Introducing the Boutique Hotel AI Visibility Index

For us as hotel marketers, visibility in these responses has become an increasingly important part of discussions with hotel owners and management teams. Yet there is still very little practical data about how AI platforms select, rank, and cite hotel recommendations, especially in the boutique and independent hotel space.

Saying “we are not 100% sure” or “visibility depends on a variety of factors” may be technically true, but it does not give hotel teams much to work with. One-off screenshots and isolated examples are also difficult to build on.

So we decided to run our own research project: The Boutique Hotel AI Visibility Index.

What We Are Measuring

Our aim is to track hotel visibility, volatility, and discernible trends in AI-generated hotel recommendations.

Which hotels are recommended? Which ones appear repeatedly? Which platforms are more consistent? Which sources are cited? And how much does the answer change when the same question is asked again?

These are the kinds of questions hotel marketers, owners, and management teams will need to understand as AI-assisted hotel discovery becomes more common.

For this first phase, we are running the same set of hotel recommendation prompts across 100 destinations worldwide. The destinations include a mix of cities, islands, resort areas, and well-known boutique travel regions.

Each run includes three prompt categories: boutique, luxury, and romantic. These were chosen because they reflect the type of properties Kollective works with most often: independent boutique hotels.

Is that setup a little self-serving? Yes. But it is also where we have the clearest frame of reference. Initial results suggest the findings are likely to be useful for hotels and marketing teams across much of the hospitality sector.

The study currently covers five AI platforms: ChatGPT, Copilot, Gemini, Google AI Overview, and Google AI Mode. These engines were chosen based on the usage patterns we currently see across our hotel clients’ analytics data.

Taken together, that gives us 1,500 individual recommendation queries per run.

The Study Setup

We will run the same set of queries 15 times over the next 30 days, using the same prompt structure each time. The point is to avoid changing our own inputs and then pretending that any movement came from the platforms.

A single snapshot does not tell us very much. What matters is what happens when the same question is asked again and again.

Studies outside the hospitality vertical already suggest that AI answers can be highly variable across repeated runs. We want to test this within our own sphere of interest.

If a hotel appears once, that is interesting. If it appears consistently, that is more meaningful. If it appears in one platform but not another, that matters too. And if the top recommendation changes within 24 hours, that says something about how fluid AI visibility may be.

The Early Data Is Already Intriguing

Our first runs have already shown a high level of variation.

In Run 1, more than 4,000 distinct hotels were named and close to 3,000 different sources cited. By Run 2 the very next day, over 6,000 distinct hotels appeared, and only around 58% of the hotels overlapped between the two runs. Put another way, roughly 2,800 hotels surfaced on day two that were not named the day before.

Those numbers are early, and we are not treating them as final conclusions. But they do suggest that AI visibility is not a fixed position. It may vary by platform, destination, source ecosystem, query type, and timing.

This is not a perfect study. It is not trying to model every possible traveller profile. A family, a honeymoon couple, a solo traveller, and a price-conscious guest would all have different needs, and AI platforms should reflect that.

For this first phase, we are keeping the prompts broad. We are looking at boutique, luxury, and romantic hotel recommendations because that is the segment we know best, and because it keeps the research focused enough to run repeatedly.

We also cannot see exactly why each platform recommends a particular hotel. What we can measure is what appears: the hotels named, their ranking position, how often they return, and which sources are cited or associated with the response.

That matters because boutique hotels often depend on a more fragmented visibility ecosystem than large global brands. Their presence may be shaped by their own website, reviews, editorial coverage, destination content, PR, social platforms, OTAs, and third-party hotel guides. Knowing more about how these signals appear to be used by AI platforms, and how they may affect visibility, is critical for exactly the kinds of properties we work with daily.

Over the coming weeks, Lynn will be sharing selected observations from each run on LinkedIn. Once the research period is complete, we will analyse the full dataset and publish a detailed report with the final findings, methodology notes, charts, and implications for boutique hotels and hospitality marketers.

We expect the final analysis to be published in early September, in time for the next phase of hotel marketing planning after the summer season.