How AI assistants pick a hotel

In a March 2026 Phocuswright survey, 56% of US travellers had used AI for planning, booking or help during at least one trip in the past year, up from 33% in the first half of 2025.

Nobody outside these companies knows exactly how an answer is put together. But each of them has published what its assistant reads. This guide sticks to that: their own documentation, plus a few public studies, with a link for every fact.

An assistant doesn't answer a hotel question from memory alone. It runs searches and reads what comes back.

OpenAI says ChatGPT search “leverages third-party search providers, as well as content provided directly by our partners”, and that it “typically rewrites your query into one or more targeted queries”. Google describes the same idea for AI Overviews and AI Mode and calls it “query fan-out”: several related searches for one question.

So your guest's one sentence becomes several searches. A place that shows up in several of them has more chances of being read.

This is the part most owners don't expect. The assistants don't share one list.

• Gemini reads Google's own data. It “gets real-time information — think locations, reviews, prices, and more — from Google Maps, Flights, and Hotels”.
• Google AI Mode brings together “real-time Search data for flights and hotels, details from Google Maps like photos and reviews and relevant information from sites across the web”.
• ChatGPT uses web search, plus apps that travel companies built inside it. Booking.com and Expedia were there from October 2025. Tripadvisor followed, promising to “sift through our millions of reviews and ratings”.
• Perplexity uses web search with numbered sources, plus a partnership with Tripadvisor for reviews.
• Claude uses web search, with direct citations to the pages it used.

In practice, three things about your hotel get read: your Google Business Profile, your listings and reviews on the big travel sites, and your own website.

Here the public studies disagree, and it's worth knowing that they do.

• Cloudbeds looked at answers about 145 hotels in six destinations (July 2025). Online travel agencies were 55% of the sources, led by Booking.com. Hotel websites were 13.6%.
• Koddi collected 2,000 ChatGPT answers about hotels (September 2026). About 44% of the sources were hotel or brand websites, and about 17% were OTAs.
• Yext counted 6.8 million sources across industries (October 2025). On Gemini, 52% were the business's own website. On ChatGPT, almost half were third-party sites, and for hospitality the main one was Tripadvisor.

All three come from companies that sell to hotels, and they measured different assistants at different times. The fair reading: your own site and your listings both get read, and the mix depends on the assistant. Keep your listings, and make sure your own site can be read too.

Ask the same question twice and you often get a different list. SparkToro and Gumshoe had 600 volunteers run the same prompts 2,961 times (January 2026), and found a less than 1 in 100 chance that ChatGPT or Google's AI would give the same list of brands in any two answers. Koddi saw the same with hotels: one prompt, asked five times, and the recommended hotels overlapped by only 37.3% on average.

Two things follow. One screenshot of ChatGPT naming you, or not naming you, proves very little. And anyone who promises you a fixed place in AI answers is promising something these systems don't do.

None of this needs a big budget. Each step comes from the assistants' own guidance.

1. Let their search crawlers in. Sites that opt out of OpenAI's search crawler, OAI-SearchBot, “will not be shown in ChatGPT search answers”. Ask whoever looks after your website to confirm that the site and its security settings don't block OAI-SearchBot, PerplexityBot, Claude-SearchBot or Googlebot.
2. Put your facts in plain text on your own site. Room types, location, what's nearby, how to book. Google's checklist asks that “important content is available in textual form”, and a Vercel and MERJ study found that none of the major AI crawlers ran JavaScript, apart from Google's. If your rooms only appear inside a booking widget, several assistants may never see them.
3. Keep your Google Business Profile current. Google names it among the things that help a business show up in AI answers. Check the category, the hours, the photos and the link to your own site.
4. Say the same thing everywhere. Name, address, phone, what kind of place you are. The assistants read your site, Google, Tripadvisor and the booking sites side by side, and OpenAI warns that “search results and citations can be incomplete, outdated, or incorrect”. Conflicting details make that worse.
5. Earn real reviews, and answer them. Reviews are what Tripadvisor's and Google's tools summarise. Never buy or write them: Google says chasing “inauthentic mentions” doesn't help.
6. Skip the magic files. Google says there is “no special schema.org markup you need to add” for AI answers. Clear pages beat tricks.
7. Check for yourself. Write down five questions your guests would really ask. Ask each assistant, more than once, on different days, and note who gets named and where the links go. Our step-by-step guide shows how, in 15 minutes.

Nobody can promise that an assistant will name your hotel. What you can do is make sure that when it goes looking, the facts about your place are there, easy to read, and the same everywhere.