In a recent audit, an upscale hotel had everything generally recommended to improve its visibility in AI: clear positioning, rich content, consistent external sources and clearly identified attributes.

Yet in the recommendation scenarios tested, it almost never appeared.

The problem, then, was not that AI systems did not understand the hotel.

They understood it. They simply did not think of it when it came time to choose.

This is an important distinction.

Being understood is not the same as being considered

When we asked the engines directly about the property, its identity came through clearly: a natural setting, serenity, privacy, experiences for couples and wellness.

External sources largely confirmed these attributes.

So this was no longer a problem of understanding.

The breakdown occurred further along: between the point at which an AI system knows the property and the point at which it spontaneously thinks of it as one of the possible options.

This is where “creating more content” reaches its limits

When AI visibility is low, the most common response is to enrich the website.

Create more pages. Add FAQs. Elaborate on the experiences. Repeat certain attributes.

This can be useful when information is missing.

But when the website and external sources already tell the same story, producing another variation of the same content does not necessarily solve the problem.

The question then becomes:

what needs to be strengthened for the property to become a credible answer when it is time to choose?

This is where GEO goes beyond content production alone.

Look at those being chosen

In the case we studied, the property occupied a highly distinctive positioning space.

That was a strength.

But some competitors benefited from a broader information footprint. They were associated with more guest segments, occasions for a stay and experiences.

They therefore offered more pathways for AI systems to reach them.

This does not mean diluting your positioning in an attempt to meet every request.

It does, however, require asking two questions:

in which travel situations do we want AI systems to think of us spontaneously?

And above all:

in which areas do we genuinely have what it takes to become a reference point?

This is where competitive analysis becomes useful: not to copy businesses already being recommended, but to understand why they make the shortlist where we do not.

Evidence also makes the difference

Another finding: some competitors benefited from a more recent and more diverse ecosystem of documentation.

More independent sources. More recent content. Experiences covered from different angles. Mutually reinforcing evidence.

So it is not just about the volume of information available.

It is also about the vitality of the evidence graph.

Positioning can be very well established and yet gradually lose salience if it is no longer sufficiently confirmed by credible, diverse and recent sources.

At this stage, producing an additional page on your own website does not necessarily add much value.

Sometimes the real challenge is to ensure that the company's claims are better corroborated elsewhere.

Start again with the experience

For every important travel situation, we ultimately need to return to a fairly simple chain:

situation → reason to choose → actual experience → distinctive characteristic → customer evidence → external evidence.

Take a hotel that wants to be recommended for a romantic stay.

It is not just about writing that it is romantic.

You need to be able to explain why. What specific experience makes the stay different? What do guests remember about it? Which external sources confirm it? And compared with the other available options, why does this experience genuinely deserve to be chosen?

It is this chain that gives the recommendation credibility.

Above all, it helps determine where to act:

• should you create content that is genuinely missing;
• provide stronger evidence;
• build your presence in certain external sources;
• document an experience more effectively;
• make better use of the voice of the customer;
• or fix a technical problem.

Content then becomes a means again.

Not an automatic response.

Key takeaways

The case we studied demonstrates one simple point:

a business can be well understood without receiving enough consideration.

And until it becomes part of the set of options being considered, it cannot be recommended.

This is why simply measuring visibility or producing more content is no longer enough.

The real challenge now is to understand where the breakdown occurs:

Understood → Considered → Recommended

Then act precisely at that point.

In our view, this is one of the most important changes that GEO is demanding of tourism businesses today.