58% of Travelers Check AI Hotel Recommendations Against the Property's Own Site — Most Sites Fail the Check

Only 17-20% of travelers say they trust AI-generated hotel recommendations outright — yet 58% visit the property’s own website afterward to verify the claim anyway, and 62-63% do that verification check regardless of how much they trust the AI output in the first place. That’s the core finding in a September 11 Hospitality Net opinion piece arguing hotel websites have quietly shifted role: no longer the primary discovery platform, but the verification stop travelers make before booking. The stakes are higher than the trust gap alone suggests, because most sites aren’t winning that visit either: citing Tharro research covering 695 searches, the piece finds under 10% of AI citations reference hotel websites directly, with most instead drawing on OTA and third-party content — meaning the direct site rarely gets first crack at the guest’s attention, only the follow-up verification.

The friction points are specific and fixable. OTA listings typically show all-in pricing — taxes and fees included — immediately, while hotel sites often bury the same total behind multiple clicks and form submissions; cancellation terms and fee breakdowns get buried rather than surfaced; and hotels routinely fail to proactively correct AI-assistant answers that may already be wrong by the time a guest lands on the site. The recommended fixes are tactical: show all-in pricing on the first screen, name fees in plain language rather than at checkout, display actual room types at actual prices, and swap automated chat widgets for property staff who can verify details in real time.

The pattern matches an economics story worth pairing with it: GuestCentric’s Joao Trindade found 56% of engaged hotel-website visitors show high booking intent, yet 98% still abandon before completing a purchase — largely over the same unresolved questions about fees and policies. Different framing, same conversion leak: AI search didn’t create the trust-and-friction gap between a ready-to-book guest and a completed reservation, it just made the gap more visible.