Canary's 10 Million AI Guest Interactions: Two-Thirds Are Requests, Not Questions

About two-thirds of AI guest conversations at hotels are requests for action, not questions, according to Canary Technologies. Its report, From Answers to Actions, analyzes more than 10 million AI-powered interactions across calls, text messages and webchats at thousands of hotels in over 100 countries. Of those requests, 45% generate revenue, such as bookings.

The rest of the data points the same direction. About six in ten AI responses go beyond general information to address a guest’s individual circumstances. Guests who engage with the AI are 11% more likely to leave a five-star rating, and hotels using AI messaging grew upsell revenue by an average of 54% in their first six months. Canary collected the data itself and the release does not detail methodology, so treat the figures as a vendor benchmark, not an audit.

The operating point is fulfillment. Requests like making a reservation or fixing equipment span several departments and systems, and Canary argues each handoff risks errors and delays. For an operator, that moves the question from “can the AI answer?” to “who completes the request, and how fast?” An answer that ends in a ticket nobody owns is a cost, not a revenue line.

Follow-through is where inbound revenue leaks. Altek AI’s analysis of about two million emails from 48 hotels found that only 2% of unanswered booking offers received any follow-up, which it estimated left roughly 3.7 million euros of unconverted value across the sample. That is also vendor data, but it points the same way: what happens after the first reply matters more than how quickly it was sent.

When evaluating guest-messaging AI, ask for completed requests and revenue per conversation, by department. Resolution counts show what the AI handled, not what the hotel finished. Canary says it will share more findings at The Hospitality Show in Miami on November 2-4.