The first time hotels gave away the relationship
When I launched BookDirect in 1997 — one of the first online hotel reservation platforms — the pitch to hotel operators was straightforward: build your own direct booking infrastructure now, before someone else owns that channel.
We integrated with the Amadeus Global Distribution System, giving properties real-time inventory management and direct booking capability at a time when most hotels were still taking reservations by fax. We signed hundreds of properties, including Delta Hotels and Westmont Hospitality. The technology worked. The strategic argument was sound.
But here’s what I watched happen in parallel: for every hotel operator who invested in direct digital infrastructure, ten others decided it was easier to let Expedia handle it. The commissions were manageable. The technology was someone else’s problem. The guests kept coming.
What those operators didn’t understand — what the whole industry mostly failed to understand until it was too late — is that distribution infrastructure isn’t just a cost centre. It’s where the customer relationship lives.
OTAs didn’t beat hotels by being better at hospitality. They beat hotels by being better at data. They knew which guests were price-sensitive and which were loyal. They knew which markets were undersupplied on Thursday nights. They knew, before any individual hotel did, what the demand curve looked like across an entire destination.
Hotels had the rooms. OTAs had the intelligence. And intelligence, as it turns out, compounds.
Hotels lost direct customer relationships not because OTAs were better at hospitality, but because hotels underinvested in the infrastructure required to own that relationship themselves. The commission model felt efficient until the day it didn’t — and by then, switching costs were enormous.
What’s actually happening with AI right now
Here’s the thing about the OTA disruption that most post-mortems miss: OTAs captured travellers at the search stage. The guest had already decided to travel, had a rough destination in mind, and was comparison-shopping. That’s where Expedia won — by being better at search and better at presenting options.
AI disrupts one layer earlier than that.
When someone asks ChatGPT, Perplexity, or a voice assistant “where should I stay in Toronto for a two-night business trip?” they haven’t opened a browser yet. They haven’t visited an OTA. They’re at the aspiration stage — the moment when travel intent is first forming and destination or property preference hasn’t solidified.
AI disrupts hotel distribution at the aspiration stage, before brand preference forms.
The AI synthesizes a recommendation from everything it knows: structured data about the property, published reviews and content, reputation signals from authoritative sources, and entity records from databases like Wikidata and Google’s Knowledge Graph. It produces an answer. The traveller, more often than not, follows it.
If your hotel isn’t legible to that AI system — if it doesn’t appear in the data sources the model draws from, if there’s no authoritative content describing what makes you the right choice, if your entity record is thin or inconsistent — you are invisible at the exact moment the decision is being made.
Not invisible on page two of Google. Invisible before the traveller ever opens a browser.
Why this is a governance problem, not a marketing problem
Here’s where I want to speak directly to boards.
The instinct at most hotel companies, when confronted with a technology disruption, is to hand it to the marketing team. Social media strategy? Marketing. OTA optimization? Revenue management, maybe marketing. AI visibility? Probably marketing.
This instinct is wrong, and it’s the same instinct that made the OTA problem worse.
The OTA problem wasn’t a marketing failure. It was a governance failure — a failure of boards and ownership groups to ask, early enough and persistently enough, what the long-term strategic cost of ceding the customer relationship was going to be. Marketing optimized within the system. Boards should have questioned the system itself.
The AI visibility problem has the same structure. It is not fundamentally about content calendars or SEO tactics. It is about data ownership, entity infrastructure, and the question of whether your hotel is building the kind of authoritative digital presence that AI systems treat as a credible source — or whether you’re outsourcing that credibility to OTAs and review platforms that will benefit from your guests’ loyalty instead of you.
Three questions every hotel board should be asking management today:
First: What is our AI legibility score? Have you tested what major AI assistants say when asked to recommend a hotel in your primary market? Does your property appear? If it does, what sources does the AI cite — your own website and content, or an OTA profile? The answer tells you who owns your reputation in the AI layer.
Second: Do we own our entity record? A hotel’s entity record is the aggregated set of structured data — across Google’s Knowledge Graph, Wikidata, your own website’s schema markup, and third-party data sources — that AI systems consult when building a recommendation. Most hotel companies have never audited this. Many would find it thin, inconsistent, or dominated by OTA data.
Third: Are we publishing content that AI systems can cite? AI recommendation engines don’t invent answers from nothing. They synthesize from indexed, authoritative sources. A hotel that publishes nothing — no articles, no structured FAQs, no content that directly answers the questions travellers ask — gives AI systems nothing to cite on its behalf. That vacuum gets filled by whoever is publishing: OTAs, review aggregators, travel bloggers.
The compounding advantage of moving now
One thing the OTA era taught operators who were paying attention: first-mover advantage in distribution infrastructure is real and it compounds.
The hotels that invested in direct booking infrastructure in 1998 and 1999 had lower OTA dependency ratios for the next twenty years. The hotels that waited until 2005 or 2010, when the problem was undeniable, paid much higher switching costs — both financially and in terms of guest relationship reconstruction.
The AI era is earlier than that. Most hotel companies haven’t started. Most hotel boards haven’t asked the question.
That means the window for a meaningful early-mover advantage is still open. Legibility to AI is an operational discipline — it involves structured data implementation, consistent entity management, and a content strategy oriented around the questions travellers ask AI systems rather than the keywords they type into search engines. None of this requires enormous capital expenditure. It requires deliberate attention and board-level prioritization.
The hotelier who told me in 1997 that the internet was just a fad wasn’t foolish. He was working from the best available evidence at the time. The evidence available now is different. The pattern is visible. The window is open.
Hotels lost the OTA war because they didn’t see the distribution layer shift until it had already happened. The AI layer shift is happening in plain sight. The question is whether boards will treat it as a governance priority before it becomes a crisis — or after.
Download the free 12-point AI Visibility Checklist for Hotel Operators →
