Travel and Hospitality: Getting Recommended by AI

Three out of four French travelers have never asked an AI engine to help plan a trip, but those who do use it from the very first step: picking a destination, then a place to stay. The available studies on what ChatGPT or Gemini answer next agree on one point: large hotel chains and booking platforms dominate these answers by far, much more than the average guest review score. For an independent hotel, a tourist office or a local travel agency, the question is no longer whether this matters, but how to show up in an answer that, by default, names the biggest players first.

Here is what the available studies measure about how travelers already use AI, who AI engines actually recommend between chains, booking platforms and the property's own site, the real weight of guest reviews, and what changes when France's public tourism operator builds its own AI recommendation layer.

How many travelers already ask AI for advice?

A minority, but one growing fast among younger travelers, and already decisive at the earliest stages of a trip. According to a survey run by Evaneos with YouGov among 1,689 people interviewed online between March 17 and 25, 2026, 76% of French people have never used AI to organize a trip (Evaneos, press release, April 23, 2026). In other words, roughly a quarter of French travelers have already done so, a minority share but far from negligible for a mainstream use case that only emerged a few years ago.

Among those who do use AI to travel, Evaneos found usage concentrated early in the trip: general information about a destination (50%), inspiration (45%) and planning activities (45%). The published findings do not break out how many users directly ask for a hotel or accommodation name: looking for lodging logically comes after the destination and activities are already set, once a conversational AI has already helped build part of the trip with the user. Another figure from the same study: 71% of users rate the experience as satisfactory, which helps explain why usage keeps climbing despite the majority's caution.

What the Evaneos study does not measure, and what other surveys approach with different methods (how "usage" is defined, panel, timing), is the detail of what gets asked once the destination is chosen. That is exactly the stage where a hotel's, an agency's or a tourist office's visibility is decided: once the traveler knows where they are going, the AI still has to point them to the right property.

Who do AI engines actually recommend: chains, booking platforms, or the property itself?

Large chains and booking platforms first, independent hotels' own sites far behind. At its Luminate conference, hotel software vendor Lighthouse presented a study covering 4,545 prompts put to ChatGPT across nine destinations and five traveler personas, and counted 49,707 named hotel mentions spread across only 2,721 distinct properties (Hospitality Net, 2026). In other words, a handful of properties capture most of the citations, and Paris is one of the nine destinations covered.

Two findings from this study matter directly for an independent hotel. First, the property's category outweighs its reputation: according to coverage of the presentation, ChatGPT recommends luxury-category hotels 83% of the time, leaving most three-star properties unmentioned (Hospitality Today, 2026). Second, large chains generally outperform independents, except in Paris, where the study found an exception. The method does not let us explain that exception with certainty, but it suggests that the density and international reputation of Paris's independent hotel scene is strong enough to resist the brand effect of chains, something that is not guaranteed in a less well-known destination.

Still according to that same presentation, 82% of AI-generated hotel recommendations reportedly come from editorial content and booking platforms rather than from hotels' own sites (Hospitality Net, opinion, 2026). It is the same mechanism we describe for another sector where third-party comparison sites outweigh the seller's own site in how ChatGPT chooses its sources: when third parties (trade press, platforms) produce content that is richer, more recent and more comparative than a hotel's own static homepage, AI logically leans on them instead.

These figures come from a commercial study presented by a vendor that itself sells AI-visibility tools to hotels: treat them as a directional signal on a given sample (ChatGPT only, nine destinations), not as an exhaustive measurement valid across every engine and every city.

Do guest reviews really matter for being recommended?

Less than their popular reputation suggests, according to the same Lighthouse study: the correlation between a hotel's Booking.com score and its visibility in ChatGPT's answers is weak (Hospitality Net, opinion, 2026). A highly rated property is not automatically cited more often, and a middling score does not disqualify a hotel that is frequently mentioned in press coverage or rankings.

That does not mean reviews are useless, only that they likely matter less as a score to optimize than as a source of content the AI can read and cite. An analysis of the lodging and restaurant sector notes that it is above all reviews published on Google, Booking, Tripadvisor and The Fork that feed AI answers (CustPlace, lodging and restaurant sector guide): what counts is the presence of recent, detailed content spread across several platforms, not just one averaged number shown on a single one of them. It is the same principle we cover for other sectors in guest reviews and AI visibility: a dated, detailed review that answers a concrete question (access, noise, breakfast, accessibility) gives the AI citable material that a bare score cannot provide.

For a property that depends heavily on local queries ("hotel near the train station," "inn with a lake view"), this echoes what we detail in local search and AI: Gemini draws directly on Google Maps and Google Business Profile, while ChatGPT and Perplexity rely more on indexed third-party content. A property that leaves its Google listing incomplete, or never replies to reviews, denies every AI engine part of the material it could otherwise cite.

MarIAnne: when the public tourism operator builds its own AI assistant

While hotels wonder about their visibility in ChatGPT or Gemini, France's official tourism site is building its own AI recommendation layer. Since France.fr was redesigned in 2024, Atout France has offered MarIAnne, a conversational assistant that generates personalized itineraries for visitors wanting to discover France, built on a database of more than 65,000 points of interest across the country (Atout France, France destination presentation document, May 2025; French government, official announcement).

For a lodging business or a local tourist office, this tool is a concrete reminder: being listed in a database of points of interest, with a complete, up-to-date and structured entry, is no longer only a matter of visibility on Google, but a condition for appearing in the itineraries this kind of assistant proposes directly to visitors. Atout France has also opened training sessions for tourism professionals for the second half of 2026 on structuring an AI project and getting hands-on with tools like ChatGPT, Gemini or Claude (Atout France, training calendar, second half of 2026), a sign that the organization now treats this as an industry-wide issue, not a passing curiosity.

The parallel with other sectors we have covered is direct: just as an electricity supplier benefits from linking to the public comparison tool rather than ignoring it (see energy suppliers and AI visibility), a French lodging business benefits from making sure its listing is accurate and current everywhere an AI, public or private, might look for it.

What should you do in practice?

For a hotel, a bed and breakfast, a campsite or a tourist office, visibility in AI answers depends less on the property's size, which you cannot change, than on the quality and freshness of what is published about it, on your own site and elsewhere.

1. Structure your property's data. Use schema.org's Hotel or LodgingBusiness types on your site to describe name, address, amenities and lodging type in a machine-readable way. Our practical schema.org guide covers implementation, criterion by criterion.
2. Keep your Google Business Profile current. Hours, recent photos, a precise description of access and surroundings: this is the source Gemini draws on first for local queries (local search and AI).
3. Reply to reviews across several platforms, not just the one you consider the priority. Google, Booking, Tripadvisor and The Fork are the sources most cited in the available sector analyses; a detailed reply to a review (about noise, access, an amenity) creates dated, specific content that AI can reuse.
4. Publish content the platforms do not cover. Real walking distances to tourist sites, seasonal advice, accessibility, a precise cancellation policy: this is what a generic OTA listing most often lacks, and what can make an AI prefer your own page when it needs a precise answer.
5. Check your presence in official tourism databases (tourist offices, Atout France, regional databases), which feed tools like MarIAnne in addition to mainstream search engines.
6. Regularly measure what AI engines say about your property against your direct competitors: our guide to measuring AI visibility compares the available methods.

For a full measurement that compares your property or destination against your direct competitors on the questions travelers actually ask, Beeleven's AI visibility audit asks ChatGPT, Gemini, Claude and Perplexity these precise queries and measures who is cited, in which position, and against which competing properties. It is the same logic of dated, verifiable proof we cover across sectors in how AI chooses the products and properties it recommends.

Frequently asked questions

Do French travelers really use AI to choose their hotel?

Not yet massively for lodging specifically: according to Evaneos and YouGov, 76% of French people have never used AI to organize a trip, and those who do use it mainly for choosing a destination and activities. But the share is growing, and the lodging step logically follows the destination step.

Does an independent hotel stand a chance against large chains?

Yes, but it is harder: available studies show chains and booking platforms dominate citations, except for local exceptions such as Paris in the Lighthouse study. A property's category (luxury, mid-range) weighs more than its reputation alone.

Should you chase a higher Booking.com score to be better recommended by AI?

Not directly: the Lighthouse study found only a weak correlation between a hotel's average score and its visibility in ChatGPT. Reviews matter mainly as a source of detailed, recent content AI can cite, not as a score to maximize.

What is MarIAnne, the AI assistant on France's official tourism site?

It is a conversational trip planner built into France.fr since the site's 2024 redesign, generating personalized itineraries from a database of more than 65,000 points of interest. For a property, being well listed in this kind of official database is becoming a visibility issue in its own right, alongside mainstream search engines.

Sources

• Evaneos and YouGov, survey on AI and trip planning, press release, April 23, 2026 (fieldwork March 17-25, 2026, 1,689 respondents, in French).
• Hospitality Net, "The AI recommendation is the new battleground for hotel distribution," 2026 (coverage of the Lighthouse study, Luminate conference).
• Hospitality Net, "AI can find your hotel, it just won't recommend it," 2026.
• Hospitality Today, "In AI search, hotel brand beats guest reviews," 2026.
• CustPlace, "Hébergements et restaurants : leurs nouvelles stratégies face aux recommandations générées par l'IA" (in French).
• French government (info.gouv.fr), "MarIAnne, des voyages sur-mesure en France avec l'IA" (in French).
• Atout France, France destination presentation document, May 2025 (in French).
• Atout France, training calendar, second half of 2026 (in French).

→Local search and AI: how to appear in proximity answers

→Automotive: how AI helps people choose a car

→Guest reviews and testimonials: their weight in AI visibility