Hotel AI Radar: Google still dominates, but AI is already changing search

Artificial intelligence is transforming the way travelers search for information, compare options, and discover hotels or vacation rentals. However, measuring this transformation requires distinguishing between concepts that are often conflated: users, queries, web references, citations, recommendations, and search engine rankings do not all mean the same thing.

In this first edition of EMEXS’s Hotel AI Radar, we analyze the latest available data on search and artificial intelligence and debunk one of the most widespread misconceptions regarding visibility on ChatGPT.

AI hasn't replaced Google; it's being integrated into Google

In July 2026, Google accounted for 91.31% of global search engine referrals recorded by Statcounter. This percentage does not represent the total number of searches conducted, but rather the clicks that lead to pages in the Statcounter network from a search engine. The company compiles its statistics based on more than 3 billion monthly page views across more than one million websites.

At the same time, an independent study that analyzed 55,393 Google searches between March 13 and April 21, 2026, found that AI Overviews provided a response in 13.7% of the searches analyzed. When the search query was phrased as a question, that percentage rose to 64.7%. The sample was limited to trending searches conducted from the United States and distributed across 19 subject categories.

The two sets of data are not directly comparable because they are based on different methodologies and time periods. Taken together, they show a clear trend: artificial intelligence is not suddenly replacing traditional search, but rather changing the user experience within the search engine itself.

Current Landscape: AI-powered responses are making strides in commercially-driven searches

An analysis published by Semrush on July 2, 2026, examined more than 600,000 keywords from its U.S. desktop search database, spanning ten industries and tracked between November 2025 and April 2026.

During that period, the presence of AI Overviews in search results for queries with commercial intent increased by 71%. This does not mean that they appeared in 71% of all commercial queries, but rather that their frequency increased by 71% compared to the beginning of the period analyzed. The study also found that commercial queries displayed AI Overviews more frequently than transactional queries across all sectors in the sample.

This distinction is relevant to the hotel industry. Searches in which users are still comparing neighborhoods, categories, experiences, or types of accommodations are more likely to include a generative response than queries that are closer to an immediate transaction.
 

Hospitality Industry: Intermediaries provide information, but they don't necessarily get the click

An independent study on Google AI Mode ran 4,000 hotel search queries on February 2 and 3, 2026. The study used 40 search queries, eight cities (including Barcelona), five types of search intent, and four simulated geographic locations. In total, it collected 84,329 citations, 6,130 interactive links, and 1,146 unique hotels.

Metasearch engines and review sites accounted for 29.2% of the cited sources, while OTAs contributed another 17.3%. Collectively, intermediaries accounted for 46.6% of the citations. Official hotel websites accounted for 21.2%, Google properties for 19.9%, and editorial media for 10.9%.

However, the distribution changes when we look at the destination of the links associated with the hotels. Of the 6,130 interactive links analyzed, 79.1% first directed users to a Google Business Profile listing, 16.6% linked directly to the hotel’s website, and only 3.6% led to an OTA.

This raises two distinct questions: where does the AI obtain the information it uses to construct its response, and where does it direct the user when the user selects a hotel?

Myth of the month: “My hotel ranks third on ChatGPT”

We’re hearing customers say things like this more and more often: “My hotel ranks third on ChatGPT” or “I asked about the best hotels in Barcelona, and we came in second.”

It’s understandable that we try to apply the same logic we’ve used for years to measure Google rankings to AI assistants. However, ChatGPT and other assistants don’t function like a traditional search engine, nor do they provide a consistent ranking of hotels.

The same question can generate different answers depending on who asks it, how it’s phrased, which country the query comes from, the context of the conversation, or the time of day it’s asked. Even if the same person repeats the exact same question, the hotels mentioned and the order in which they appear may change.

A study on business recommendations ran approximately 6,000 trials with different rephrasings of questions and another 6,000 control trials on models from OpenAI and Anthropic.

When repeating the exact same question, the sets of recommended brands showed a similarity of between 50% and 61%. When using slight rephrasings of the same intent, the match rate dropped to 28.8%. When adding conditions such as country, language, or a more specific profile, it fell to 13.5%.

This study is not limited to the hotel industry, but it demonstrates the methodological problem of turning an isolated response into a permanent ranking.

A screenshot may show that a hotel appeared in a specific search query. It does not prove that it holds a fixed position for all users.

That’s why, instead of asking ourselves, “Where does my hotel rank?”, we should ask ourselves other questions:

• How often does it appear?
• In what types of searches?
• For what types of travelers?
• In which markets or languages?
• What sources does the AI use to recommend it?

Visibility in artificial intelligence should not be measured by a single search, but rather through a broad and recurring set of queries, phrasings, locations, and assistants. In AI, an isolated occurrence may be anecdotal. It is frequency and consistency that begin to provide us with a useful signal.

The EMEXS Data

Price discrepancies of up to 12.6% across four simultaneous channels (and how this affects what the AI reads). 

Discrepancies between a hotel and its suppliers regarding check-in on August 6, 2026

In our ongoing monitoring of price discrepancies, we see that price discrepancies rarely occur in just one channel. A real-world example from this very week: for an official rate of $419, we simultaneously detected four providers (Super.com, Trip.com, Bluepillow.com, and dealbase.com) selling below that price. Super.com had the largest price gap, with a difference of $53 (at $366). 
Why is this relevant when discussing ChatGPT or search engines? Because artificial intelligence systems and metasearch engines constantly crawl the web to provide answers to users. If, when the assistant searches for information about your hotel, it finds that you’re cheaper on four third-party channels than on your own website, not only do you lose control of your distribution, but the traffic (and the “best price” recommendation) will invariably go to the OTA. 
 

EMEXS Reading

When we talk to hotels or vacation rentals about artificial intelligence, one of the first questions that usually comes up is: “Do we have to change our entire SEO strategy right now?”

The answer is no. The data from this edition does not indicate that SEO has ceased to be important or that all the work done so far is no longer useful. Google remains the primary gateway to websites, and its traditional systems continue to influence the information retrieved and displayed by artificial intelligence tools.

What is changing, however, is the way that information reaches the user. Previously, travelers would receive a list of results and decide which ones to visit. Now, more and more often, they receive a pre-formulated answer in which AI selects, combines, and summarizes information from various sources.

And this is where one of the main challenges for hotels comes in: AI doesn't just check the official website. It can also use Google Business Profile listings, OTAs, metasearch engines, reviews, specialized media, destination guides, and other websites where the property is mentioned.

That’s why, when a hotel doesn’t appear in an AI response, it’s not always a problem that can be solved by adding specific words to its website. In many cases, the first step is to check something much more basic: whether the hotel’s information is clear, complete, and consistent across all its channels.

Priorities should focus on four areas:

• Keep the hotel's information complete, up-to-date, and consistent on the official website, Google Business Profile, OTAs, and other channels.
• Create specific content that addresses experiential intentions: location, atmosphere, services, traveler profiles, and reasons for choosing the establishment.
• Work on digital reputation and the quality of reviews, without confusing observational correlations with the effects detected in controlled experiments.
• Measure visibility through multiple queries and executions, avoiding the presentation of a single response as a fixed position.

Visibility in artificial intelligence isn't achieved through a trick or a single action. It is built by enabling systems to find information that is clear, consistent, up-to-date, and supported by various sources.

Ultimately, it's not about starting from scratch, but about doing the essentials better: organizing data, strengthening the hotel's digital presence, and offering content that truly helps travelers make a decision.