OTAs Win 55% of AI Hotel Citations as Agents Reshop Stays

While Booking Holdings told investors that large language models generate well under 1% of its room nights, online travel agencies already dominate 55.3% of the sources cited in artificial intelligence hotel recommendations [[1], [2]].

The mechanics of travel commerce connect consumer discovery queries directly to autonomous booking agents. How properties surface during conversational research dictates whether a guest books directly or routes through an intermediary. Two parallel developments reveal how this shift functions: information engines query third-party platforms to answer guest requests, while autonomous transaction bots negotiate checkout rails without direct merchant contracts.

Online Travel Agencies Dominate the Reference Layer

AI search models replace standard lists of web links with direct conversational answers drawn from multiple databases [1]. According to eHotelier, an analysis conducted by Cloudbeds showed that online travel agencies accounted for 55.3% of sources cited by AI systems recommending hotels [1]. Direct hotel domains, by contrast, generated only 13.6% of citations [1].

Google confirmed that its AI Mode and AI Overviews rely on regular web retrieval systems rather than separate technical infrastructure [1]. In 2026, Google introduced specific Search Console metrics enabling operators to monitor impressions from AI Overviews and AI Mode [1]. If property data varies between direct pages and distributor feeds, algorithmic discovery favors the most consistent third-party repository [1].

Autonomous Software Bypasses Hotel Channels at Checkout

Search visibility alone does not secure the commercial booking rail. As Skift reported, autonomous agents are establishing transaction routes independently of accommodation suppliers [2]. When Meta’s Muse agent searched for a New York hotel stay, the tool opened an independent browser and picked between Expedia and Hotels.com without prompting, bypassing direct hotel channels entirely [2]. In the airline sector, Muse executed bookings through a direct Duffel integration, but hotel connections defaulted straight to online distributors [2].

Even when hotels join these systems, transaction control diminishes. When bookings occur via Google’s AI Mode agentic feature, the hotel or distributor remains the merchant of record [2]. Yet Google passes along only basic guest data: name, property, occupancy, and payment [2]. The upstream conversation that shaped the booking remains inside Google, denying hotel revenue managers direct behavioral context [2].

Algorithmic Reshopping Weakens Standard Loyalty Pricing

Direct commercial terms face pressure from automated repricing tools. A joint study from McKinsey and Skift Research showed that 50% of 1,000 surveyed U.S. travelers altered an existing booking after finding a superior option [2]. Hilton stated that it anticipates agents will continually reshop reservations up to the day of departure [2].

Skift Research documented that 44% of U.S. travelers bypass preferred loyalty tiers when presented with cheaper rates, while 55% follow an AI recommendation to secure a lower price [2]. Loyalty membership also placed last among traveler justifications for trusting an AI recommendation [2].

The Mistake Operators Make With AI Distribution

The operational mistake hotels make is delaying commerce adjustments because current transaction volumes appear negligible [2]. Bernstein analyst Richard Clarke estimated AI referrals at below 2% across Booking and Marriott [2]. Because these percentages remain low, operators frequently defer tech investment [2].

According to Skift, distribution rules and commercial alignments are solidifying before 2027 commerce budgets finalize [2]. Hoteliers who postpone data cleanup and distribution terms will find third-party intermediaries have already captured the payment wallets, customer data, and referral feeds feeding AI commerce engines [2].