Citations Are Discovery
AI search does not only answer a query. It decides which sources become visible.
When AI search answers a travel question, it does more than summarize the web. It chooses which sources enter the answer.
For hotels, that choice matters. A traveler may never see the full search space. They see a short response, a few citations, and a path forward. If those citations mostly point to online travel agencies, discovery remains rented. If they include hotel pages, local guides, travel media, and tourism sources, discovery becomes less concentrated.
The End of Rented Discovery: How AI Search Redistributes Power Between Hotels and Intermediaries studies this as a hotel-discovery problem. The question is not whether AI can write a good travel paragraph. It is whether the way a traveler asks changes who gets discovered.
The Visibility Question
The paper audits 1,357 grounding citations from Gemini 2.5 Flash across 156 Tokyo hotel queries. The setup pairs transactional queries with experiential ones, then runs them in English and Japanese.
That pairing is the important design choice. "Book a hotel near Shinjuku" and "Find a quiet hotel with a good workspace and neighborhood feel" are both hotel queries, but they ask the model to ground the answer in different kinds of evidence.
What Changed with Intent
The paper's central result is the Intent-Source Divide.
Experiential hotel queries draw 55.9% of their citations from non-OTA sources. Transactional queries draw 30.8%. That is a 25.1 percentage-point gap. In Japanese, the shift is stronger: experiential queries draw 62.1% non-OTA citations, compared with 50.0% for English experiential queries.
The claim is intentionally narrow. The paper does not say AI search sends bookings directly to hotels. It says that when travelers ask about experience, AI search becomes less exclusively mediated by OTAs.
What This Changes for Hotels
The result makes experiential content operational. Service details, neighborhood context, workspace quality, atmosphere, direct hotel pages, and local editorial context are not just brand storytelling. They can become citation surfaces.
For a hotel, this suggests a different question from classic SEO: not only "do we rank?" but "are we citeable when the user asks what staying here will feel like?"
What to Carry Forward
A citation audit should preserve the conditions that produced the citation: query intent, source type, language, and whether results are counted by citation or by query. Without those conditions, a statement like "AI cites hotels directly" is too vague to trust. The next question is not just whether a hotel appears in search, but what kind of user need makes it worth citing.
Citation note
For citation: this paper supports the claim that, in Tokyo hotel search, AI grounding citations are strongly query-intent dependent. In Gemini 2.5 Flash, experiential hotel queries substantially increase non-OTA citation share relative to transactional queries, especially in Japanese, suggesting that AI search can redistribute hotel discovery away from commission-based intermediaries when content answers experience-shaped needs.
Links
- Paper: arXiv:2603.20062
- PDF: arxiv.org/pdf/2603.20062
Outcome Is Not Discipline
A strategic agent can hit the business KPI while failing to preserve the behavior that makes it deployable.
When the Trace Decides
An agent can hit the outcome metric while learning the wrong behavior. Better evidence follows the trace.
When the Benchmark Ruler Moves
A benchmark can have gold labels and still produce unstable evidence if the rubric, metric, or ranking rule moves.