The booking is no longer where you lose

AI doesn't need to process the transaction to control where travelers go. That's the part most hotels are missing

When OpenAI announced it was stepping back from in-chat hotel checkout, the reaction in distribution circles was relief. One less competitor at the point of transaction. The OTAs could breathe again.

That reading misses what is actually happening.

Whether or not ChatGPT processes your guest's credit card is not the question that should be keeping you up at night. The question is whether your hotel gets mentioned at all — before the traveler opens Booking.com, before they type anything into Google, before they have formed an opinion about where they are going to stay. That earlier moment, the one where a shortlist gets built and everything else gets filtered out, is where the real competition is now being decided.

How the funnel used to work

For the past twenty years, travel distribution was a funnel that started wide and narrowed at checkout. A traveler had a vague idea — a city, a date range, a rough budget — and they went looking. They searched. They browsed. They compared. OTAs built entire business models around that browsing behavior, and they were very good at it. Wide inventory. Easy filtering. Side-by-side price comparison. Package deals. Review scores. All of it designed to keep the traveler inside the platform long enough to buy.

Hotels complained, correctly, that this made them interchangeable. But at least they were visible. If you were listed, you had a chance. The guest was looking at twenty options. You just had to be better than nineteen of them on the metrics that mattered to that particular traveler on that particular night.

What AI changes

An AI assistant does not show you twenty options. It shows you three. Or one.

The traveler types something conversational — "a good independent hotel in Lisbon near the waterfront, not a chain, somewhere with character" — and the model responds with a recommendation. It has already done the filtering. The guest did not browse. They did not compare. They received an answer, and in most cases they will act on it.

This is not a small change to the funnel. It is a structural change to where decisions get made. The moment that used to happen inside an OTA — the moment of narrowing down — now happens inside an AI interface, before the OTA is ever opened. By the time the traveler clicks through to book, the choice may already be settled.

For OTAs, this creates a problem that goes beyond losing booking volume. Their merchandising model — upselling room upgrades, bundling car rentals, promoting sponsored placements, steering guests toward higher-margin inventory — depends entirely on the traveler being present and browsable inside the platform. If the traveler arrives knowing what they want, that whole layer collapses. The take rate holds but the upsell opportunity is gone, and the upsell is where much of the margin lives.

What it means for your hotel

Independent hotels are, in one sense, well-positioned for this shift. The characteristics AI tends to highlight when travelers ask for "something with character" — distinct design, local ownership, genuine story, specific location — are exactly what independent properties have and what chain hotels struggle to manufacture. When someone asks an AI for a boutique hotel with a rooftop bar in a particular neighborhood, the answer will not be a Marriott.

But being well-positioned in theory is not the same as being recommended in practice.

AI models surface hotels they know about. They draw on information that exists: review content, editorial coverage, descriptions on the web, data from booking platforms. A hotel with a thin digital footprint — outdated website copy, few quality reviews, no presence in the kind of editorial content that AI training data tends to include — is simply not in the pool, regardless of how good the property actually is.

There is also a subtler problem. AI recommendations are not neutral. They are shaped by what information is available, how that information is structured, and how clearly a property communicates what it is. A hotel that describes itself in vague aspirational language — "an escape," "a home away from home," "a sanctuary" — gives an AI model very little to work with when a traveler asks a specific question. A hotel that clearly states it has a rooftop bar, is a ten-minute walk from the main cathedral, has been operating under the same family for thirty years, and has a restaurant run by a local chef gives the model something it can match against a query.

This is not SEO. It is something closer to being legible — making sure that what you are, and what makes you worth staying at, is expressed clearly enough that an automated system can recognize and communicate it.

The upstream problem

The honest answer to where independent hotels should be focusing attention right now is upstream of the booking, not at the booking itself.

The distribution conversation in hospitality has been dominated for years by commissions, rate parity, direct booking incentives — all of which matter, but all of which assume the traveler has found you. The question that is becoming more urgent is whether the traveler ever encounters you at all.

This means paying attention to how your hotel is represented across every surface where AI models might look: your own website, major review platforms, travel publications, editorial directories, booking platform descriptions. It means writing about your property in concrete, specific terms rather than generic promotional language. It means earning the kind of coverage — in trade publications, in travel media, in owner forums and review sites — that gives AI systems something substantive to draw on.

OpenAI stepping back from checkout does not change any of this. The transaction was never the point. Control over the recommendation is the point, and that battle is already underway.

The hotels that understand this now will be in the conversation. The hotels that wait for the booking data to confirm the trend will be wondering why occupancy is softening with no obvious cause.

by Markus Busch, Editor/Publisher Hospitality.today

Enjoying this analysis? Hospitality.today delivers daily insights on hotel distribution, AI trends, and travel commerce — straight to your inbox. Subscribe for free at Hospitality.today →