The AI guest is still rare

Two numbers describe the guest who arrives from a chatbot, and both are true.

In July, hotel and short-term rental websites got four times the share of their visitors from AI assistants that they got a year earlier. After all that growth, it was still less than one visit in a hundred.

The hype runs on the first number. The yawn runs on the second. A hotelier deciding what to spend time on this winter needs both, and a few more that have come out this year.

How small it is

The traffic figures come from Similarweb, published by Bain & Company in a report on September 30. Between July 2025 and July 2026, the share of visits that AI platforms sent to hotel and short-term rental websites quadrupled. Every travel category Bain looked at, OTAs and cruise lines included, stayed under 1% of total traffic. Similarweb tracked only three to five sites per category, so this is how a handful of large brands fared. A forty-room independent may look quite different.

The biggest seller of hotel rooms counts the same way. On its earnings call in August, Booking Holdings said traffic from AI assistants, paid and unpaid together, was significantly below 1% of its room nights, and that the share had not moved in quarters.

The AI guest is real. She is also rare.

Why rare doesn't mean passing

The people planning trips with AI are not rare at all. Phocuswright's March report found that 56% of US leisure travelers used AI for at least one trip in the previous twelve months, up from 43% in the second half of 2025 and 33% in the first half. Its researchers called it the fastest change in traveler behavior they had measured in more than a decade.

They keep using it, too. Bain's own consumer research found that 49% of US travelers who have used AI for travel now rely on it most or all of the time. Across the UK, France and Germany the figure is 41%. Bain doesn't publish when the survey ran or how many people it asked, which is worth knowing before anyone quotes it in a budget meeting.

Put those together and the small traffic number reads differently. Most of the planning happens inside the conversation, where your analytics can't see it. What reaches your website is the last step.

She now books like everyone else

Adobe measures what happens when that last step lands. Its figures cover visits to US travel websites, and in July 2026 traffic arriving from AI sources was up 119% on the year before.

The more useful number is conversion, the share of visits that end in a purchase. In July 2025, visitors arriving from AI converted 47% less often than everyone else. By May 2026 the gap was 28%. By July it was 1%.

A year ago the chatbot visitor looked far more than she bought. Across travel sites, she now arrives, compares and books at almost exactly the rate of the guest who came from Google or an email. Adobe sells tools for tracking AI traffic, so it has an interest in the trend, but these are counted visits and purchases, not answers to a survey about intentions.

She still checks you

The conversion figure doesn't show where she went between the chatbot and the booking. Phocuswright found that half of travelers who use AI answers in search engines still click through to the source website afterward.

In June we reported what that visit is for: after an AI recommendation, well over half go on to the property's own site, and the travelers who trust AI check almost as often as the ones who don't. She arrives having read about you somewhere else, perhaps fresh from an OTA listing that showed her a total for the whole stay. Your site gets one look to confirm it.

Fifteen years of the same number

Bain's report holds a chart that deserves more attention than its advice. Built on Euromonitor data, it shows travel spending booked in advance growing from €1.1 trillion in 2011 to €2.4 trillion by 2024 and 2025. The share booked direct sat at about 45% in every year shown.

Through the rise of the OTAs, and through the arrival of the smartphone, that line barely moved. The chart covers all travel, flights and cruises included, so it can't be read as a hotel figure.

Bain's own reading is that the flat line hides the work behind it. Travelers moved to the OTAs in large numbers, and hotel brands spent heavily on software to manage their guest relationships just to hold the share where it was. A steady number was the result of a long fight, not an absence of one.

The fight over who books the room is old. The assistant at the start of the trip is the newest place it is being fought.

What it means for hotels

The AI guest is too small to rebuild anything around this winter, and growing too fast to leave unmeasured. The useful move is to find her in your own numbers.

Open your website analytics and filter the last twelve months for visits referred by chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. Ask three questions. How many visits came that way. How they converted against Google organic. Which page they landed on.

For most independents the first answer will be small. The second tells you whether Adobe's July figure holds for your hotel. The third shows you what she checked when she arrived, and whether that page answered her.

Then run the same report every quarter. It is the one AI number you will have that nobody is selling you.