“Today’s helpful partner can become tomorrow’s dominant gatekeeper.”
Americas Great Resorts wrote that on March 15, 1998. About Online Travel Agencies.
We named the merchant model. We warned that OTAs would control pricing, presentation, and the customer data generated from every transaction. We said hotels risked ceding ownership of their guests’ profiles, preferences, and booking histories to intermediaries. We said that once lost, that relationship becomes difficult to reclaim.
The industry called it progress.
The original 1998 article is available here: Online Travel Agencies: A Short-Term Boon or Long-Term Threat to Hotel Independence?
Twenty-eight years later, OTA commission dependency is structural. For most independent luxury hotels, the guest discovery relationship belongs to Expedia and Booking.com. The data belongs to them. The repeat traveler is often re-entered through them. The hotel provides the room.
That is not a distribution problem. It is the outcome of ignoring a warning.
In 2026, the same pattern is running.
AI platforms are not neutral information conduits. They are positioning themselves as the primary interface between travelers and hotels. ChatGPT, Gemini, Copilot, Grok, and Perplexity synthesize an answer. The hotel that contributed the raw material for that answer may receive no attribution, no traffic, no guest identity, and no relationship.
OTAs controlled booking. AI-mediated systems are increasingly present in discovery. What began as discounted distribution became structural commission dependency once OTA platforms concentrated demand. AGR sees a comparable intermediary-risk pattern in AI-mediated discovery, but this is a strategic analogy, not evidence that the underlying platform mechanism or future monetization path is identical. As AGR wrote in 1998, intermediaries who control a customer gateway can acquire structural leverage over suppliers.
This is no longer only our observation. In March 2026, Boston Consulting Group published “AI-First Hotels,” predicting that the OTA commission model will evolve into AI-era distribution fees charged for prominence and relevance in algorithmic recommendations. The commission survives. Only the landlord changes.
The industry is calling it an opportunity.
The question in 1998 was whether hotels would lead the online revolution or become subject to it.
Most chose subject to it. Not through ignorance. Through the perfectly rational decision to fill rooms now rather than build infrastructure for later. The OTA commission felt manageable in 1998. By 2010 it was structural. By 2026 it is embedded for most.
The question in 2026 is whether hotels will build owned, machine-readable identity before AI systems harden their canonical answers about which properties are worth recommending.
We published the warning in 1998 because the pattern was visible then.
We are publishing this in 2026 for the same reason.
KFO is the structural response: build the hotel’s machine-readable identity before someone else defines it.
In 2034, the ones who didn’t will call it unavoidable, and will be submitting applications to a new AI platform that won’t be built until 2032, asking to be considered.
