The 69% Who Book From an AI Summary Never Saw Your Website

A 2026 Amadeus report found that 69% of travellers trust AI summaries enough to book without further research.

Read that as a hotel. More than two-thirds of the market is prepared to act on a paragraph - two or three property names in a confident answer - without opening a list of results, without reading reviews, without visiting your website. The decision happens inside the answer.

What the funnel used to be

Search. Ten blue links. Open three. Compare on an OTA. Read reviews. Maybe visit the hotel's site. Book.

Every step was a place a hotel could appear. Rank on page one. Have a good OTA listing. Earn reviews. Convert on the site. A property that was weak at one stage could recover at another.

What it is becoming

Ask. Read a paragraph. Book.

There is one stage. The hotel is either named in the paragraph or it is not. If not, it was never considered - not ranked low, not on page two, not losing on price. Absent. And there is no later stage at which the property gets a second look.

The 69% is the share of travellers for whom this is already the funnel.

Why the trust is extended

The answer is confident, specific and short, and it arrives in the same place the question was asked. It reads like a recommendation from a knowledgeable friend rather than a list or an advertisement.

Whether that trust is warranted is a separate question. What matters commercially is that it is being extended, at scale, by a majority of the market, and it is deciding where they stay.

What the assistant needs to name you

Three things, none of them advertising.

Extractable facts. Pages that state, in a sentence, what the traveller asked about - parking, pets, check-in, cancellation, location. A page that says contact us for details contributes nothing to an answer.

Structured data, so the engine reads the facts without interpreting prose. Pages with valid structured data are roughly 2.3× more likely to appear in Google's AI Overviews.

Consistency across sources. The assistant cross-references the site, the OTA listing, the map profile and the reviews. Where they agree, it quotes with confidence. Where they disagree, it hedges or picks a competitor whose sources agreed.

None of this is a campaign. It is the property's information, made legible.

The traveller who arrives after the answer

The 69% who book from the summary are the headline. The rest - the ones who read the answer and then look further - are where the direct booking is won or lost.

They arrive with a specific question the answer did not cover. We land at 7am, can we check in early. Is the parking underground. Is the pool heated. They ask it on WhatsApp, or on the OTA's messaging, or on a contact form.

The property that answers in seconds, in their language, from accurate information, takes the booking direct. The one that replies tomorrow morning loses it to the OTA listing that answered instantly. The messaging assistant is the conversion point for the traveller an AI search delivered.

The honest caution

One report, one year. The 69% will move, and the methodology behind it is Amadeus's, not independent. Treat it as a direction with a credible source, not a constant.

But the direction is not in dispute. Every major assistant is building toward answering travel questions with named recommendations, and every survey of traveller behaviour points the same way. The properties that are legible to the assistants will be named. The rest will discover the problem as a slow decline in bookings from guests who used to find them easily.

What to do

Make the twelve most-asked facts about the property clear, marked up, consistent and dated. Then make sure the traveller who arrives with the thirteenth question gets an answer before the OTA gives them one.

That is the whole response to the 69%. It is a day of work and a messaging assistant, and it is what being in the answer now requires.