximately 70% of its nine million active listings are exclusive to the platform, restricting agents' ability to locate identical properties elsewhere. Moreover, Airbnb attracts about 45% of web traffic directly, reducing its reliance on paid search. Its customer relationship extends beyond bookings, as around 90% of guests communicate with hosts post-reservation, providing Airbnb with additional opportunities to retain users on its platform.

The revenue model adopted by AI agents could ultimately determine their impact on OTAs. User-funded agents seeking unbiased results pose the most substantial threat to OTAs, while advertising-supported models may preserve a more significant portion of the existing travel distribution economics.