We recently audited a hotel that, on paper, seemed to have everything it needed to feature prominently in AI recommendations.

Clear positioning.
Sources talking about it.
Guest reviews.
A properly built website.

Yet in certain search scenarios closely aligned with its offering, it rarely appeared, if at all.

That is where the analysis became interesting.

Looking at the sources used to describe the property, we found a fairly strong promise, fully consistent with its positioning.

On paper, it could clearly provide a compelling reason to recommend the hotel.

But when we looked at what guests actually described, that promise appeared far less often.

No glaring contradiction.
No major satisfaction issue.

Simply put, few guests spontaneously confirmed that experience.

And for an AI system choosing between several hotels, that kind of gap can matter.

AI does not just look at what you say about yourself

This is probably one of the most interesting aspects of GEO in tourism.

There is a lot of discussion about content, citations, structured data and presence in the right sources.

All of that matters.

But when a traveller asks an AI system to recommend a hotel, it is no longer just trying to understand which properties exist.

It has to choose.

And to choose, it can cross-reference several signals: the hotel's website, platforms, media, guides, reviews and the experiences travellers describe.

The real question then becomes quite simple:

“Do enough pieces of evidence tell the same story?”

If the hotel makes a claim, an external source confirms it and guests describe the same experience, the promise gains credibility.

If the promise exists mainly in marketing messaging or in a single isolated source, it can become less robust.

This is where GEO becomes much more interesting

Because the issue very quickly goes beyond website optimisation.

The question is no longer just:

“How do I make AI aware of my hotel?”

But rather:

“What reasons does it have to recommend me over someone else?”

And those reasons are not all found on a web page.

They are found in the experience actually delivered, in the details of the product, in reviews, in the sources discussing the property and in the consistency of it all.

This is also why we believe that GEO in tourism will sometimes end up reaching back to the product itself.

Describing an experience better can be useful.

But if that experience is not strong enough, distinctive enough or sufficiently lived to be described naturally by guests, the problem lies elsewhere.

Guest reviews then take on a different role

Traditionally, we look at them to track satisfaction, reputation or conversion.

We probably also need to start viewing them as raw material for AI recommendations.

What do guests actually describe?

What comes up spontaneously?

Which moments during a stay seem memorable enough to mention?

Which brand promises are confirmed time and again?

And which, by contrast, appear far less often than we imagined?

Sometimes, that gap is the most instructive part.

Because it does not just show us how to communicate better.

It can also reveal where the experience needs strengthening.

The main lesson this case taught us

Being visible does not necessarily mean being worth recommending.

And repeating a promise more often does not automatically make it more credible.

For tourism businesses, that is rather good news.

It means the battle for AI recommendations will not be determined solely by who publishes the most content or has the best command of a few search optimisation techniques.

It will also depend on something much more tangible:

“The ability to create an experience clear, strong and consistent enough for guests, external sources and the brand to end up telling the same story.”

That is probably where the real work begins.