What Three AIs Understood About How Travelers Discover Hotels

A hotel can use AI to forecast demand, answer guest messages, and run its operation more efficiently. A traveler can use AI to decide where to stay and never hear the hotel’s name.

Gemini, Google AI, and ChatGPT all understood that distinction when we asked what they thought of Hotels Are Solving the Wrong AI Problem. Gemini focused on which travelers a hotel suits. Google AI followed what happens when the hotel misses the shortlist. ChatGPT asked how a hotel would measure the commercial value of getting on it.

Series · AI Reviews of AGR Articles
Original article · Hotels Are Solving the Wrong AI Problem
Capture record · One supplied response per system, reproduced below. Google AI’s citation numbers are retained as plain text. Google redirect URLs are omitted. ChatGPT’s source labels and run-together text are reproduced as supplied.

Exact prompt

Tell me your thoughts on this article: https://www.americasgreatresorts.net/hotel-ai-discovery/

AGR’s reading of the responses

Asking AI about a hotel by name shows what the system can retrieve about a property already under consideration. The original article asks executives to try a different question. Describe the trip without naming the hotel. See whether the system recommends the property, whether it describes it accurately, and where the links lead.

The article’s OTA analogy concerns how early hotels recognize the influence of a new interface. OTAs brought measurable bookings before the longer-term effects on distribution and customer relationships became clear. An AI assistant can influence which hotels the traveler considers before the traveler chooses where to book.

Gemini. Gemini understood why the purpose of the trip matters. Its examples of family trips and romantic getaways illustrate how one property can suit different traveler needs. It warned that an unclear public record could leave a hotel omitted or placed in the wrong competitive tier. That was Gemini’s prediction. The article asks how systems classify a hotel and describes how answers vary across questions and platforms. Gemini also brought in standard schema markup and AI visibility audits. Neither appears in the original article.

Google AI. Google AI understood why omission is hard to see in a hotel’s analytics. A hotel can measure a referral that reaches its site. An answer that omits the property leaves no comparable referral record. Google AI described that as losing the customer. The observable loss is a place on that shortlist. The traveler’s eventual choice remains unknown.

Google AI said most AI interfaces default to OTAs or metasearch for pricing, availability, and booking links. It provided no evidence with that claim. AGR’s link study found Google Travel links in 270 of 296 assessed Gemini answers (91.2%) and Google viewer links in 235 of 290 assessed Google AI Mode answers (81.0%). The Google AI Mode answers also included links to hotel websites. These figures establish how often particular links appeared on those surfaces. The study did not measure inventory feeds, clicks, or bookings, or establish a default across most AI interfaces. AI Hotel Recommendations: Who Gets the Link?

Gemini and Google AI both picked up the article’s commercial context. Gemini questioned the emphasis on a specialized framework. Google AI read the piece as a funnel for AGR’s services. The article explicitly connects the discovery problem to Knowledge Formation Optimization and the question of who receives the demand to owned demand infrastructure. Both readings noticed that connection and drew different conclusions about how it shaped the piece.

ChatGPT. ChatGPT gave the article 8.5 out of 10 and turned several of its diagnostic questions into a five-part measurement table. The source already asks which competitors appear, which traveler needs bring up the property, and whether the hotel is accurately differentiated. ChatGPT grouped those questions alongside recommendation frequency and recommendation quality, and proposed repeating the measurements across systems and over time.

ChatGPT and Google AI both wanted more attention to independent sources. The article explicitly includes independent coverage, reviews, comparisons, and authoritative references among the evidence AI systems can draw on. Google AI raised a narrower question about how heavily systems rely on particular third parties. The article gives no source weights. ChatGPT added a practical example of what happens when competing properties have stronger independent coverage.

ChatGPT used a March 2026 Phocuswright finding to reinforce the case for paying attention. Phocuswright reported that 56% of active U.S. travelers used AI for planning, booking, or in-destination assistance for at least one trip in the preceding 12 months. ChatGPT acknowledged that this covers a broader set of behaviors than the trip-research figure in the original article. The two figures cannot establish a change in the rate of AI use for trip research. Phocuswright research summary

ChatGPT asked what it would take to connect discovery to bookings. The article gives the hotel a place to start. Run the unbranded question and record three things. Did the hotel make the shortlist? Did the answer describe it accurately? Did the link give the traveler a direct path to the property? Those are three results a hotel can check before it opens the booking report.

Gemini Response

The article from Americas Great Resorts hits on a critical blind spot in how hospitality brands approach artificial intelligence today.

Key Takeaways & Strengths

• Operational AI vs. AI Discovery: The core premise draws a sharp, much-needed line between internal operational efficiency (using AI for chatbots, revenue management, or automated concierge services) and external AI discoverability (how large language models like ChatGPT, Gemini, and Claude actually find, perceive, and recommend a property to travelers).
• The “Consideration Set” Shift: Traditional digital marketing has focused heavily on SEO and capturing traffic after a traveler decides where to look. The article underscores that AI models build upstream “consideration sets”—meaning a hotel can be entirely filtered out of a user’s trip planning before a traditional search ever happens.
• Beyond Basic Listing Data: It rightly points out that having accurate names, addresses, and standard schema markup isn’t enough. AI engines synthesize brand positioning, traveler audience alignment (e.g., whether a resort is recommended for families vs. romantic getaways), and competitive sets based on the broader web ecosystem.

Critical Perspective

While the diagnostic perspective is sharp—identifying that hotels are often optimizing backend operations while remaining invisible or misrepresented in AI outputs—it leans heavily into framing this as a specialized, proprietary problem requiring dedicated frameworks (like their “Knowledge Formation Optimization” or AI visibility audits).

For independent hoteliers, the takeaway shouldn’t necessarily be panic over a new technical acronym, but rather the realization that LLMs are the new travel gatekeepers. If a hotel’s digital footprint across third-party authorities, press, and structured data doesn’t clearly articulate who the property is best for, generative search engines will either ignore it or misclassify its competitive tier.

Google AI Response

This article presents a sharp, well-reasoned strategic argument about a genuine blind spot in the hospitality industry, though it is ultimately written as a thought-leadership piece designed to sell the agency’s proprietary framework. [1, 2]

Here is a breakdown of what the article gets right, where it shows its commercial bias, and the key takeaways:

What the Article Gets Right (Strengths)

• The Operational vs. Discovery Divide: The core thesis is spot-on. Hotels love investing in internal AI (revenue management, chatbots, forecasting) because the ROI is easy to measure on a quarterly spreadsheet. Meanwhile, they are largely ignoring how generative AI is changing top-of-funnel consumer behavior. [1, 2]
• Recognition vs. Recommendation: Pointing out the flaw in how executives test AI (“Tell me about our hotel” vs. “Where should I stay in Maui?”) is a great practical insight. Being in an LLM’s training data is completely different from being surfaced as a top recommendation for an unbranded, intent-based prompt. [1, 2]
• The “Invisible Loss” Problem: The comparison to the late-1990s rise of Online Travel Agencies (OTAs) is effective. In traditional search, hotels can track lost clicks and cart abandonment. In AI discovery, if a property is left off a 3-hotel shortlist, there is no analytics trail—the hotel never even knows it lost the customer. [1, 2]
• Critique of Vanity Metrics: The author rightly pushes back against single “AI visibility scores,” noting that LLM outputs are probabilistic, vary across platforms (ChatGPT vs. Gemini vs. Perplexity), and depend heavily on specific user intent.

Caveats & Context (What to Keep in Mind)

• It’s Content Marketing for an Agency: Americas Great Resorts is a luxury hospitality marketing agency, and this article is a funnel for their “Knowledge Formation Optimization (KFO)” and “Owned Demand” services (visible right in the site navigation). While the diagnosis of the problem is accurate, the framing is designed to create urgency so hotel executives hire them to fix it. [1]
• Overlooking the “How” of AI Citations: The article emphasizes publishing specific facts and building “external corroboration”, which is true, but it glosses over how heavily AI models currently rely on third-party aggregators (like TripAdvisor, Reddit, Condé Nast Traveler, and major OTAs) to form those recommendations. A hotel can optimize its own website, but influencing the broader web consensus is much harder than the article implies. [1]
• The Current Reality of Booking Paths: Even when an AI recommends a hotel directly, most AI interfaces today still default to pulling pricing, availability, and booking links from OTAs or metasearch engines rather than direct hotel booking engines—a technical hurdle that a hotel’s own content strategy can’t easily override on its own.

Overall Verdict

It is a persuasive and timely strategic read for hoteliers and marketers. Even accounting for the agency pitch at the end, the warning is valid: treating AI purely as a back-office efficiency tool while ignoring how it reshapes consumer discovery risks repeating the exact same distribution mistakes hotels made with Expedia and Booking.com twenty-five years ago.

Would you like to dive deeper into any part of this, such as practical ways to optimize content for AI discovery (GEO/AEO) or how this applies to a specific industry?

ChatGPT Response

My overall take: I think this is a very good article, and it identifies a significant strategic shift in the hotel industry that many hotel owners and executives are not paying enough attention to.

More importantly, I think the article is making a distinction that deserves considerably more attention:

There is a difference between using AI to operate a hotel more efficiently and making sure AI recommends your hotel to prospective guests.

Those are two very different business opportunities.

I agree with the central argument, but I also think there are some important implications that the article could have explored more deeply.

1. The article gets the biggest issue right: discovery is changing

The most compelling point is the distinction between branded and unbranded searches.

Consider these two questions:

• “Tell me about The Breakers Palm Beach.”
• “What is the best luxury beachfront resort in South Florida for a couple looking for a relaxing getaway?”

The first question assumes the traveler already knows the hotel.

The second asks AI to help decide which hotel deserves consideration in the first place.

That second interaction is where the real competitive battle is developing.

A hotel can have an excellent website, outstanding reviews, a recognizable brand, and strong traditional SEO, yet still be overlooked when someone asks ChatGPT, Gemini, or another AI platform for a recommendation.

The article correctly identifies that recognition is not the same as recommendation.

Americas Great Resorts

2. The OTA comparison is particularly insightful

I like the comparison with the emergence of online travel agencies.

When Expedia, Booking.com, and other OTAs emerged, hotels initially saw them as additional distribution channels.

The immediate benefits were obvious:

• More exposure.
• Additional reservations.
• Access to new customers.
• Better opportunities to fill unsold inventory.

But over time, hotels discovered that they had surrendered considerable control over customer acquisition and distribution economics.

The article suggests AI discovery could create a similar strategic challenge.

I think that is a reasonable analogy, although AI and OTAs are not identical.

There is an important difference: an OTA controls a booking marketplace, whereas an AI assistant may influence the traveler’s decision before a booking marketplace is even selected.

That potentially makes AI discovery an even earlier and more consequential point of influence.

3. I think the article understates the importance of independent information

This is where I would take the argument a step further.

AI recommendations aren’t simply determined by how well a hotel describes itself on its own website.

AI systems draw on a broader information ecosystem, including:

• Independent travel publications.
• Hotel and destination reviews.
• Travel blogs and editorial coverage.
• Reputable travel industry websites.
• Destination and tourism organizations.
• Consistent descriptions of amenities, location, experiences and services.
• Comparisons with competing properties.

This creates an interesting challenge for hotels.

A hotel might describe itself as the premier luxury resort in a destination. But if independent sources consistently describe its competitors as having better beaches, superior service, more distinctive accommodations or better dining, those competing properties may have stronger supporting evidence.

AI visibility is not just a content-writing exercise. It is an authority, credibility and differentiation exercise.

I consider this one of the most important commercial implications of the article.

4. The article is right about measurement, but the industry needs better measurements

I particularly agree with its criticism of simplistic AI visibility scores.

Asking ChatGPT whether it knows a hotel is not a meaningful competitive analysis.

I would want hotel management to measure at least five things:

Measurement	What it tells management
Recommendation frequency	How often the hotel is recommended
Competitive share	Which competing hotels are recommended instead
Search intent coverage	Which types of travelers and occasions generate recommendations
Brand accuracy	Whether AI accurately describes the property’s strengths
Recommendation quality	Whether the hotel is recommended for the right reasons

I would also measure consistency across AI platforms.

For example, a hotel might be recommended frequently by ChatGPT but rarely by Gemini or Perplexity.

That matters.

And measurements should be repeated over time, across different traveler profiles, destinations, price points and trip purposes.

The article correctly recognizes that a single visibility score cannot capture all of this.

Americas Great Resorts

5. Where I think the article could go further

There are three important areas I would have liked to see addressed.

A. AI recommendations must translate into actual bookings.

Getting recommended is valuable, but it doesn’t automatically generate revenue.

Hotels need to understand the entire journey:

1. The traveler asks AI for a recommendation.
2. The hotel enters the consideration set.
3. The traveler evaluates the property.
4. The traveler visits the hotel’s website.
5. The traveler books directly or through an intermediary.

The commercial objective should be to increase qualified demand and direct bookings, not simply AI mentions.

B. Reputation management becomes even more important.

Hotels have traditionally managed reviews partly to influence prospective guests.

Now those reviews also become part of the broader information environment that AI systems may use to understand and compare properties.

Consistent guest feedback, credible reviews and accurate property information become increasingly important.

C. The competitive implications for independent hotels could be enormous.

Large hotel chains have established brands, enormous marketing budgets and extensive digital footprints.

Independent luxury hotels often have something equally valuable: distinctive experiences, personality, exceptional service and unique positioning.

But those attributes need to be documented and supported by credible information.

A smaller independent hotel that communicates its differentiation effectively could compete for AI recommendations against much larger brands.

That is a particularly exciting opportunity.

6. The research supports the urgency

The article cites Phocuswright’s finding that nearly four in ten U.S. travelers used generative AI for trip research in 2025, an increase of 11 percentage points in one year.

Phocuswright

+1

There is also more recent evidence.

Phocuswright reported in March 2026 that 56% of U.S. travelers had used AI for planning, booking or in-destination assistance for at least one trip during the preceding 12 months.

Phocuswright

+1

One qualification: these statistics measure different forms of AI usage, so they shouldn’t be treated as a direct year-over-year comparison.

Nevertheless, the direction is clear. This is no longer something hotels can comfortably dismiss as an emerging technology experiment.

7. My biggest takeaway: Hotels need to rethink what digital marketing means

I think this is ultimately the most important message.

For years, hotel digital marketing has largely revolved around:

• Search engine optimization.
• Paid search.
• Social media.
• Online travel agencies.
• Email marketing.
• Website conversion.
• Reputation management.

All of these remain important.

But there is a new strategic question:

When a traveler asks an AI platform where to stay, why should that AI recommend this particular hotel instead of its competitors?

That question requires a different approach to digital marketing.

It requires hotels to establish a clear identity, build credible supporting information, communicate differentiated experiences and ensure that the information available across the internet consistently reinforces their positioning.

I particularly like the article’s concept of Knowledge Formation Optimization (KFO). It recognizes that the objective is not merely to make a hotel visible, but to help AI systems understand why the property is relevant to particular travelers.

I would, however, be careful about treating KFO as an established technical discipline with guaranteed ranking outcomes. There is no universal formula for securing AI recommendations, and the underlying systems continually change.

My final assessment

Strong strategic argumentI would give the article an 8.5 out of 10.

Its strongest contribution is reframing AI from an internal hotel efficiency tool into an external customer acquisition and distribution challenge.

Its biggest weakness is that it identifies the problem more thoroughly than it explains how hotels should execute the solution or measure its financial return.

But I think the underlying message is absolutely worth taking seriously.

One final thought: I believe the hotel industry may be approaching a period in which there are two distinct types of digital visibility:

1. Traditional search visibility: Can travelers find your hotel when they search?
2. AI recommendation visibility: Does your hotel enter the conversation when travelers ask where they should stay?

The second could become a major competitive differentiator.

And the hotels that start understanding and addressing that distinction now will have an opportunity to establish an advantage before their competitors recognize the importance of the shift.

One thing I’d be particularly interested in exploring further is this: How would we actually test whether ChatGPT, Gemini and Perplexity recommend a particular hotel when travelers ask destination-specific, unbranded questions, and then build a measurable strategy to improve those recommendations?

I think turning that question into a practical hotel AI discovery audit would be a very valuable next step.