Why hotel websites stop being found by AI (and how to fix it)

A guest asks ChatGPT for a quiet riverside hotel within an hour of Bangkok. Another asks Perplexity for a family resort with a kids' club on Koh Samui. A third types a question into Google and reads the AI Overview before any blue link.

In each case, a machine decides which hotels to name. It can only choose from hotels it can find, read and trust. Many hotel websites fail at the first two steps, and the owners have no idea.

AI answers are the new front page of travel. Here is how those answers are built, what they can read on your site, and how to check yours.

How AI assistants choose which hotels to name

ChatGPT search, Perplexity and Google's AI features work in broadly the same way. They take a question, run searches behind the scenes, fetch pages, and write a summary with links to the sources they used. Google describes a "query fan-out" technique, where one question becomes several related searches across subtopics.

That means two things for a hotel:

1. Your pages must be reachable. If the crawler that feeds an assistant's search results cannot fetch your site, you are not in the pool of candidates.
2. Your pages must be readable. Once fetched, the system needs to pull clear facts from the page: what you are, where you are, what rooms you have, what they cost, and who you suit.

Reputation still matters: credible mentions, consistent information across the web and depth on your topic. But reputation cannot help a page the machine never reads.

Crawler access: the silent blocker

Every AI company publishes the names of its crawlers. The important detail is that most run more than one, for different jobs.

Company	Crawler	What it does
OpenAI	OAI-SearchBot	Finds sites to show in ChatGPT search answers
OpenAI	GPTBot	Collects content that may be used to train models
OpenAI	ChatGPT-User	Visits a page when a user's request needs it
Perplexity	PerplexityBot	Finds and links sites in Perplexity answers
Perplexity	Perplexity-User	Visits a page when a user asks a question
Google	Googlebot	Crawls for Search, including AI Overviews and AI Mode
Google	Google-Extended	A control token for Gemini training and grounding outside Search

This split gives you a real choice. Blocking training crawlers such as GPTBot or Google-Extended is a reasonable position. But if you also block OAI-SearchBot, OpenAI says your site will not be shown in ChatGPT search answers. Google says Google-Extended does not affect inclusion in Search, and that Googlebot rules are what control AI Overviews.

There are two common ways hotels block AI by accident:

• A blanket rule in robots.txt. Some sites carry a Disallow: / left over from a staging build, or a list of "bad bots" copied from a forum that includes search crawlers.
• Bot protection at the host or CDN. Google's own guidance tells site owners to check that crawling is allowed "by any CDN or hosting infrastructure", not only in robots.txt. Perplexity publishes firewall instructions for the same reason. A security setting can quietly reject crawlers that your robots.txt welcomes.

Structured data: telling machines what you are

Structured data is a block of code, using the shared vocabulary at Schema.org, that labels the facts on a page. A human sees "Deluxe River View, 42 square meters, king bed". Structured data tells a machine that this is a hotel room, with this size, this bed type and this occupancy.

For a hotel, the core types are:

• Hotel (a type of LodgingBusiness) on the homepage: name, address, phone, geo-coordinates, check-in and check-out times, star rating, amenities.
• HotelRoom on each room or suite page: size, bed configuration, occupancy, room amenities.
• Offer on packages and rate pages: what is included, price, currency, validity dates.
• FAQPage where a page genuinely shows questions and answers.

Be realistic about what this does. Google states there is no special schema needed to appear in AI Overviews, and it removed FAQ rich results from Search in May 2026. Structured data will not win you a placement on its own. What it does is remove ambiguity, giving every system that reads your page the same clean, labeled facts. Google asks that the markup matches the visible text. It is common to find hotel sites with none at all, or with a single generic block from a theme that describes the hotel as an "Organization".

llms.txt: a summary written for machines

llms.txt is a proposed standard, first published in September 2024, for a plain-text Markdown file at the root of your site. It gives AI agents a short description of the site and links to the most useful pages, without the menus, pop-ups and scripts that make HTML hard to parse.

Be honest about its status. None of the AI search products covered here list llms.txt as a requirement, and Google says you do not need new machine-readable files for its AI features. The proposal's author reports that thousands of sites now publish one and that Chrome's Lighthouse checks for it. For a hotel, it is a short file stating your facts plainly: who you are, where you are, your rooms, your policies and how to book. Cheap to add and easy to keep current, but it will not fix a site that crawlers cannot reach.

Entity-rich content: facts beat adjectives

An "entity" is a thing a machine can identify: your hotel, your town, the river, the airport, the night market down the road. AI systems build answers by connecting entities. Copy that is all mood and no facts gives them nothing to connect.

Compare two descriptions of the same room:

• "Indulge in timeless elegance in our beautifully appointed sanctuary."
• "A 38-square-meter room on the second floor with a king bed, a rain shower, a private balcony over the river and space for one extra bed. Sleeps up to three."

The second can answer a real question: "Which hotels near the river have rooms for a family of three?" The first cannot answer anything.

Pages that tend to help:

• One page per room type, with size, beds, occupancy and view.
• A location page with real distances and travel times to the airport, landmarks and transport.
• Clear policies: check-in, children, pets, cancellation, accessibility.
• Dining, spa and meeting pages that name opening hours, capacity and what is actually offered.
• The same name, address and phone number everywhere: your site, Google Business Profile and OTA listings.

Fast, clean HTML

Speed matters for guests first. Hy Digital, our sister hotel marketing agency, advises hotel sites to load in under three seconds. Structure matters for machines. If your room descriptions, prices or address only appear after a slider script or a booking widget loads, a machine reader may never see them. The safest assumption is that a crawler reads the HTML your server sends. Put your important facts in plain text in that HTML, with proper headings, and keep the heavy decoration optional.

A real example: fast site, invisible facts

A technical review in October 2026 looked at a riverside resort near Bangkok. The site loaded in under two seconds, so speed was not the problem. Everything else was:

• No Hotel, HotelRoom, Offer or FAQ structured data on any page.
• No llms.txt, and no deliberate rules for AI crawlers in robots.txt.
• No working analytics, so the resort could not see AI referrals, or any other traffic, at all.

A fast website can still be close to invisible to machines. The fixes were not dramatic: a list of missing basics.

The checklist: run it on your own site

You do not need a developer for most of this. Set aside an afternoon.

Access

• Open yourhotel.com/robots.txt. Look for Disallow: / under User-agent: *, and for any rules naming OAI-SearchBot, PerplexityBot or Googlebot.
• Decide on purpose whether to allow training crawlers (GPTBot, Google-Extended). Keep search crawlers (OAI-SearchBot, PerplexityBot, Googlebot) allowed.
• Ask your host or agency whether firewall or bot-protection settings block AI crawlers.
• In Google Search Console, run URL Inspection on your homepage and a room page. Confirm both are indexed.

Readability

• In URL Inspection, view the crawled HTML. Can you find your room names, address and phone number in it?
• Paste your homepage and a room page into the Schema Markup Validator (validator.schema.org). Look for Hotel, HotelRoom and Offer.
• Check that the structured data matches what is on the page: same prices, same room names.
• Visit yourhotel.com/llms.txt. Does it exist, and is it accurate?

Content

• Does every room type have its own page with size, beds, occupancy and view?
• Is there a location page with real distances and travel times?
• Are your name, address and phone identical on your site, Google Business Profile and OTA listings?

Reality check

• Ask ChatGPT, Perplexity and Google three questions a real guest would ask about your area. Note which hotels are named, which sources are cited, and whether anything said about you is wrong.
• In GA4, look at traffic acquisition for referrals from AI assistants. If you cannot see any traffic at all, your analytics may not be loading.

If you only fix three things this month, make them crawler access, room-level facts in plain text, and Hotel and HotelRoom structured data.

For a wider view of how AI search is changing SEO, Hy Digital's SEO and AI in 2026 is a good next read.

Keeping it fixed

The hard part is not fixing these things once. It is keeping them right while rooms are renamed, packages change and someone installs a new plugin.

HotelierKit is a managed website platform for hotels. Structured data is generated on every page from your real content, llms.txt and AI crawler rules are maintained for you, and content health scoring shows which pages need attention. You are experts in hospitality. We are experts in the digital infrastructure underneath it.

If you want to see how your site would look to an AI assistant, book a demo and we will walk through it with you.