Booking Holdings said referrals from large language models still account for under 1% of Booking.com’s room nights. Chief financial officer Ewout Steenbergen said on the company’s third-quarter 2026 earnings call on November 3: “LLM referrals have accelerated over the last several weeks but still account for significantly less than 1% of Booking.com’s room nights”, in his prepared remarks (call transcript).
Booking Holdings sold 345 million room nights across its brands in the quarter, up 7% from a year earlier, according to its earnings release. The company did not give the number of room nights that came from LLM referrals.
Business Pill · Tool Use
An evergreen Business Engineer explainer on how an AI model goes from text to action. It teaches the general idea only and does not discuss Booking or Uber.
The key insight: Booking Holdings said on its third-quarter call that LLM referrals have accelerated in recent weeks but still account for significantly less than 1% of Booking.com’s room nights, while users of its own AI assistants converted at about twice the rate of non-users.
Steenbergen said organic referrals increased after the release of the latest ChatGPT models, which he said “improved citation rates and drove more traffic to our platforms”. He said paid referrals are also growing as Booking continues to partner with OpenAI by expanding into additional markets.
Asked by J.P. Morgan analyst Doug Anmuth whether newer agents such as Muse and Grok Bot differ from earlier chatbots, Steenbergen said: “what we are seeing that this traffic is still very small in aggregate”. He added that, based on third-party data sources, “the vast majority of that traffic is coming to us”.
Chief executive Glenn Fogel said many people use AI for travel discovery but that “It doesn’t end up with a lot of bookings”, adding: “the referral numbers are pretty darn small”.
Share of volume from LLM referrals or AI agents, as Booking Holdings (3 Nov) and Uber (4 Nov) disclosed on their Q3 2026 earnings calls. Booking gave only a ceiling, significantly less than 1 percent, drawn at 1 percent. The Uber share is our arithmetic: about 504 agent orders and rides of about 320 million.
What Was Posted on X
$BKNG Booking: “LLM referrals have accelerated over the last several weeks but still account for significantly less than 1% of https://Booking.com’s room nights.”
A post by The Transcript at 19:12 UTC on 7 November quoting Booking’s line from its third-quarter call that LLM referrals remain significantly less than 1% of Booking.com’s room nights. It quotes the account’s own 4 November post on Uber’s Halloween-week figures: about 320 million rides and Eats orders, four rides and about 500 Eats orders from AI agents. The line that agentic commerce ‘still has a long way to go’ is the poster’s view.
Where Booking Reports AI Gains
Steenbergen said Booking.com’s daily AI search visitors were up 10x with trip search in the third quarter compared with the second, and that users of Priceline’s Penny and Agoda’s AI Assistant converted at approximately twice the rate of non-users.
He said OpenTable’s native integration with OpenAI “had in September 190,000 seated covers”. Fogel said customer service costs declined in the third quarter despite higher booking volumes, giving it as an example of AI helping Booking operate more efficiently.
The Uber Comparison
Uber disclosed its own figures on its third-quarter call on November 4. Chief financial officer Balaji Krishnamurthy said Uber handled about 320 million trips across Uber and Uber Eats in Halloween week, of which around 500 Eats orders and four rides came through AI agents. We covered that disclosure in detail (our piece).
Visa said this week that 22 AI models from 12 labs failed 42.8% of its shopping test runs (our piece).
Our Numbers
By our arithmetic, Uber’s four agent rides were about 0.0000013% of the week’s 320 million trips, and the 500 agent Eats orders about 0.00016%. Booking’s figure is a ceiling: under 1% of Booking.com’s room nights, with no exact share given.
OpenTable’s 190,000 seated covers through OpenAI in September average about 6,300 a day over 30 days, also our arithmetic.
The Structural Read
As we read the call, Booking’s AI numbers sit mostly inside its own products: 10x daily AI search visitors on Booking.com and about twice the conversion for Penny and Agoda’s assistant users.
As we read it, the referral figure is a ceiling rather than a count: under 1% of Booking.com’s room nights, with no exact share and no figure for Priceline, Agoda or Kayak.
By our arithmetic, Uber’s four agent rides were about 0.0000013% of a 320 million-trip week, so both companies reported agent-sourced volume as a very small share of the total.
Glenn Fogel, Booking Holdings CEO, Q3 2026 call
“the referral numbers are pretty darn small”
Three Implications
FOR TRAVEL PLATFORMS Steenbergen said that, based on third-party data sources, the vast majority of LLM and agent travel traffic is coming to Booking.
FOR AI ASSISTANTS Steenbergen said paid referrals are growing as Booking continues to partner with OpenAI by expanding into additional markets.
FOR AGENT COMMERCEUber’s CFO said about 500 Eats orders and four rides came through AI agents in a week of about 320 million trips.
The analysis says: “With fewer than one in a hundred individual users on the agentic surface, the conversational interface still overwhelmingly owns the consumer base.”
As we read it, Booking’s under-1% referral share and Uber’s four agent rides are company-reported data points on the same gap between AI discovery and AI transactions. The analysis does not discuss Booking or Uber.
What Is Not Established
Booking gave a ceiling, not a count, for LLM referrals, and the figure covers Booking.com only, not Priceline, Agoda or Kayak. The call did not say how many travellers used an AI assistant to research a trip and then booked directly. We have read the call through a published transcript, not Booking’s audio. This piece makes no forecast.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Booking Holdings said LLM referrals accelerated in recent weeks but remain significantly less than 1% of Booking.com’s room nights, while its own AI assistants converted at about twice the rate of non-users. Uber reported four agent-sourced rides in a 320 million-trip week.
A note on sourcing. Booking’s figures and quotes are from its Q3 2026 earnings call on November 3, read in a published transcript, and its earnings release. Uber’s are from its Q3 2026 call on November 4, as covered in our earlier piece. The percentage shares and the daily OpenTable average are our arithmetic. This is not investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
