Ask an AI assistant where it would send your guest

You can probably describe your OTA conversion, your direct-booking share and your website performance in detail. It is worth asking whether you can say the same about what an AI assistant recommends when someone looks for a luxury hotel in your destination.

It takes about ninety seconds. Ask for a five-star or luxury stay in your city or on your stretch of coast, in the segment you actually compete in, and read the answer the way a guest would.

Four things are worth noting. Whether the property appears at all. Which properties do appear, and whether they are the competitive set you would have named. How the property is described if it is there. And which sources the answer seems to draw on, where those are visible.

None of this is research. Outputs vary between assistants, between phrasings and between one day and the next, so a single answer is an anecdote. Absence from one response does not establish that a property is invisible, and presence in one does not establish that it is well represented. What turns the exercise into evidence is running the same prompt, in the same form, at intervals, and keeping what comes back.

The reason to do it anyway is that a recommendation layer now sits between a guest’s intention and the property. It describes the property to people who may never reach the website, and it sits outside the reporting the property already has.

Whether that matters commercially at your property is a separate question, and the answer will not come from one prompt. Knowing what the answer currently says is the cheap part.