Swiss hospitality faces a new invisible ranking: the one set by AI

A single Swiss hotel brand can be cited up to seven times more often on one artificial intelligence engine than on another, within a seventy-two hour observation window. La Réserve Genève records six mentions on ChatGPT against forty-two on Gemini. Hotel Bristol Genève cumulates zero mentions on ChatGPT and thirty-four on Perplexity. This dispersion across engines stands out as the most notable observation of the AI Visibility Index conducted by Repliq between 17 and 19 May 2026.

Beyond classical search engine optimisation criteria and the dynamics specific to online travel agencies, Swiss hospitality faces the emergence of AI answer engines. This paradigm shift moves the point of contact: the internet user no longer explores a list of links, but consults a recommendation written in synthetic form.

To assess the impact of this transition for the Swiss brand, an analysis was carried out across 1,248 measurements in Geneva, Lucerne, Zermatt and Interlaken. The data indicates a strong correlation between the structure of official websites, the weight of hotel consortiums and the frequency of citation by the algorithms.

Summary of the first findings of the 2026 AI Visibility Index. The full measurement pipeline can also be explored visually on Repliq's interactive observatory (open access).

A protocol based on the analysis of real scenarios

To establish this index, Repliq queries each AI engine the way a real traveller would, through its public interface and not through a technical API. This method gives a faithful view of responses as they appear to end users, rather than a programmatic version that may differ, be sanitised or cached. The analysis covers four destinations in Swiss tourism (Geneva, Lucerne, Zermatt and Interlaken) and addresses the main models on the market: ChatGPT (OpenAI), Google AI Mode, Gemini (Google) and Perplexity AI.

The sample covers seven stay categories matching the planning intents observed in travellers: luxury, wellness, family, budget, business, ski and lake. Each query was issued in the four languages relevant to the Swiss market (French, English, German, Italian), at several moments, on the four engines.

A few example queries:

• Best 5-star hotel in Lucerne?
• Spa hotel in Interlaken for a quiet weekend?
• Family hotel in Zermatt with a swimming pool?
• Ski-in ski-out hotel in Zermatt?
• Lakefront hotel in Lucerne?

For each response, the study records the properties cited, their position in the response and the sources mobilised by the engine to support its recommendation.

First findings: three dynamics shape visibility

Three main observations emerge from the analysis.

First, a mix of sources mobilised by answer engines. The hotel's official site, booking platforms, specialised guides, Wikipedia entries and hotel consortiums act together. The relative weight of each channel varies by query and destination.

Next, the impact of editorial specialisation. Properties positioned clearly on a given use (wellness, family, ski-in ski-out, conferences) obtain a higher mention rate on their niche segment, sometimes exceeding established palaces. A dedicated product page and a coherent semantic description weigh more than overall renown.

Finally, a notable linguistic asymmetry. The same property can be cited heavily in one language and remain almost invisible in another, depending on the sources the engine consults. Multilingual coverage of the official site and of third-party content sits among the decisive levers.

A significant visibility gap between engines

The most notable observation of the study concerns the scale of inter-engine variance for a single brand. Across the top twenty properties identified, the gap between the most generous engine and the most sparing one exceeds a factor of two in most cases, and reaches a factor of seven to eleven for certain brands.

In Geneva, La Réserve Genève records six mentions on ChatGPT against forty-two on Gemini, a factor of seven. In Interlaken, Salzano Hotel Spa Restaurant is never cited by ChatGPT but appears nineteen times on Gemini. Hotel Bristol Genève cumulates zero mentions on ChatGPT and thirty-four on Perplexity. In Zermatt, The Omnia totals two citations on Google AI Mode and twenty-one on Perplexity, a factor of ten.

This dispersion calls into question any aggregate reading of AI visibility. For marketing teams, measuring overall exposure without breaking it down by engine amounts to obscuring most of the information. ChatGPT also captures between 59 and 83 percent of the consumer AI assistant share by various measures, which makes the visibility specific to that engine particularly sensitive for brands.

Destination-by-destination mapping

Geneva: a fragmented market, a tight top 5

Geneva presents the most fragmented market of the study. No single property stands alone. The main palaces share the recommendations, and a budget-segment actor (Geneva Hostel) emerges in the top 8.

Lucerne: Schweizerhof in a leading position

Lucerne is the inverse case: a single property structures the market. Schweizerhof Luzern is the most-cited property of the entire study, across all destinations. Its presence is observed across the majority of stay intents tested.

Zermatt: concentration at the top, a surprise on the ski segment

Zermatt appears more polarised. Mont Cervin Palace leads the overall ranking, followed by a trio of Riffelalp, Zermatterhof and Omnia. More unexpected: the "ski-in ski-out hotel" query brings up a non-palace boutique property (Matterhorn Focus Design), which suggests that answer engines strongly segment their recommendations by stay intent.

Interlaken: specialisation prevails over star rating

Interlaken presents the most concentrated market of the study. Victoria-Jungfrau Grand & Spa structures the recommendations on luxury and adventure. On the wellness segment, however, a specialised 4-star property (Salzano Hotel Spa Restaurant) positions itself above the local palace. This observation suggests that algorithmic selection favours semantic coherence over prestige.

Where these recommendations come from

Beyond the properties cited, the study documents the sources that engines rely on to formulate their recommendations. This mapping is probably the most operational contribution of the analysis for marketing and communication teams in the sector.

Two indicators are measured for each source. The coverage rate indicates the percentage of AI responses in which the source appears at least once. The citation rate indicates the average number of citations per response for that source.

Across the full set of identifiable citations (6,220 citations in the corpus, excluding Knowledge Graph envelopes and opaque redirections), a structural distribution emerges. Hotels' official websites account for about 35 percent of measured citations, OTA booking platforms for 27 percent, and listicle-type editorial content for 6 percent. A brand that invests primarily in its owned surfaces and in the quality of its OTA listings reaches nearly two thirds of the AI citation volume in cumulative terms. The reflex of chasing editorial placements exclusively captures a more modest portion than expected. This breakdown sits among the most operational contributions of the study for marketing directions.

Several observations emerge from this breakdown.

The properties' official websites remain the leading source mobilised, present in more than six responses out of ten. The structure, semantic richness and readability of the official site weigh heavily in the formulation of the recommendation. The booking platforms (Booking, Tripadvisor) act as the second source.

The specialised guides (Michelin Guide, The Hotel Guru, Tablet Hotels) and Wikipedia position themselves in the upper half of the ranking, at a comparable level. An entry in the Michelin Guide or a structured Wikipedia page weighs more than an isolated article in the generalist business press.

The hotel consortiums (Swiss Deluxe Hotels, Leading Hotels of the World) constitute a specific visibility channel, mobilised more regularly by the engines than traditional business press outlets.

The Swiss business press (Le Temps, htr.ch, NZZ, 24 heures, Hotellerie Gastronomie Zeitung) appears marginally in the sample of hospitality responses. Combined, the four best-known generalist titles total fewer than five citations across the full corpus. The articles exist and are read, but they are not the principal source that answer engines mobilise to formulate a recommendation to a traveller. Conversely, specialised thematic blogs and the Wikipedia encyclopedia capture most of the editorial share in the measured AI recommendations.

The weight of time in editorial citation

The analysis of publication dates of cited editorial content highlights a rarely commented phenomenon. Among the press, specialised guide, thematic blog and Wikipedia content for which a publication date could be extracted, the median age stands at nine years. The observed window covers an amplitude of twenty-three years, from the oldest (a Wikipedia article on Lucerne dated 2003) to the most recent.

In concrete terms, a Wikipedia entry written in 2015 on Mandarin Oriental Palace Luzern generates twenty-one citations across the May 2026 measurement period. A 2017 article published on The Hotel Guru on the hotels of Lake Lucerne remains cited today. The persistence of editorial content over time is a structural characteristic of AI visibility, distinct from the audience-peak dynamic associated with the short cycle of current affairs.

This observation carries a direct implication for Swiss business newsrooms. An article well positioned on a property, a destination or a stay category retains a capacity for citation across a five to ten year window. The editorial investment applied to structured and durable content is measured no longer at publication, but in cumulative citation capacity over several years.

Outlook for the 2026 summer season

Understanding the architecture of these new visibility channels becomes a performance issue for hotel chain managers, communication directions and tourism promotion bodies. As the summer season approaches, mastery of this algorithmic share of voice represents a complementary acquisition lever that is still under-exploited.

The full report, enriched with detailed rankings by property category, comparative analyses across engines and technical recommendations by destination, will be published in June 2026.

What the full report documents

On the basis of the 1,246 retained measurements, the in-depth analysis highlights several observations that reconfigure the reading of AI visibility for the Swiss hospitality sector:

• A single brand can be cited up to seven times more often on one engine than on another, within the same seventy-two hour window. The inter-engine variances documented in the report invite marketing teams to move beyond an aggregate reading of AI visibility.
• Out of the 3,539 properties detected at least once, only twenty-six are the subject of a consensus across the four engines. The position of "AI leader" is statistically narrow.
• Nearly two thirds of the citations mobilised by the engines refer to surfaces controlled by the brand itself (official site: 35%) or by its distribution (OTA: 27%). The share of independent editorial content is more modest than expected.
• The median lifespan of an editorial citation measured in the corpus stands at nine years. An article published in 2015 retains an AI citation capacity equivalent to that of a recent article. The report quantifies this longevity by publication.
• The sustainability and business events segments do not appear in any measured AI response. The absence of established editorial authority for these themes constitutes a territory to invest.
• The interactive observatory complements the report. The /en/observatoire/hotellerie page, in open access, visually renders the entire measurement pipeline (stay categories, destinations, engines, source types, hotels cited) in the form of an explorable Sankey diagram. Four complementary views detail the sources: top 20 cited domains, distribution by type, breakdown by engine, amplifiers by destination.

Receive the AI Visibility Index — Hospitality 2026

Register by email to receive the full report on publication, during June 2026. Detailed rankings by stay category, comparative analyses across engines and technical recommendations by destination.

M'inscrire

En vous inscrivant, vous acceptez de recevoir le rapport par e-mail et, occasionnellement, des informations sur les éditions suivantes de l'Index de Visibilité IA Repliq. Vous pouvez vous désabonner à tout moment.

For organisations wishing to measure the AI visibility of their property now, without waiting for the report's publication, Repliq offers an AI visibility audit in 11 days, from CHF 990, or continuous monitoring on the platform from CHF 89/month.

---

Study conducted by Repliq between 17 and 19 May 2026. 1,248 measurements, 4 AI engines, 4 Swiss destinations, 4 languages. Detailed methodology and full data available on request for press and research.