What ChatGPT and Google AI tell guests about your hotel
Ask ChatGPT, Google AI and Claude the questions guests ask, then check if your hotel is named, which link and price they give, and which policies they repeat.
To find out what ChatGPT and Google AI tell guests about your hotel, ask them the questions a guest would ask, in a fresh chat, and read the answers closely. Check four things every time: whether your hotel is named, which booking link is given, which price is quoted, and which policies are repeated.
Why does it matter what AI says about your hotel?
Guests can now plan a trip by asking an AI assistant instead of typing a few words into a search box. They ask full questions, such as “a quiet beach hotel with a kids’ club for four nights in April”, and the answer can shape which hotel they look at next.
That answer is built from sources you do not fully control. It may describe your hotel from your own website, from an OTA listing, from a review site or from an old social post. It may quote a price from last season. It may send the guest to a booking page that is not yours. If you never look, you will not know.
We covered how assistants choose hotels in how ChatGPT recommends hotels. This guide is the practical follow-up: how to check what they actually say about you, and what to do with what you find.
Can you see what guests ask AI about your hotel?
No. Nobody can see the private questions real guests type into ChatGPT, Claude or Google AI. There is no report for hoteliers that lists them, and no tool can read them.
What you can do is ask the same kinds of questions yourself and save the answers. That gives you a fair picture of what a guest is likely to see. It is a sample, not a full record, so treat it as a regular health check rather than an exact measure.
It also helps to be clear about one more limit. Nobody can edit an AI answer directly. You cannot log in somewhere and correct what ChatGPT says about your pool hours. You can only fix the sources the assistant reads, which we cover below.
How do you test what AI assistants say about your hotel?
Set aside 30 minutes and follow these steps. We use a fictional hotel, Seaview Resort, Seabay, as the example.
- Use a fresh session. Sign out, or use a private browser window, so past chats and your own location do not shape the answer too much. Note the date and which assistant you used.
- Ask a discovery question without your name. For example: “Which family resorts in Seabay have a private beach?” This shows whether you appear at all when a guest has not heard of you.
- Ask a direct question with your name. “Tell me about Seaview Resort, Seabay. Is it good for families?” This shows how the assistant describes you.
- Ask about price. “How much is Seaview Resort, Seabay for two adults on the second weekend of next month?” Compare the answer with the live price on your own booking engine for the same dates.
- Ask how to book. “Where is the best place to book Seaview Resort?” Click every link in the answer and note where it lands.
- Ask about policies. “What is the cancellation policy at Seaview Resort? Is breakfast included? What time is check-in?” Compare each point with your current policies.
- Repeat in each assistant. Run the same questions in ChatGPT, Google AI (AI Overviews or AI Mode) and Claude. Answers often differ between them.
- Save everything. Copy each answer and its links into a simple document or spreadsheet. Next month, you will want to compare.
Ask in the languages your guests use, too. A hotel with many guests from the Gulf might ask in Arabic as well as English, because the sources an assistant finds can differ by language.
What should you look for in each answer?
Read each answer as a guest would, then check it against the facts. This checklist covers the points that most often cost a hotel a direct booking.
| Check | What good looks like | Warning signs |
|---|---|---|
| Is the hotel named? | Your hotel appears for the discovery questions that fit it | Only nearby competitors are named, or your hotel is described under an old name |
| Which booking link? | The link goes to your own website or booking page | The only links go to OTA listings, or to an old booking system you no longer use |
| Which price? | The price quoted is close to the live price on your booking engine | A price from an old offer, a price without taxes, or a price only an OTA shows |
| Which policies? | Cancellation, check-in, children and meal details match your current rules | A free cancellation promise on a non-refundable rate, or a closed facility described as open |
| Which facts? | Room types, facilities and location are correct | Renovated rooms described in their old style, or the wrong distance to the beach |
| Which sources? | Cited pages include your own website | Cited pages are mostly third-party sites, or old blog and social posts |
The booking link deserves special attention. If an assistant names you and recommends you warmly, but every link points to an OTA, the guest you earned will most likely book there. The reasons this happens, and who should fix each one, are in why AI assistants send your guests to OTAs.
What should you do when an AI answer is wrong?
Because you cannot edit the answer, you work on what the assistant reads. Start with the source it cites, if it shows one. Then work outwards.
- Your own website first. Make sure every Book Now button goes to your own booking engine, and that rooms, facilities and policies are written as plain text on the page.
- Old offers and pages. Remove or update expired offer pages, old press releases and outdated FAQ answers. Assistants can still find them.
- Rate names. If a rate is called “Flexible” but is non-refundable, an assistant may repeat the wrong promise. Make names match the rules. Our cancellation policy guide helps here.
- Listings. Keep your Google Business Profile and OTA listings consistent with your website. Being on Google Hotels free booking links also puts your own rate where guests compare.
- Live rates. Give assistants a way to read real prices, through an MCP connection to AI assistants, instead of guessing from old pages.
Then be patient. Fixing sources helps, but nobody can promise an AI answer will change, or how fast. Changes tend to show over the following weeks, and it varies by assistant and by question. That is why checking regularly matters more than checking once.
For a full list of things to put right, see the AI readiness checklist for independent hotels.
How goDirects AI Visibility runs these checks for you
Doing this by hand every month across three assistants is slow, and it is easy to skip. goDirects AI Visibility is built into the goDirects booking engine to do the routine part for you.
Every month it asks Google AI, ChatGPT and Claude the questions guests ask and saves the answers. For each answer it shows:
- whether the assistant named your hotel;
- where it sent the guest, your booking site or somewhere else;
- which price it quoted, next to the live price on your goDirects booking engine;
- any wrong policy it repeated.
You also get an AI readiness score, checked every morning, with a short fix list. Each fix has an owner: you, your website team, or goDirects. You can add some questions of your own, for example about your spa or your airport transfer.
The “AI Guests” view shows price checks and room holds made through AI assistants connected to goDirects. It shows counts, never guest names. AI Visibility is included at no monthly fee: there is no monthly fee and no setup fee; you pay only on confirmed, paid bookings. Read more on the AI Visibility page.
If you would like to see what AI assistants currently say about your hotel, book a demo and we will walk through it with you.
Frequently asked questions
How often should a hotel check what AI says about it?
Once a month is a sensible rhythm for most independent hotels. Answers change as assistants update and as new pages appear about you. Check again soon after a big change, such as a new rate structure, a renovation or a new booking engine, and keep your saved answers so you can compare month to month.
Can I see the questions guests ask ChatGPT about my hotel?
No. The questions real guests type into AI assistants are private, and no tool can show them to a hotel. You can only ask similar questions yourself and save the answers. That gives a useful sample of what guests are likely to see, but it is not a complete record.
Can I ask ChatGPT or Google to correct an answer about my hotel?
You cannot edit an AI answer directly. What you can do is correct the sources the assistant reads: your website, your listings, old offers and rate names. Fixing sources helps, but no one can promise when an answer will change. It usually shows over the following weeks, and it varies.
Why does ChatGPT quote a different price from my website?
Assistants often read prices from pages they find, which can be old offers, OTA listings or prices shown without taxes. Unless an assistant can read your live rates, it may guess. A live connection such as MCP lets assistants connected to your booking engine quote your real price for the dates asked.
Do I need to test in more than one AI assistant?
Yes. ChatGPT, Google AI and Claude can give different answers to the same question, because they search and summarise in different ways. A hotel can look correct in one and wrong in another. Test the same questions in each, on the same day, and note the differences.