When an agent books a hotel, who decides where the booking lands?

The hotel is the company actually providing the room.

If the price and conditions are the same, the AI agent should book directly with the hotel instead of through a middleman like Booking.com.

Direct booking can give guests:

• better prices
• loyalty points and perks
• easier changes and cancellations
• more accurate information
• better customer support.

It also helps hotels keep the customer relationship instead of paying large commissions to OTAs.

If AI agents are supposed to work in the customer’s best interest, a “direct-first” booking approach makes the most sense.

Google has already announced: they will not be the merchant of record and they're not even charging a transaction fee.

Needless to say, this is a great precedent but it doesn't mean other AI platforms will follow this example.

At the same time, there's nothing that prevents us from pushing AI platforms in the direction of a direct booking path integration. And we will continue pushing.

I am firmly of the view that Google will eventually present all offers openly. Complementing that will be the savvier user who explicitly directs their preferred booking channel. If I am a Bonvoy member, I want to book at marriott.com to protect my loyalty assets.

Right now, it's a question of structured supply. The most expedient path for AI Search players has been connecting with the industry giants first—a sad indictment on smaller operators who haven't gotten their technology houses in order.

However, Google’s core mantra is to shorten the distance between the consumer and the product; they don't want to create middlemen. This is an operational lifeline for hoteliers. Because travel is a high-ticket purchase, currently a massive 89-point trust gap exists: while 91% of travelers plan with AI, only 2% allow full autonomy to book. Consumers demand visual validation before transacting, natively routing high-intent traffic directly to brand.com.

Ultimately, this is a democratic extension of search. Everyone who structures their digital shelf to be accessible, will be. If you are worth finding, you will be found. If you are forgettable, you are forgettable today.

My earlier published article on the details surrounding this development at large - https://travhotech.com/google-agentic-shift-hospitality-distribution/

Google’s Universal Commerce Protocol raises a much deeper issue than booking convenience. Henri’s prompt highlights an important ethical question about AI use: Who should control the routing logic behind hotel reservations?

Travelers should be able to compare hotel websites, OTAs, rates, loyalty benefits, and booking conditions before making an informed decision. Now or in the near future, many of these comparisons are handled invisibly by AI agents to enable faster decision-making. If an AI agent decides whether a booking goes through a hotel website or an OTA before travelers see the options, then distribution power quietly shifts from consumers to algorithms.

This matters because booking paths determine far more than transaction processing. They influence commissions, customer ownership, loyalty participation, data access, and long-term brand relationships. Routing logic, therefore, becomes a new form of digital gatekeeping and deserves policymakers’ attention.

Transparent comparison remains essential. For consumers, AI agents should present multiple booking paths openly and allow price, loyalty benefits, cancellation policies, and user preferences to shape booking recommendations. Meanwhile, hotels and OTAs should also be informed about the standards and ranking criteria influencing those decisions. Responsible AI use should prioritize transparency, fairness, explainability, and consumer agency rather than opaque algorithmic control.

From an industry perspective, the real question isn’t just who completes the booking – it’s who makes the decision in the first place. If AI agents become the main way people search and book, the process shifts from comparing options to being guided toward a choice.

Ideally, the agent would show all relevant options clearly, letting things like price, conditions, loyalty programs, and user preferences drive the decision. That keeps things balanced and supports both direct bookings and partners like OTAs.

For hotels, maintaining flexibility is key. They rely on a mix of channels to reach travelers. As this evolves, transparency in how options are selected and presented will be critical.

At the end of the day, the goal should be simple: make booking easier for travelers without taking away choice.

IMHO, this feels like the natural evolution of what Google already attempted years ago with “Book on Google”: reducing friction and keeping users within the SERP ecosystem.

I loved the idea back then, and I love it now, but with a very big BUT.

To me, the biggest unresolved issue is the infrastructure, or lack thereof, underneath it: most hotel booking engines today are not designed for this. Hell, many of them never even fully caught up with Metasearch!

Which means that a third-party orchestration layer becomes necessary to connect inventory, pricing, availability, loyalty logic, and checkout flows.

I was working with WIHP when we launched Meta I/O, so I have seen firsthand how these things tend to evolve...

Unless booking engines wake up very quickly, we risk creating yet another category of gatekeepers: the platforms that connect the whole ecosystem.

And let’s be honest here, these are not really “direct bookings.” They still carry an acquisition cost, depend on an intermediary layer, and shift part of the relationship and control OUTSIDE the hotel itself. Google says the brand remains the merchant of record, which is technically true. But the merchant of record does not necessarily mean the owner of the customer relationship anymore...

When an agent books a hotel, Google decides where it lands, through routing logic it has not published. That is the distribution decision.

Take a Marriott property bookable through Marriott direct and through Booking.com. Both are in the pilot. A user asks for a room, that hotel matches, and something chooses which connection fulfils it. Whichever one does keeps the guest and the data.

Google has done this kind of fork before. In AI Mode, its restaurant feature handles the discovery and sends the user to a partner like OpenTable or Resy to finish, in an order Google sets and has never explained. Different mechanism from UCP, same shape.

UCP ships free, open source, no protocol fee, hotel as merchant of record. Google does not give away a rail to earn nothing: it charges for the layer above, and it has said so. At Marketing Live in May it confirmed hotel Direct Offers will run from Hotel Center, the console that already prices Hotel Ads on bid against a quality signal.

tl;dr: plan for the version where the routing is sold, where placement in the agent's answer is bought the way position already is on Google's travel surfaces.

The hospitality industry has lived through similar transitions before metasearch, opaque ranking models, and OTA merchandising layers, but the difference here is that agentic systems collapse the funnel further upstream, into intent formation itself.

If the agent becomes the interface of discovery and decision mediation, then transparency of routing logic is not a philosophical preference; it becomes a structural market fairness issue. In that context, “all available offers surfaced equally” is not just an ethical stance; it is a prerequisite for maintaining competitive price discovery.

The unresolved variable is not whether hotels are bookable through agents, but whether distribution remains contestable at the point of intent, or is pre-allocated at the point of algorithmic interpretation.

That is where the real power shift sits.

I think the answer will depend on whose agent (or chatbot) you're using. Google already has a great business model where they locate and shop authoritative sources and present them in order of bid. The simple pay-to-play format supports their strategy and, hey, who can argue with that kind of success? When Meta starts selling travel through their agent, it'll likely do the same.

OpenAI and Anthropic are likely to be different--they'll leave it up to the traveler. They aren't chasing ad revenues or taking a booking fee the way others are. They are looking at a much bigger prize: establishing their agents as the new first point of entry into consumers' digital lives. Need to buy, see, think about, or do something? Let our agent help you. Dominating this space creates value (and lock-in) that makes Microsoft Windows' business model look like child's play. It's in their interest to give the traveler the offers that best match their needs (unaffected by ads) and let them book however they want. Personally, I think they're likely to drop the traveler off at the brand/hotel's front door and let them finalize a direct booking because direct bookings will provide the richest offers.

When your hotel shows up on Google in the future, we should expect Google’s Universal Commerce Protocol to give preference to whatever works best for Google. And that means showing the best choice that encourages guests to return to Google for every reservation. 

Your rates are out of parity? Google will boost the site with the lowest rate. You don’t display quality photos or verifiable details around your rooms or property? Google will show the site that does. All else being equal, they’ll almost certainly give preference to the company that’s most willing to pay. They just had the most profitable quarter in their history. They’ll want to do that again. 

Google is generally good for consumers and frequently good for our businesses. I don’t think they’re ill-intentioned or evil. They are, however, in the business of growing their business, not yours. As a result, they’ll make UCP work in whatever way allows them to drive that growth. Which means that, longer term, your job is to grow your brand beyond just Google. Otherwise, you’re turning over the keys to your success to a gatekeeper who will always act in their own best interest. And that’s the universal truth.

Let's face facts. Google is not doing this out of the goodness of its heart. It's doing it to make money from its near monopoly over online visibility.

Who comes up where, and how often, will not be driven by altruistic notions of equity or fairness, but by which player is prepared to pay the most. As we have seen in search, those with the deepest pockets get prime position, with others regulated to the long tail where no-one ever looks.

Unless hotels fundementally revisit how much they spend on customer acquisition (as well as put in place appropriate systems to facilitate the process), the winners will undoubtedly be the OTAs who, as tech platforms themselves, already understand the rules of the game.

Think of it this way — there are really two ways this could play out.

In the first, more open model, the agent behaves like a helpful advisor. It shows you all your booking options — direct hotel, OTAs, everything — and lets you compare what actually matters: price, cancellation flexibility, loyalty perks, and even your own past preferences. It might recommend something, but the choice is still yours. That keeps the ecosystem healthy — hotels can still drive direct bookings, OTAs have to compete on real value, and the agent simply helps you decide.

In the second, more opaque model, things are less visible. The agent quietly picks one option for you, based on its own internal logic — partnerships, incentives, conversion likelihood, things you don’t see. You’re shown a “best” option, but you don’t know why, or what you might be missing. Over time, this shifts control away from the market and into the hands of the platform, even sidelining direct channels when they might actually be better.

So the big question is: do we keep this transparent, or let it become invisible? Because that decision will determine whether AI becomes a trusted guide — or a gatekeeper

My answer is simple: the agent should present all available options and make a recommendation.

The recommendation can take into account price, conditions, loyalty benefits, user preferences, past behaviour, or whatever context is relevant. That's already incredibly valuable.

What it shouldn't do is quietly decide where the booking goes before the user even sees the alternatives.

If the routing decision is made by the agent rather than the consumer, then we're no longer talking about search or recommendation. We're talking about distribution.

And if a company controls both discovery and distribution, that's where things become interesting from a competition perspective.

As a consumer, I want the recommendation.

I don't want the decision made for me.

I do not think this question has a simple ‘either/or’ answer. The appropriate booking path is often determined by the traveler's intent, preferences, and the inventory available at the time of booking.

Some travelers are highly particular about where they stay and how they book. They may prefer booking directly with a hotel brand because of loyalty benefits, room preferences, elite status recognition, or confidence in dealing directly with the property. In those cases, presenting direct-booking options alongside OTA offers provides meaningful value and transparency.

Other travelers care less about the booking channel and more focused on convenience, location, price, cancellation flexibility, or proximity where they need to be. For them, an OTA may offer the best combination of inventory, rates, and comparison capabilities.

The real question is not whether search engines should route bookings through one partner or another. It is whether decision is driven by traveler context and trip purpose. 

The ideal experience is one where the agent transparently evaluates available inventory, rates, conditions, and traveler preferences, then recommends the best fit or clearly presents choices. The traveler, not an invisible routing rule, should remain at the center of the decision.