The Booking paradox: record revenue, falling profit
One figure from Booking's 2025 results sums up the moment in travel distribution better than any analysis: the company billed more than ever — 22.82 billion euros, up 13.4% — and earned 8% less: 4.58 billion in net profit, according to its official results presented in February 2026. Selling 1.235 billion room nights (+8%) and making less money demands an explanation. And the explanation directly concerns any hotel paying commissions.
Where the money went
Two items explain the drop. The first: around 770 million euros of extra marketing spend, aimed mostly at Google, plus metasearch engines, social networks and television. The second is more symbolic: a 388-million-euro accounting write-down of Kayak, its metasearch engine — an acknowledgement that the asset is worth less because AI search is eating its ground. Kayak is, in a way, the first "accounting casualty" of conversational search: a business built on comparing links, in a world where the answer arrives already compared.
The industry context fits the pattern: in the same year, Airbnb earned 2.51 billion dollars (−5%) on revenue of 12.24 billion (+10%), and Expedia 1.29 billion (+5%) on 14.73 billion (+8%). Growing revenue costs ever more profit.
The chain of tolls
Here is the paradox in the title. Booking charges hotels typical commissions of 15-25% for delivering demand. But a good part of that demand Booking buys, in turn, from Google. The OTA that is the hotel's toll pays its own toll to the search engine — and that toll is growing: in the first quarter of 2026, Booking spent 2.1 billion dollars on sales and marketing, 16% more than the year before (38% of its revenue). Expedia spent 1.9 billion (+6%, 54% of revenue) and Airbnb 751 million (+33%). The four big OTAs spent 20 billion dollars on marketing in 2025, up from 17.8 billion in 2024 (PhocusWire).
The final irony: the collector of the second toll, Google, is becoming a direct competitor. Its AI Mode already processes hotel bookings in the United States with partners such as Marriott, and its AI products are reshuffling the travel search Booking has been buying for twenty years.
Wasn't AI supposed to make marketing cheaper?
The efficiency narrative is not showing up in the budgets: all the big OTAs are spending more, not less (PhocusWire, June 2026). Their own executives explain why: Glenn Fogel, Booking's CEO, argues that the performance-marketing expertise accumulated with Google is transferable to AI platforms; Ariane Gorin, Expedia's CEO, declared that answer engine optimisation (AEO) is her company's fastest-growing channel, and that its ChatGPT ads yield small but promising returns. AI hasn't cut marketing: it has opened a new front on which to bid again.
What this means for a hotel
Three practical readings, from fifteen years negotiating with clients and suppliers at a tour operator:
- Your commission funds a war that isn't yours. Part of what you pay the OTA becomes Google bids and, now, advertising inside assistants. It is legitimate — it is their model — but worth knowing when you negotiate and when you decide how much direct business you want to claw back.
- Pressure on OTA profits doesn't make them softer, it makes them more aggressive. An intermediary with narrowing margins defends its share: loyalty programmes, parity, more marketing. Expecting AI to "weaken Booking" in the short term is bad strategy; building your own channel in the meantime is not.
- The underlying shift favours whoever owns the direct relationship. If the chain of tolls gets longer (hotel → OTA → Google → AI assistant), every link takes margin. The only move that shortens the chain is the customer who books, writes and returns to you directly. Everything you invest there pays no toll to anyone.
The Booking paradox is not a passing weakness: it is a snapshot of a model renegotiating who charges for access to the traveller. While the giants split that bill, the best position for a small business is to owe the toll chain as little as possible.
