There’s a structural irony at the heart of AI booking that the hospitality industry hasn’t fully reckoned with yet. Every wave of digital disintermediation in travel has been read, in the moment, as a threat to brand. Hotels panicked about Expedia in 1996 for the same reason they’re anxious about AI booking today: when an algorithm sits between you and the guest, does the brand still matter? The answer from the OTA era is instructive, and it runs exactly counter to what most operators assume.
I’ve had an unusual perch on this question. In the 1990s, I founded BookDirect, one of the first online hotel booking platforms, at exactly the moment the first wave of digital disintermediation was reshaping the hospitality industry. I watched from the inside as Expedia launched, Booking.com followed, and an entire generation of hotel operators was forced to reckon with what they were actually selling and to whom.
Then I spent the last five years as Chair of the Travel Industry Council of Ontario, the consumer protection regulator for Ontario’s travel industry, watching what happens when trust breaks down at scale in travel: what it costs, who bears it, and how hard it is to rebuild.
The pattern I’m seeing now is not new. The technology is different and the speed is considerably faster, but the strategic error hotels are about to repeat is identical to the one most of them made in the late 1990s. The operators who avoid it will be the ones who genuinely understood what disintermediation takes from you.
The first time this happened, the industry drew the wrong lesson
When Expedia launched in 1996 as a Microsoft subsidiary and Booking.com expanded aggressively through Europe in the early 2000s, hotels faced a genuine distribution crisis. Suddenly, there was a platform that could reach millions of potential guests more efficiently than any hotel’s own website, marketing budget, or sales team. The pressure to participate was enormous, and largely irresistible.
Hotels responded by listing on every platform they could find. They competed on rate. They accepted commission structures that by the mid-2010s averaged 15 to 25 percent of every booking on the major OTAs. The industry rationalized this as the cost of distribution, a necessary tax on visibility.
What most operators missed in that calculation was what they were actually surrendering. It wasn’t just margin. It was the guest relationship. When a traveller books through Expedia, their primary loyalty, if they have any, is to Expedia. The hotel is a vendor. The confirmation email comes from the platform. The loyalty points accrue on the platform. The review gets posted on the platform. The hotel becomes, in the guest’s mental model, a fulfillment center for a transaction that happened somewhere else.
The hotels that survived the OTA era best were not the ones who optimized their OTA presence most aggressively. They were the ones with strong enough brand identities that guests sought them out despite the platform layer sitting between them. The Four Seasons didn’t lose pricing power because guests valued the Four Seasons experience independent of where they found it. The boutique hotel with a loyal local following didn’t lose its regulars because those guests had a direct relationship with the property that no OTA could replicate or replace.
When the transaction gets intermediated, brand trust is the only thing that still belongs to you. Everything else is rented from the platform.
Now it’s happening again. This time faster, and with higher stakes.
AI booking agents — whether integrated into Google’s ecosystem via Gemini, available through ChatGPT’s Operator function, or built by purpose-designed travel AI companies — are beginning to change the mechanics of how hotel reservations get made. The shift is not yet mainstream, but its trajectory is clear: within the next three to five years, a growing share of hotel bookings will be initiated not by a human searching a platform, but by an AI agent acting on a human’s behalf.
Most operators read this backwards.
The intuitive reaction is that AI will make brand trust less important. When an algorithm is choosing the hotel, human affinity for a brand name becomes irrelevant — or so the thinking goes. Why would an AI care whether a hotel has a strong reputation if it can just optimize on price, availability, and review scores?
Here’s why that logic fails: brand reputation is the closest signal an AI has to substitute for the experiential judgment a seasoned traveller would bring.
When a person books a hotel, they bring a complex set of judgments: personal experience, intuition, social proof, pattern recognition across dozens of previous stays. An AI can’t replicate that directly. What it can do is use accumulated brand signals — reviews, ratings, consistency of experience, recognition — as a stand-in for the kind of reliable, low-risk recommendation a knowledgeable human would make.
A hotel whose name consistently signals reliable, consistent, no-surprises experience will be algorithmically favoured. Not because the AI is loyal, but because brand reputation is the most efficient available signal of quality when the AI is operating without direct experiential knowledge. The more autonomous the booking agent, the more it relies on reputation as a filter.
A hotel with no distinctive brand signal competes purely on price and availability. In an AI-mediated world, that is a commodity position. Commodities race to the bottom.
What this means for Expedia, Booking.com, and the platforms built to replace them
The OTAs built their dominance on a single insight: they could own the discovery layer between guest intent and hotel room more effectively than any individual hotel could. For two decades, that insight held. Guests went to Expedia first, not the hotel website.
AI booking disrupts that dynamic structurally. When a guest stops going to Expedia and starts asking an AI agent to handle the booking, the OTA loses its front door. It becomes, at best, a backend inventory source — a supplier to the AI rather than a destination for the traveller. That is a fundamentally weaker position, and the major OTAs are well aware of it. Their aggressive moves into AI-enhanced search and recommendation are, in part, a defensive response to this structural threat.
New AI-native booking platforms are racing to occupy the discovery layer before the OTAs can adapt. The central question they’re competing to answer: when an AI agent books a hotel on a guest’s behalf, does it route through OTA inventory — continuing to pay 15–25% commissions to Expedia and Booking.com — or does it go direct to the hotel?
For hotel operators, this is where the calculus genuinely changes. If an AI books direct with your property, you recover the commission margin you’ve been surrendering for two decades. You also recover something more valuable: the relationship starts with you. The Book Direct campaigns hotels have been running since the early 2010s — with limited success because the platforms were simply too convenient — might finally have the infrastructure to actually work.
The significant catch is that an AI only books direct with your hotel if it trusts your hotel enough to recommend it without the OTA’s review layer, guarantee infrastructure, and consumer protection framework underneath. The guest isn’t making that judgment anymore. Their AI agent is. And that agent is making its recommendation based on accumulated brand signal: your reputation, your consistency, and what your name actually conveys.
The hotels that built genuine brand trust over the past twenty years didn’t just survive the OTA era. They’re the ones whose names an AI will surface at the top of its recommendation set and book direct.
What operators should actually be doing right now
The current conversation in hospitality about AI is dominated by operational applications — revenue management optimization, guest communication automation, housekeeping scheduling. That work is real and it matters. But it’s not the strategic conversation.
The strategic conversation is about the guest relationship. Specifically: what is your brand’s standing with an AI agent that has never stayed at your property, has no personal loyalty, and is working entirely from what your accumulated record says about you?
Ask yourself: if a well-informed human travel consultant were recommending hotels to a client with your target guest profile, would your property be on the list? Not because of your OTA ranking, but because of your actual reputation. What your reviews say. What your brand name implies. Whether your story is specific and recognizable or generic and forgettable.
Practically, this means: stop optimizing for the platforms — start optimizing for what the platforms measure. Your Google Business Profile should be complete, current, and actively managed, because AI treats it as a primary source of truth about your property. Your website needs Hotel schema markup, not just good copy. Your review responses need to be intentional rather than templated — they’re indexed. And your post-stay email sequence needs to exist at all, and it needs more than one touch. None of this is expensive. Most of it takes an afternoon. All of it compounds over years in ways that can’t be manufactured at the moment an AI is deciding.
The operators who survive this wave will be the same kind as the ones who survived the last one. Not the biggest, not the most technically sophisticated. The ones who understood, even in the OTA era, that the booking was never the point. The relationship was the point.
For everyone else, the window to build the brand signal that will matter in an AI-mediated booking environment is narrowing. The time to start — or to be honest about whether you’ve actually been doing it — is now.
It was never the room. It was always the trust.
Download the free 12-point AI Visibility Checklist for Hotel Operators →
