Booking.com reports that less than 1% of Booking Holdings’ room nights can be linked to AI tools such as ChatGPT, while this percentage has remained almost unchanged in recent months. The company estimates that, based on its own measure, AI has not yet begun to disrupt the way bookings are made in any significant way.
The figure released by Booking is specific: it concerns only those who used an AI tool and subsequently went on to make a booking through the platform. The percentage, however, does not capture travelers who used AI for their research and then booked directly with a hotel or a short-term rental provider.
What the percentage measures and what it does not
In its update on its second-quarter 2026 results, Booking Holdings CFO Ewout Steenbergen said that traffic from artificial intelligence tools based on large language models, whether paid or unpaid, remains “well below 1%” of room nights and that this share “hasn’t moved much recently.” The company is therefore referring to completed bookings, rather than simple visits or clicks.
CEO Glenn Fogel added that, according to third-party data, Booking frequently appears in AI tool responses when users search for travel information, but this appearance does not necessarily mean corresponding referrals to the platform. As he said, the company sees that it “shows up a lot,” but referrals to the platform do not appear to be particularly high.
He also acknowledged that he does not know whether people who begin their research with AI ultimately book directly with hotels or other providers. He estimated that it may not represent a “huge” volume of business, but clarified that this is speculation and not a measured figure.
Why the finding matters for accommodation providers
For hotels, property managers and hosts, the issue is not only whether AI sends bookings directly to Booking. The key question is whether it affects the research phase, meaning where travelers form their opinions before deciding where to spend their money.
Booking argues that much of its value still relies on users’ trust. Users may conduct their research wherever is most convenient for them, but when it comes time to make the final decision, they turn to a platform or brand they already know. In this context, the figure below 1% indicates that the company still retains most of the booking funnel.
At the same time, Booking itself is moving in both directions. It is participating in tests of paid placements within third-party AI tools, while also investing in its own AI features for discovery and booking within its own environment.
The Outlook for the future remains open
Booking also acknowledged that traditional search through Google is under pressure, attributing the change to AI-generated answer summaries. At the same time, however, it maintains that its direct relationship with travelers has not changed substantially.
For the hospitality market, this means that AI has not yet been proven to be a direct threat to bookings that go through major OTAs. But the measurement gap remains: no one knows precisely how many travelers use AI as a research tool and then book outside the platform.
This is also the point that matters most to short-term rental and hospitality professionals. If AI evolves into a primary way of searching, visibility within these tools could become more important than what a platform’s official booking metrics currently show.
Frequently Asked Questions
Why doesn’t the “below 1%” figure tell the whole story?
Because it measures only the bookings that went through Booking after the use of an AI tool. It does not measure those who researched through AI and booked directly elsewhere.
What does this mean for an independent accommodation?
It means that visibility in AI research tools may become more important, but Booking has not yet shown, based on its own data, that AI has shifted a significant volume of bookings away from its platform.
