Ask ChatGPT About Your Hotel. Now Ask It the Question Your Next Guest Actually Asks.

You have probably already run the first test. Most hotel marketers have, and nearly every owner has asked someone on the team to do it. You opened ChatGPT or Gemini, typed your property's name, and read the answer. It was accurate. It named your restaurants, your spa, your design story, your awards. It was flattering enough to screenshot for the Monday meeting. You concluded that AI knows your hotel, and you moved on.

That test gave you the right answer to the wrong question.

The answer was not the problem. The conclusion was. Asking an AI engine about your hotel by name is how you search. It is not how the guest you have not already reached searches. That guest does not know your name yet. They ask the category question. Best luxury hotel in your destination. Best spa resort. Best honeymoon hotel. That question is where the shortlist is built, before a rate is seen, before a photograph is seen, before your website gets its chance.

So run the second test now. It takes sixty seconds.

Open the same engine and type this, replacing only the bracketed words: "best luxury hotel in [your destination]." Do not type your property name. Read the answer. If your property sells on something more specific, run that instead: "best spa resort in [destination]," "best honeymoon hotel in [destination]," "best golf resort near [destination]," "best wedding venue in [destination]." Look for your name in the answers.

We ran the same test across nine luxury properties in five markets. Every property was tested in live single-run captures on ChatGPT and Gemini. Additional captures were run on Perplexity, Grok, Copilot, and Google AI Overview where relevant to the audit.

The name test succeeded every time. Nine properties out of nine. Asked directly, the engines returned accurate, detailed, current profiles. The design pedigree, the Forbes ratings, the signature restaurants, the award-winning golf course, the private cottage model, the full-floor suites. Not one property had a recognition problem.

The category test failed almost every time in that same nine-property set. A property holding a Forbes Five-Star rating for both its hotel and its restaurant was absent from the best-luxury query on all six platforms tested. A resort whose every suite is built as a private spa did not appear on the spa weekend query for its region. A property with the largest event footprint in its market was invisible on every event and wedding query while a competitor with a fraction of its space was recommended by name. A branded residence was named by one engine under a heading of what not to buy.

Known when named. Absent or buried on the discovery queries these properties would commercially want to own. Nine properties is a small set, and every result above is a live capture, not a projection. Run your own and you will have a tenth.

The gap follows a retrieval pattern. For a category question, the engine is not starting with your brand narrative. It builds an answer from the sources it retrieves for that category. When the question contains your name, it retrieves sources about you: your website, your brand pages, your Forbes listing, your press. Those sources are accurate, so the answer is accurate. When the question is a category, it retrieves category sources: third-party roundups, aggregator lists, review-site rankings, travel-guide listicles. If those lists do not contain you, neither does the answer.

Cloudbeds' 2025 study of 810 prompts across ChatGPT, Perplexity, and Gemini found that online travel agencies account for 55.3 percent of AI-generated hotel citations, and hotel websites for 13.6 percent. That study is not about these nine properties. It corroborates the mechanism: category answers rely heavily on third-party sources. The evidence that shapes these queries is narrative and comparative: editorial roundups, list pages, review summaries, destination guides, category rankings. A property can be fully distributed on every OTA, highly rated, and accurately described when named, and still be absent from the lists the category answer is assembled from.

The answer moves with the source.

This is why the name test feels so reassuring and means so little. It measures the query where the guest already knows you exist. The category query is shaped by sources most properties have never inventoried, let alone managed.

The commercial point is short. Your own record is winning the query where the guest already knows you exist and losing the query where the initial shortlist is formed. Name searches convert past guests, referred guests, and guests your sales and advisor channels already touched, and that traffic matters. The category search is where the incremental guest, the one no channel has touched yet, decides which names deserve consideration. Those are the travelers your property is invisible to.

If your property came back in the category answer, run the next category, and the next, until you find the one you are missing. If it was absent, the issue is not whether AI knows your hotel. It does. The issue is whether the sources that define your category know your hotel.

Correcting that starts with the sources, not the AI engine. It means identifying the third-party sources that shape the category answer in your market, then making sure those sources carry an accurate, current, category-relevant version of the property. The discipline Americas Great Resorts operates to do this is Knowledge Formation Optimization.

The findings referenced here come from AI visibility audits of nine luxury properties in five markets conducted by Americas Great Resorts in June and July 2026, using live query testing with each prompt run once in a fresh session and the returned answer recorded. Each capture records the platform, date, prompt, and returned answer, and whether the audited property appeared. Nothing was modeled or projected. The dated captures are retained by Americas Great Resorts.