Is Your Hotel Marketing Agency Selling You AI Pixie Dust?

Americas Great Resorts examined 1,177 AI-generated hotel recommendation answers containing 6,110 hotel appearances, across ten destinations and four engines: ChatGPT, Google AI Mode, Gemini, and Copilot.

Among 379 question-and-engine groups with three assessed answers, the recommended hotel lineup changed at least once in 363.

That is 95.8%.

The repetitions were collected on the same date. The difference could be as small as one hotel entering or leaving the list. This does not mean every answer changed completely. It means the lineup usually failed to remain identical across even three observations. Read the research.

A Screenshot Is Not Proof of Improvement

Now open the screenshot your marketing agency included in this month’s report.

The one where your hotel appears.

The one presented as evidence that the AI strategy is working.

What, exactly, does it prove?

That your hotel appeared in that answer when somebody asked that question.

It is an observation. Promoting it to proof of improvement requires evidence.

If the screenshot is the entire argument, congratulations. Someone successfully operated a keyboard.

The harder questions are still waiting.

How often does your property appear?

Is the description accurate?

Where does the link go?

What changed after the work?

What evidence connects the work to the result?

If the report cannot answer those questions, the screenshot needs a smaller job title.

Buying Software Is Not the Same as Doing the Work

Your agency may be excellent at photography, website design, social media, email marketing, and paid search. Those capabilities deserve credit.

They do not automatically establish competence in AI visibility.

Neither does buying a subscription.

Buying AI software proves someone bought AI software. The invoice arrives whether the person using it understands the problem or not.

And putting “AI-powered” in front of a service description does not explain what anybody actually does.

It does, however, take up space on a proposal.

A visibility tool can tell you where your property appeared in its tests. A good platform can track changes over time, investigate sources, and support corrective work. Some offer substantial capabilities.

The question is what your provider does with them.

A thermometer tells you the temperature. Read it repeatedly and you have a useful record.

Nobody should confuse that record with repairing the heating.

If you are paying someone to improve the conditions, a new temperature chart is not the whole deliverable.

Nor does anyone have a thermostat inside ChatGPT that guarantees your hotel a position. AGR does not. Your agency does not. Anyone promising that control should demonstrate it before you sign anything.

There are conditions outside the model that can be investigated and changed: inaccurate descriptions, contradictory specifications, outdated listings, missing coverage, broken links, and incomplete information.

What has your provider found?

What has it corrected?

What remains unresolved?

Those are ordinary questions about paid work. They should not require an extraordinary meeting.

You Got the Mention. Who Gets the Click?

And a mention is only the beginning.

In AGR’s companion research, Google Travel links appeared in 270 of 296 assessed Gemini answers. One captured card named Las Ventanas al Paraíso, a Rosewood resort. Its immediate destination was Google’s travel interface.

The hotel got the mention. Google got the next step.

We did not measure subsequent bookings, and that link does not establish an OTA transaction. It establishes something simpler: a hotel’s name in an answer does not guarantee a direct link to its website. Read the link analysis.

So when somebody tells you, “You’re showing up,” the next question is:

Showing up as what, and sending people where?

A green box beside your name cannot answer that.

AI Knows Your Name. Does It Recommend Your Hotel?

There is another question worth asking before anyone celebrates.

Did your property appear because the traveler described the experience they wanted, or because somebody handed the engine your hotel’s name?

Those are different achievements.

Ask an engine about your hotel directly and it may produce a beautiful description. Ask where to stay for the kind of trip your hotel is built to deliver, and your property may disappear from the answer.

Knowing your hotel exists is not the same as recommending it.

The traveler who already knows your name has found you somehow. The traveler asking where to stay is still choosing.

Before the next marketing meeting, try that experience yourself.

Ask the AI services your prospective guests might use questions appropriate to your property:

• What are the best luxury hotels in this destination?

• Which hotels would suit a family vacation here?

• Where should a couple stay for an anniversary?

• Which hotel is best suited to a corporate retreat or private event?

Choose the questions that match what you actually sell. Start without naming your hotel. Record the answers and repeat the exact questions in fresh conversations.

Read the descriptions. Inspect the sources. Follow the links.

A few answers will not establish your overall recommendation frequency. They can expose specific problems worth investigating.

One answer can contain an incorrect link. One description can name a restaurant that closed. A separate check of your website can reveal contradictory facts.

You do not need to watch the same mistake happen twenty times before investigating it.

Repetition helps establish how consistently something happens. Investigation establishes what is wrong with the information you found.

Your property might be absent. It might appear under an outdated identity. It might be accurately described but linked to the wrong website.

Those are different problems.

“Your visibility score is 62” does not explain which one you have.

“It Takes Time” Is Not a Progress Report

“Our agency handles that.”

Fine. Ask to see the investigation.

“We’re working on it.”

What has been completed?

“It takes time.”

Of course it takes time. So does renovating a lobby. You can still ask what has been demolished, what has been ordered, and why nobody has touched the ceiling.

Time is a condition of doing work. It is not evidence that work is being done.

Ask for the observed problem, the supporting evidence, the corrective action, and the way the result will be assessed.

If the answer is another dashboard, you have learned something about the service.

See What an Actual Audit Contains

AGR is a hotel marketing agency too. The same standard applies to us.

Our research documents how AI recommendations behave. It does not, by itself, prove that our service improved a client’s results.

Our Knowledge Formation Optimization work investigates and corrects the public information surrounding a property, strengthens supporting evidence, and examines what changes in subsequent AI answers. We must document the work and distinguish observed results from explanations that remain uncertain.

The audit is where that investigation begins.

It examines descriptions, cited sources, competing recommendations, and factual inconsistencies. It checks the property’s own information alongside the outside record.

An incorrect website link is actionable even if your name appeared.

An outdated restaurant listing deserves correction even if your visibility score improved.

Conflicting specifications on your own pages remain a problem after the presentation ends.

You do not have to take our description of the audit on faith.

Our published June 2026 Estate Yountville AI Visibility Audit combines dated ChatGPT and Gemini observations with an investigation of booking-platform descriptions, outdated restaurant information, and gaps between the property’s documented strengths and its online representation.

Read the six-page example. Then look at what you currently receive.

Does it identify specific issues, show the evidence, and explain what needs attention?

Or does it mostly confirm that someone checked?

Americas Great Resorts offers a free, no-obligation AI Visibility Audit for independent luxury hotels and resorts.

Our current standard audit covers ChatGPT, Gemini, and Google AI Mode. We record the first answer to each query in a fresh, logged-out session, then separately investigate your website and relevant public sources. The audit identifies documented problems and corrective priorities; it does not measure how frequently your property will appear across repeated runs.

Where we find issues, you receive a detailed PDF documenting:

• Where your property appeared or was absent in the tested answers.

• Which competitors appeared instead.

• How your property was described and where that description conflicts with verified information.

• Which sources and links the answers presented, where available.

• Outdated information, factual inconsistencies, and relevant technical problems found during the investigation.

• What needs correcting, reconciling, or further examination.

The findings are dated. We do not turn one missing recommendation into a declaration of permanent invisibility. We also do not ignore a documented error because somebody found a favorable answer elsewhere.

If we find no issues requiring a report, we tell you.

See where your property stands. Request your free AI Visibility Audit.

Your next guest may already be asking AI where to stay. Is your hotel in the answer?