Does ChatGPT recommend my hotel?

A growing share of travelers plan trips by asking ChatGPT, Gemini, Claude or Perplexity where to stay. The assistant does not return a page of links to compare. It returns three to five hotel names.

For that traveler, those names are the whole shortlist. If your hotel is not among them, it is not considered at all, however good it is.

You can get a rough picture yourself tonight. Open ChatGPT, and if you can, one other assistant such as Perplexity or Gemini. Start a fresh conversation for each question so earlier answers do not influence later ones.

Ask the questions your guests actually ask. Do not ask about your hotel by name, because a traveler who does not know you will not either. Adapt these to your city and your guest:

• “Best boutique hotel in {your city} for a quiet weekend”
• “Where should I stay in {your city} for a first visit?”
• “Romantic hotel in {your city} for an anniversary”
• “Family-friendly hotel in {your neighborhood}”
• “Hotel in {your city} near {the landmark or station your guests use}”

Write down every hotel named in every answer. Five questions on two assistants gives you ten answers, which is enough to see a pattern.

Four things tell you where you stand.

1. Do you appear at all? Count how many of the answers name your hotel. This is the core of your AI visibility.
2. Where do you appear? First in a list of four is a different position from fourth.
3. How are you described? Read the sentence the assistant writes about you. It often reveals which guests it thinks you suit, and it may not match how you see yourself.
4. What does it link to? If the assistant cites sources, check whether it points to your own website or to an OTA listing. A recommendation that sends the guest to an OTA still costs you commission.

Also note the hotels that appear instead of you. Those are your real competitors in AI answers, and they are not always the ones you would expect.

When a good property is missing from these answers, the cause is rarely the property itself. It is usually one of these.

• The AI crawlers are blocked. If your robots.txt blocks crawlers such as GPTBot, ClaudeBot or PerplexityBot, that assistant cannot read your site at all. This often arrives with a security setting nobody revisited.
• Your facts are not machine readable. Without Hotel structured data, an assistant has to infer your location, size and amenities, or take them from an OTA listing written by someone else.
• Your positioning is not written down. If your site does not say plainly who you are for, whether couples, families or business travelers, an assistant cannot match you to a question asking for exactly that.
• Your name is inconsistent across the web. Different spellings of your hotel name on your site, your Google Business Profile and OTA listings make it harder for an assistant to recognize you as one property.
• Few trusted sources mention you. Assistants lean on third-party sources. Properties featured in travel press and credible guides get named more often.

A manual check is a good start, but it has limits.

AI answers vary between runs. Ask the same question twice and you may get different hotels. The four major assistants also behave differently from each other, so a hotel can be visible on one and absent from another. And your guests ask many different kinds of questions, so appearing for one type of trip says little about another.

A reliable picture needs many questions, across all four assistants, each asked more than once and averaged. That is the part that is hard to do by hand, and it is what a proper measurement does.

In rough order of impact.

1. Check your robots.txt and allow GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended.
2. Publish Hotel schema markup with your address, coordinates, room count, price range and amenities.
3. State who you are for in plain words on your main pages, including your neighborhood and what staying there is like.
4. Use one spelling of your name everywhere, on your site, your Google Business Profile and your listings.
5. Publish an llms.txt, a short plain-text summary of your property written for AI systems.
6. Give assistants something to cite on your own domain, such as a detailed rooms page and a neighborhood guide, so recommendations point to you rather than to an OTA.

If you want to see what a complete measurement and fix plan looks like, our sample report shows one in full, using a demo property.