On 15 October 2026, my colleague Nick Hollows and I presented an Expert Session for the IHG Owners Association: “The Invisible Hotel: How AI Search Decides Which Hotels Get Seen and Booked”. More than 150 hotels registered, with owners, operators and commercial teams joining from across EMEAA, led by the UK and Ireland, Iberia and Italy.

When members registered, they answered three short questions. The answers set the tone for the whole session:

• 57% had heard of AI Search but not yet acted on it.
• 35% were actively working on it, and 8% had never heard of it.
• 59% had checked whether their hotel appears in ChatGPT or Google AI Overviews for their city.
• 46% said no one at their hotel has primary responsibility for AI Search yet.

That last number stayed with me. AI is already deciding which hotels guests see, and in nearly half of these hotels, nobody owns it.

The best part of the session was not the slides. It was the questions owners typed into the chat. They were practical, sharp and honest, so rather than recap the deck, this post shares the questions that came up and the answers we gave.

Three ideas carried the session.

Guests are asking, not searching. A traditional search is three to four keywords. A conversation with AI averages 20 to 25 words, full of personal circumstances, wants and needs. What comes back is not a page of blue links. It is an answer, and a citation.

Being the answer is not enough. When I recently asked Google for a CBD hotel in Melbourne, its AI recommended a hotel and cited Expedia as the source. If AI recommends your hotel but cites an OTA, the OTA may still get the booking.

Demand is not falling. Discovery is moving. Google search interest in hotels is up 25% in London, 53% in Rome, 25% in Berlin and 23% in Dublin, comparing 2026 with 2024. Yet one hotel we shared lifted its traditional organic rankings by 57% over a 90-day window, while its organic direct bookings fell 56% and organic direct revenue fell 48.8%. Strong rankings no longer guarantee visibility.

And the next step is already in testing. On 7 August, Google began a limited US test that lets travellers complete a hotel booking inside AI Mode, without leaving the answer. Today, the invisible hotel loses the recommendation. Tomorrow, it loses the booking.

1. What Is the Main Shift in Website Content That Improves Our Presence in AI?

Stop publishing adjectives and start publishing facts. “Centrally located” tells AI nothing. “Six minutes’ walk from the main station” gives it something to quote.

Think of it as a content hub with two sets of pages:

• Pages about your area. Guests usually book the flight first, then research the area. Walking times to the stations and venues they name, what is open late, where to eat on a Tuesday night, what to do when it rains. Publish this and you become the answer before they have even chosen a hotel.
• Pages about your hotel. The details guests ask AI about. When I asked what equipment a hotel gym had, the hotel’s own site offered a line of text and a pixelated photo. AI pulled the answer from OTA reviews instead, and I ended up booking on the OTA. Gym equipment, pool length, opening hours, room configurations: if the answer is not on your site, AI will find it somewhere else.

2. Our Website Is Managed by the Brand. What Can We Actually Control?

More than most owners think, because much of what AI reads about your hotel does not live on your website at all. Three actions need no web developer:

1. Generate reviews, across the hotel, restaurant, bar and spa. Aim for consistency, not perfection. The sweet spot is 4.2 to 4.8 stars, where guests see reviews as both credible and trustworthy.
2. Reply to every review. More on this below.
3. Keep your facts consistent everywhere. Rates, policies, room names and facilities should match on brand.com, your Google Business Profile, social media and every OTA.

One hotel we work with kept seeing the wrong checkout time in AI answers, even though its own website was correct. The source was a single OTA listing. We corrected that listing, and the right answer flowed through.

3. Does a Microsite Help a Branded Hotel in AI Search?

Yes, and a great deal. A microsite is one of the smartest moves a branded hotel can make, for AI Search and well beyond it.

• You control your own destiny. You decide what is published and when, without relying on anyone else’s template, queue or roadmap.
• You can move first. While competitors wait, you can publish your area and hotel content now and become the answer before they do.
• You are far more flexible. Add pages for your restaurant, bar, meetings and events or weddings, answer the questions guests actually ask, and update facts the day they change.
• It does not conflict or compete with brand.com. A microsite taps into another market and another revenue stream, reaching guests that brand.com does not.

Where we are given access, we also work to enhance and drive brand.com. The two work best together, which we cover in Microsites & Brand.com: Why Smart Hotels Need Both in 2026.

One rule applies to both: the facts must match everywhere. One new hotel we work with kept being described by AI as having a rooftop pool. It does not. The developer’s website had listed one in the original specification, and AI treated that as the source of truth.

4. Is There a Best-Practice Way to Write Review Replies So AI Treats Them as Fact?

Yes. Every review gives you the right of reply, and the guest cannot reply again. Your response is the final word in a piece of evidence that AI reads. A few pointers:

• Name the issue in plain words. “The lifts”, not “the issue you experienced”.
• State what is happening, what has changed, and when.
• Use the same names for rooms and facilities as your website.
• Keep it short and human. Write for the next guest; AI reads it the same way.

In the session I shared a large hotel whose lifts had been out for refurbishment for months. The reviews were scathing, none had replies, and AI began telling guests not to stay. A reply like this would have given AI the up-to-date picture:

“Thank you for your feedback, and we are sorry the wait for the lifts affected your stay. Several of our lifts are being refurbished, and full service resumes on [date]. Until then, our team is on hand in the lobby to help guests with luggage.”

Reviews are no longer just influencing future guests. They are influencing the AI that recommends your hotel to future guests.

5. If Every Hotel Starts Doing This, How Do We Still Stand Out?

A fair question, with two answers. First, most hotels have not started. Nearly half of the hotels registered for our session said no one owns AI Search yet.

Second, AI does not hand out one fixed top ten. It answers long, personal questions, matching specific needs to specific facts: the business traveller who needs the gym open at 6am, the family who needs connecting rooms. The hotel that has published those facts, consistently, wins those conversations.

As David Bark of Mulpha put it at AHICE 2026: “If you’re planning AI Search in your 2027 budget cycle, it’s too late.” Hotels that move in 2026 are competing against hotels that are still deciding.

6. How Do We Make Sure AI Visibility Drives Direct Bookings, Not More OTA Bookings?

Aim to be the citation, not only the answer, then measure it properly. Most hotel analytics run on last-click attribution, so a guest who discovered you through AI often arrives looking like a guest who already knew you.

We use a chain of independent evidence instead of one metric from one tool: what changed at the hotel and in the market, what changed in visibility and traffic, and the commercial outcomes in the hotel’s own reporting. Two of the examples we shared:

• A midmarket CBD hotel that added AI Search on top of traditional SEO for eight months grew direct revenue 38.3% year on year. Its direct share of public room revenue rose from 37% to 45.9%, at the direct expense of OTAs.
  
• A premium CBD hotel that replaced traditional SEO with AI Search for three months, in a falling market, grew direct web room nights 9.1%. OTA room nights grew 1%.

Other factors, including hotel marketing and rate decisions, also shaped those results, which is exactly why we look at the whole chain. We cover the method in more depth in How to Prove AI Search Actually Works for Your Hotel.

We closed the session with the questions every owner should now be asking their team:

1. Are we optimising for AI-driven search results, or still relying on traditional SEO?
2. When guests use ChatGPT or Google AI Overviews to find hotels in our area, is our hotel the answer?
3. If we are the answer, are we the source of truth, or do OTAs control the narrative?
4. Are we investing where guest discovery is actually shifting?

Members can request a complimentary AI Search and Commercial Opportunity Analysis for any hotel, any flag. There is no commitment and no obligation. It exists so you can make an informed decision about where your hotel stands, with or without us. It shows how visible your hotel is across AI-driven discovery, whether you or the OTAs control the narrative, how you compare with competitors, and the potential direct revenue upside.

More than 30 hotels have already requested theirs. The offer closes on 31 October 2026. Request your analysis here.

A sincere thank you to the IHG Owners Association for hosting us, and to every member who joined and asked such good questions.

I will leave you with the thought we finished on. The invisible hotel is not invisible because it is a bad hotel. It is invisible because today’s guests never discover it. The question is not whether guest discovery is changing. The question is whether your hotel is changing with it.

With 25+ years in hotel digital marketing, Michael has driven success for Accor, IHG, and Marriott worldwide.