In Brief: Terence Ronson warns that AI is reshaping hotel distribution by reducing discovery to a shortlist drawn largely from OTA and editorial content. He argues that independent hotels risk becoming invisible unless they maintain accurate, consistent and machine-readable property data across the sources AI trusts.
-
The AI shortlist is the team sheet — if a hotel isn’t named, it never takes the pitch. – Image Credit Pertlink
TL;DR
- AI has turned hotel discovery into a shortlist. A traveler no longer opens thirty tabs — they ask once, and an AI names two or three properties.
- Only 6% of hotels appear consistently in AI-generated recommendations, and 82% of what AI does recommend is sourced from OTAs and editorial media — not from hotels themselves.
- Marriott’s CEO has told investors AI’s biggest near-term hospitality impact is distribution, not productivity. Marriott and Hilton have already shipped conversational AI search tools; Accor and Hyatt are still testing.
- More than half the world’s hotel rooms remain unbranded and independent — the properties with the least AI-ready data, and the most to lose from being left off the pitch.
- Levelling the field isn’t about outspending the majors. It’s about becoming machine-readable: one owner, one source, one structured answer, per fact, per property.
- Guests carry risk too: a German court has held Google liable for its own AI Overviews’ false statements, and a consumer investigation found AI summaries glossing over real hotel safety complaints.
Kickoff: Discovery Has Become a Shortlist
Hospitality has lived with an enormous look-to-book ratio for years — industry estimates put hotel bookings at roughly 1,000 searches per completed booking, with a widely cited Phocuswright study finding that travelers visit dozens of sites before choosing a property. That scattergun shopping was inefficient, but it was forgiving. A hotel with modest marketing could still be found on page four of a search, inside a metasearch comparison, or three clicks into an OTA list.
AI collapses that. Ask an AI engine to find a hotel near Marina Bay with a pool, good food, and late checkout, and it doesn’t return thirty links — it names two or three properties and stops. Marriott’s own CEO has told investors this is exactly where AI will hit hospitality hardest: not back-office productivity, but distribution — whether a hotel is ever shown to the traveler at all.
That is the moment captured in the graphic opposite. The AI shortlist is the team sheet for the match. If a hotel’s name isn’t on it, its beautifully converting website, its five-star reviews, and its perfectly tuned rate never come into play. No click. No impression. No abandoned booking to retarget. From the AI’s perspective — and increasingly from the guest’s perspective, too — that hotel was never in the game.
Same Match, Different Stadiums
The contest is not being fought on level ground. Hilton launched its AI Planner in March 2026, and Marriott followed in June with its Ask Bonvoy conversational search, now spanning more than 10,000 properties and a loyalty base of nearly 283 million Bonvoy members. Accor and Hyatt have tested comparable tools but not yet launched them publicly. Marriott, for its part, is also feeding extensive portfolio data directly to Google and OpenAI, so its hotels appear accurately — and prominently — inside AI-powered search and shopping surfaces.
An independent resort in Bohol, Palawan, Bali or Phuket has none of this: no direct line into a frontier-model lab, no hundred-million-member loyalty database to train on. By Hilton’s own framing when it launched a brand aimed squarely at this segment, these properties sit within the more than half of the world’s hotel rooms that remain unbranded. That is not a niche problem. It is the majority of global hotel supply, playing the same AI-discovery match as Marriott on a pitch the majors helped design.
The scoreline so far bears that out: only 6% of hotels appear consistently in AI-generated recommendations, and where AI does recommend a hotel, 82% of the underlying information comes from OTAs and editorial sources rather than the hotel’s own data. Independents that assumed a good website and a Booking.com listing were “enough” digital presence are discovering that AI, unlike a human shopper, does not go looking for what isn’t already stated clearly somewhere it already trusts.
The OTA Half of the Pitch
OTAs are not spectators in this match — they supply most of the data AI is reading. That cuts two ways for independents. Booking.com, Expedia and Agoda listings matter more than ever as a source AI models trust, but independents are once again dependent on an intermediary for visibility, much as in the OTA-only era — except the toll is now effectively paid twice. First, a commission of roughly 15–30% (typically higher for independents than for chains with negotiating leverage) simply to be listed. Second, the accuracy of whatever that intermediary says about the property, since that is what the AI is most likely to repeat.
Google’s move to build agentic booking directly into AI Mode drives the point home. Its early integration partners for flights and hotels are reported to include Booking.com, Expedia, Marriott, IHG and Choice Hotels — a list built from scale, not merit. An independent’s fastest route into that conversation isn’t waiting for a direct invitation; it’s making sure the OTA and editorial sources that an AI already trusts say the same accurate thing about the property that the hotel would say itself.
Channel dependency is also becoming algorithm dependency, and the point travels well beyond one region. This month, changes tied to the EU’s Digital Markets Act stripped live pricing, date filters, and several descriptive tags from Google’s direct-property hotel search results in Europe, while comparison services kept a fuller presence. The platform didn’t go offline — it simply stopped showing hotels the way it used to. Nobody in this match owns the pitch.
When the Assistant Gets the Scoreline Wrong
Guests aren’t passive beneficiaries of any of this. Booking.com’s own research finds 89% of consumers want to use AI in future travel planning, yet Expedia’s 2026 research describes an “AI Trust Gap”: nearly 70% of travelers still prefer to complete the actual booking through a trusted brand rather than an AI chatbot or agent. Guests want the shortlist. They are far less sure they want the machine holding the credit card.
That caution looks reasonable given two recent cases. The Regional Court of Munich has ruled that Google’s AI Overviews constitute Google’s own statements, not neutral search results — making the platform potentially liable when its AI invents or distorts facts about a business, hotels included. Separately, consumer group Which? found that Tripadvisor’s AI-generated review summaries described several hotels in strongly positive terms, while the individual reviews underneath detailed food poisoning, harassment complaints, and hygiene failures that the summary never mentioned. Neither case targets a hotel directly — both involve the platform’s own AI output — but the hotel’s name and reputation are actually on the line when the summary is wrong.
For independents, this cuts an unexpected way: it’s also an opening. A guest who no longer trusts the shortlist to be complete still wants a verifiable answer underneath it. An independent hotel with clean, consistent, structured facts wherever an AI looks becomes the accurate exception in a landscape of confident-sounding summaries — and accuracy is a competitive advantage the biggest chain in the world cannot simply outspend.
The Pertlink Playbook: Leveling the Pitch
|
AUDIENCE
|
THE PERTLINK ACTION
|
|
Marketers
|
Put “AI Share of Recommendation” on the commercial dashboard next to SEO and OTA performance. Test the same handful of guest-intent prompts every month and track mention frequency, position and accuracy.
|
|
C-Level & Owners
|
Fund an Authoritative Hotel Knowledge Layer before funding another chatbot: one owner, one source, one machine-readable answer for every material fact about the property.
|
|
OTAs & Distribution Partners
|
Treat listing accuracy as an AI-safety issue, not a merchandising one. Every stale room description or wrong occupancy rule is now training data for someone else’s assistant.
|
|
Tourism Boards
|
Build the shared destination infrastructure — transport, experiences, payments — that lets any authorized AI transact locally. Competition happens at the interface, not the data layer beneath it.
|
|
Investors
|
Score portfolio companies on machine-readability, not AI-adoption headlines. A hotel or platform not structured for AI discovery today is a distribution problem, not a technology one, waiting to surface.
|
Final Whistle
The industry has spent two years asking whether AI is intelligent enough to help travelers choose a hotel. That question is close to settled. The one that decides who is still selling rooms in three years is different: whose data is trustworthy, structured and consistent enough for that intelligence to find, cite and recommend. The chains can win that on scale. Independents can still win it on discipline. Neither wins it by ignoring the pitch they’re already playing on.
Terence Ronson
Terence Ronson is the Founder and Managing Director of Pertlink Limited, a boutique hospitality technology and AI advisory consultancy headquartered in Hong Kong since 2000, operating across the Philippines and Asia-Pacific. He chairs the AI Education & Training Subcommittee of the HFTP AI Collective and is an inductee of both the HFTP International Hospitality Technology Hall of Fame and the CHTA Hall of Fame.
