AI-ready hotel website checklist: what to fix before guests ask AI about you

Yes. OpenAI, Anthropic and Perplexity send crawlers, programs that fetch web pages the way Google's crawler does: OpenAI's GPTBot and OAI-SearchBot, Anthropic's ClaudeBot and PerplexityBot among them. Your robots.txt file, the short text file at the root of your site, is where you tell crawlers what they may fetch, and OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers. Google works differently. Google-Extended is a name you use in robots.txt to allow or block Google's AI products, including Gemini, from using your content; it is a setting, not a separate crawler. When crawlers get in, your pages become material an assistant can draw on when a guest asks where to stay.

They do not read your site the way a guest does, though. Some of them see only the HTML your server sends, and miss anything that appears later through JavaScript. Structured data states your facts in a fixed format, which leaves less to interpret than marketing copy does. And assistants read what other sites say about you as well: in our case studies, Booking.com and TripAdvisor were among the sources they cited most. The checklist below covers all three. Tick items as you go: the ticks are saved in your browser, and the print button gives you a paper copy to hand to each person.

Before you hand this list round, it helps to know what it can and cannot do. In our published case studies, Golden Well Hotel, a boutique hotel in Prague, had the highest readiness score of the four hotels, 77.1%. AI assistants named it in 21 of 295 answers to questions that did not mention the hotel, a detection rate of 7.1%. Mosaic House Design Hotel, also in Prague, scored lower on readiness, 60.9%, and was named more often: 25 of 200 answers, or 12.5%.

So readiness did not predict visibility. A site AI cannot read will not be quoted, which is why the checklist matters. Being readable does not get you recommended on its own. That depends on whether the assistant already trusts other sources that talk about you, and on which questions guests ask. Nobody can guarantee that an AI assistant will recommend a hotel, and anyone who promises it is guessing.

Start with robots.txt, because a single broad Disallow line can shut out every AI crawler at once, and it can be left over from a staging site or a security plugin. Your developer should check the file for the crawler names above and confirm that sitemap.xml is referenced from it. The sitemap should list the room, amenity and policy pages, which are often the pages furthest from the homepage.

Next, the JavaScript test. Turn JavaScript off in the browser and reload the rooms page. If room names and descriptions vanish, a crawler that does not run JavaScript sees the same empty page. Whatever causes it, whether the theme, a page builder or the booking engine, the fix is to send the text in the first HTML response and let scripts handle the booking widget.

Speed and access come next. A site that is very slow, or that greets automated visitors with a bot check, can look unreachable to a crawler, and our own audit reports exactly that case. After that come page basics: one H1 heading per page, a page title and meta description that describe the page, and a language attribute on every language version. Last, publish llms.txt, a short plain-text summary of the site for AI agents. No major AI company has said it uses the file, so it belongs at the end of the list. Our guides to AI crawler readiness and to llms.txt cover both in more detail.

Guests ask AI assistants specific things. Is there parking, how far is the station, can I bring the dog, what time is check-in. An assistant can only quote your answer if it is written on your site in plain words. If your site has photos, a booking button and a paragraph of atmosphere, and leaves the room count, distances and policies to Booking.com, the assistant will quote Booking.com.

Write out the number of rooms and each room type with size, beds and occupancy. List amenities with the details guests ask about, such as where to park and what it costs, or breakfast hours. Give distances to the airport, the main station and the sights in minutes or kilometres. Spell out check-in and check-out times, cancellation terms, children, extra beds and pets. Then build an FAQ page from the questions your front desk answers every week, with a short answer to each.

Finally, add schema. Schema is structured data, a block of code in the page that visitors do not see, which states your name, address, phone, check-in times and amenities in a fixed format machines can read. Hotel schema covers the hotel itself, and FAQPage schema covers the FAQ page. Our free hotel schema generator writes the Hotel block from a short form, and our schema markup guide explains each field. Run the result through Google's Rich Results Test before you publish it, because a single missing comma makes the whole block invalid.

This part of the list sits outside your website, and it matters because AI assistants read those other sources too. When Golden Well was missing from an answer, the sources AI cited most were Booking.com and TripAdvisor, followed by a French travel blog and thehotelguru.com. For Mosaic House the list was Booking.com, TripAdvisor, Expedia and thehotelguru.com. These listings feed the answers about your hotel whether you look after them or not.

So the front office and revenue team should check that the hotel name, address and phone number are written the same way on the website, Google Business Profile, Booking.com, Expedia and TripAdvisor. Then compare amenities, policies, room count and check-in times. A listing that still says free parking after you started charging gives an assistant two versions of the truth, and it may pick the wrong one or skip the detail. Old listings you no longer use count too.

Reviews belong here as well. Reply to recent reviews on Google, Booking.com and TripAdvisor, and answer the specific point the guest raised, because a reply that confirms a fact, for example that the spa is open again, is more text about your hotel that says something. Then make it a habit: when something changes, update the site and every listing in the same week.

Not every item carries the same weight. If time is short, do robots.txt and the JavaScript test first, because each can hide the whole site. Then have the front office check name, address and phone across the main listings. Then write down the facts on the rooms page and add Hotel schema. The FAQ page, review replies and llms.txt can follow over the next few weeks.

Our readiness audit runs 16 checks on a hotel website, grouped under three questions: can AI reach your site, can AI understand it, and does it answer the questions guests ask. The checklist covers the same ground, so a scan run later will use the language you have already seen here.

A finished checklist tells you the site is ready. It does not tell you whether AI assistants name the hotel, and the case studies show those are different questions. The only way to know is to ask the assistants real guest questions that do not mention your hotel, and count how often you appear.

The type of question matters. Golden Well was named for occasion questions such as "best hotel in Prague for a honeymoon" and was missing from broad ones like "good hotels to stay at in Prague". Mosaic House was named when guests asked about unique hotels and was missing from questions about hotels in the city centre. If you test with one generic question, you learn very little about either.

Staylight's free scan does this for you: one scan per account, 15 guest questions on three AI engines, no card needed. It shows which questions you are named in, which you are missing from, and the readiness checks behind the score. Run it after the checklist, then again a few weeks later on the same questions, because a single check of an AI answer is a snapshot and answers vary from one day to the next.