What Is Google Doing While AI, OTAs, and Hoteliers Reshape Hotel Distribution?

The exterior storefront of a modern Google office or retail space at night, featuring a glowing white "Google" sign above the entrance and a large red Google Maps pin icon visible through the glass window.

by

The Hotel Blueprint

This is the third article in an ongoing series on AI and hotel distribution. The first established how AI is restructuring the booking path and why independent hotels are losing visibility in the process. The second documented why OTAs are winning that transition — and what it is costing hotels on the acquisition line. This piece takes the argument one step further: while AI platforms and OTAs compete over where travelers book, Google is building the infrastructure to process those bookings itself — directly, inside its own interface, without sending the traveler anywhere else.

1. The Layer Beneath the Routing Problem

The debate over AI and hotel distribution has so far centered on a routing question: when a traveler asks an AI platform to recommend a hotel, where does that recommendation send them? The answer, as the data from 2025 confirms, is predominantly to OTAs. Booking Holdings posted full-year 2025 revenue of $26.9 billion, up 13 percent, while OTA share of independent hotel bookings rose to 63.4 percent globally, according to Cloudbeds’ 2026 State of Independent Hotels Report drawn from 90 million bookings across 180 countries. AI discovery did not compress OTA dominance. It reinforced it — and if you want the financial mechanics of why that happened, the previous piece in this series covers it in full: why AI is currently making OTAs stronger, not weaker.

But the routing question, urgent as it is, is already the second act. Google is writing the third.

Google is not primarily an AI recommendation platform in the ChatGPT or Perplexity sense. It is the infrastructure company that built the original distribution machine — the search engine through which virtually every hotel booking, direct or OTA, originated for two decades. What Google is now building is a replacement for that machine, one in which the traveler’s question, the property recommendation, and the completed transaction all occur inside Google’s own interface. Not routed outward. Contained.

For hotel operators focused on the OTA commission problem, this development sits one layer below the current debate. It has structural implications for acquisition cost, channel dependency, and the question of who owns the guest relationship — implications that are not yet priced into most hotels’ distribution thinking, because the product is not yet fully live. The rate card has not been published. That is precisely why it warrants attention now.

2. Google Is Intercepting the Discovery Moment Before It Reaches Anyone

To understand what Google is building toward, it helps to understand what it has already changed. The mechanism is AI Overviews: Google’s generative summaries that appear at the top of search results, synthesizing an answer before any organic links are visible. In the first four months of 2026, 68 percent of all Google searches ended without a user clicking on any external result — organic, paid, or OTA — according to SparkToro’s June 2026 analysis using Similarweb’s cross-device clickstream panel. That figure was 60 percent in 2024. The two-year acceleration of 7.5 percentage points is the fastest in a decade.

The pattern matters differently for different query types. Pure transactional queries — a traveler ready to book a specific property on specific dates — still generate clicks, because Google’s ad revenue depends on it. Mid-funnel comparison queries — the “quiet hotels near the Amalfi Coast” category where travelers are choosing between destinations and property types — are where the erosion is sharpest. This is the segment where hotels previously competed through SEO and content, earning organic visits by writing useful destination pages. Those visits are now absorbed by Google’s own interface. The hotel contributes to the AI summary without receiving the click. The first article in this series examined why that shift makes independent hotels particularly vulnerable — the properties with the thinnest structured data footprint are the first to disappear from AI-generated results, and that is a problem that predates Google’s paid placement architecture.

The commercial consequence at this stage is not a collapse in bookings — travelers influenced at the AI Overview stage can still complete transactions downstream. What erodes is the hotel’s ability to own the discovery relationship. If the guest’s first meaningful encounter with a property is a sentence inside a Google-generated summary, that guest begins the relationship with Google, not with the hotel or the OTA.

For Booking Holdings and Expedia, this is a threat they are actively managing. Booking Holdings’ 2025 Form 10-K, filed with the SEC in February 2026, disclosed $8.2 billion in total marketing expenditure — up 12.5 percent on the prior year — with performance marketing expenses, “primarily related to the use of online search engines (primarily Google),” constituting a substantial majority of that spend. The OTAs are paying Google to remain visible inside the very interface that is reducing outbound traffic from Google. It is structurally unstable, and both companies know it. Expedia’s CEO noted on the Q4 2025 earnings call that the company is working to ensure its brands “show up prominently in AI searches” — a defensive positioning statement, not a confident one.

What to watch: revenue managers accustomed to measuring Google’s contribution through last-click attribution models will find those models increasingly misleading. A traveler influenced by a Google AI Overview and then completing a booking through a branded website or an OTA may register as a direct booking or an OTA booking with no Google attribution. The channel cost is invisible; the channel influence is not.

3. Google Hotel Ads Inside AI Responses: The Paid Placement Architecture Is Already Live

Google Hotel Ads are not new. What is new is where they appear. As of Q2 2026, Google is integrating Hotel Ads directly into AI Mode responses — the conversational interface powered by Gemini that reached 200 million monthly active users by March 2026, according to Alphabet’s Q1 2026 earnings call. A traveler who opens AI Mode and types a complex itinerary query receives hotel recommendations with live pricing pulled from Google Hotel Ads feeds. Hotels not connected to a certified Google Hotel Ads integration do not appear. Their rates are invisible to the AI.

In May 2026, Search Engine Roundtable confirmed that Google is testing direct hotel booking links inside AI Mode responses — surfacing official hotel website links with live pricing rather than defaulting to OTA aggregators as the sole booking path. This is structurally different from previous Google metasearch behavior. For the first time, a hotel with an active Google Hotel Ads feed and a direct booking engine can appear in an AI-generated recommendation with a link that bypasses the OTA.

The acquisition cost implication is direct: the gap between a 15–25 percent OTA commission on a booking routed through Booking.com, and whatever the effective cost of a Google Hotel Ads placement turns out to be for a booking that originates inside AI Mode, is the commercial question this creates. Google has not disclosed AI Mode booking conversion rates, and the feature remains in early testing. What is measurable today is the cost of non-participation. SiteMinder’s product documentation notes that paid Hotel Ads listings appear in AI Mode with an “Official site” badge and guaranteed placement; free booking links appear in a lower, less visible section with no guaranteed visibility. As AI Mode becomes the primary discovery surface for high-intent travelers, the value difference between these two positions is no longer comparable to a ranking difference on the ten blue links. It is closer to the difference between appearing and not appearing on an OTA’s first page.

Google also announced the expansion of its AI Max advertising platform to travel in April 2026, embedding hotel ad placements inside AI Overviews and AI Mode through intent-based matching rather than keyword targeting. Under AI Max, hotels define messaging and audience parameters; Google’s system decides where and when to display the ad within an AI-generated response. The gain is reach into conversational queries that no keyword campaign would have historically captured. The tradeoff is placement opacity — a hotel’s ad may surface inside an AI summary that the hotel cannot read, review, or verify in advance. Alphabet’s Q1 2026 earnings release confirmed Search and Other revenue grew 19 percent year over year, with management attributing growth explicitly to AI Overviews and AI Mode driving query volumes to an all-time high. The monetization is working for Google. The cost structure for hotels has not yet been fully disclosed.

What to watch: operators who have not verified their Google Hotel Ads feed quality and sync frequency should treat that as the immediate priority. AI Mode draws live rates and availability from those feeds; a feed that lags by more than an hour, or that carries rate discrepancies against OTA listings, produces either invisible inventory or pricing that disadvantages the direct channel inside the same AI response.

4. The Universal Commerce Protocol: Google Removes the Routing Step

At Google I/O on May 19, 2026, Google announced the expansion of its Universal Commerce Protocol to hotel booking. This is the development that reframes the entire distribution question — not as a routing problem, but as a transaction-layer problem.

UCP, announced in January 2026 and live for eligible US retailers since then, is an open-source standard designed to enable AI agents to complete purchases inside a Google interface without requiring the user to click through to an external website. Google’s developer documentation describes the hotel extension explicitly: “UCP is an open standard for agentic commerce that now unifies digital Hotel Booking. It enables direct, instant reservations, reducing friction and abandonment.” The hotel remains merchant of record. Guest data stays with the hotel. The transaction completes inside Google’s AI surface.

Booking Holdings, Expedia, Marriott, IHG, Choice Hotels, and Wyndham were named as integration partners when Google confirmed the hotel booking development in November 2025. Marriott’s CEO Anthony Capuano confirmed in February 2026 that the company is building an integration to process bookings directly through Google AI Mode. Google’s own I/O 2026 blog post described a “weekend planner agent” that pulls hotel availability and real-time pricing and builds a personalized itinerary with direct booking links, with the traveler never leaving Google.

The structural distinction from the current Hotel Ads model is precise and significant. Under Hotel Ads, Google sends a qualified visitor to a booking engine; the transaction happens on the hotel’s site or the OTA’s site. Under UCP, the transaction happens inside Google’s AI response. The click outward is removed. Google is no longer referring traffic to the distribution ecosystem — it is becoming the interface through which distribution happens.

For OTAs, this is a genuinely different threat than the AI routing question. If a traveler can search, compare, and complete a hotel booking without leaving Google — and if that booking routes directly to the hotel’s reservation system, bypassing the OTA — then the OTA’s role in a UCP-enabled transaction is nothing. This explains why Booking Holdings spent $8.2 billion on Google in 2025 while simultaneously investing in AI trip-planning features that attempt to keep travelers inside Booking.com’s own interface: they are hedging against exactly this outcome. The OTA that builds its own conversational booking layer is attempting to be the destination before Google’s destination is fully built.

For hotels, the framing is more nuanced. The preservation of merchant-of-record status and guest data is a meaningful structural difference from the OTA model — it is the thing the industry has spent years trying to reclaim. If UCP delivers on that framing at scale, it represents a genuine channel improvement. The honest uncertainty is what Google charges for the placement over time, what data signals it retains from the transaction, and whether the current cooperative framing holds as the protocol matures and Google’s leverage over the channel increases. The OTA relationship began in structurally similar terms. The evolution of that relationship over twenty years is the relevant reference point.

What to watch: UCP for hotel booking is live for the US first, with expansions to Canada, Australia, and the UK planned. Hotels whose booking engines and central reservation systems are not yet on a Google-certified integration path will be structurally absent from UCP-enabled booking flows — meaning the OTA, which is already integrated, remains the default. The question for GMs and DOSMs is whether their technology stack vendors have confirmed UCP compatibility timelines and what the integration cost looks like. That conversation is worth having before the protocol is live at scale, not after.

5. The Fee Structure Inside AI Has Not Yet Been Published — That Is the Operative Risk

The formal terms under which Google will monetize hotel bookings inside AI Mode and UCP have not been publicly disclosed. This is not an incidental gap; it is the central open question that will determine whether the shift reduces or increases hotels’ total distribution cost.

Google has confirmed it will not act as merchant of record or take a commission in the OTA sense. What it has not committed to is a specific cost model for the placements, integrations, and data feeds that enable a property to appear in AI surfaces with live rates and direct booking links. The precedent from Google Hotel Ads is that free visibility exists but carries no guaranteed placement, while paid placement — structured as cost-per-click — is the mechanism through which properties compete for position. As AI Mode becomes the dominant discovery surface, the value of paid placement within it will not remain at today’s levels. When OTAs and hotel brands bid for position inside AI Mode responses, the auction economics will resemble what happened to paid search CPCs when OTAs entered that market two decades ago. Metasearch advertising costs roughly doubled in per-click terms as OTA bidding escalated — a trajectory worth factoring into acquisition cost modeling before assuming that Google’s AI channel is automatically cheaper than the OTA commission it partially displaces.

The OTAs’ response to this is already visible in their financial disclosures. Booking Holdings is investing in its own AI trip-planning tools, attempting to make Booking.com the destination rather than the fulfilment layer. The company’s Q3 2025 filing noted that the direct channel accounted for a “mid-fifties percentage” of total room nights — their highest disclosed direct share — as they work to reduce Google dependency. Expedia is doing the same. Both companies are trying to own the conversational relationship before Google’s UCP makes that relationship Google’s. The hotel that is paying attention to this dynamic is the hotel that understands it is not a passive observer of an OTA-Google negotiation. It is the asset both sides are competing to route.

The fee structure Google publishes for UCP hotel bookings — when it publishes it — will be the most important distribution number of the next year. The next piece in this series will follow it.

Data Sources

Most read

OUR PUBLICATIONS

Also to discover