Part 2 of the series «AI visibility for hotels: The roadmap» (link to Part 1)
Introduction
In the first part of this series, you tested your own hotel in ChatGPT, Gemini and Perplexity. The result may have been sobering: your hotel barely appears, or not at all, even though the website is well maintained, up to date and visually appealing. The reason for this is rarely the design — it lies in the code. AI systems read a website differently than a human does, and this is exactly what determines whether your information gets through at all.
The problem in everyday hotel life
Many hotel websites are made for guests: nice pictures, evocative texts, a booking button. For an AI, this is often too vague. Terms like «unique feel-good ambience» or «dreamlike location» mean nothing to a machine. It cannot extract facts from them — and if it finds no clear facts, it simply won't recommend your hotel. On top of that, many hoteliers don't even know that their website needs certain technical building blocks before AI systems can read it at all.
Current findings from a study / occasion
A large-scale study of 121'425 hotel homepages in seven countries (Nicolas Sitter, Hotel Schema.org Adoption Study 2026) shows just how big the technical gap in the industry actually is:
- Only 10.6 percent of hotels have good technical markup of their website content (so-called schema markup) — the majority of the industry provides AI systems with no fact base at all, or an inadequate one.
- 41 percent use the wrong data type and incorrectly fail to mark their property as a hotel.
- The average quality score is only 14.3 out of 100 points, with the median value even at 0.
- Reviews are hardly ever stored in a machine-readable way: only 12.5 percent of websites mark up their guest reviews so that an AI can capture them automatically.
- Additionally, many hotels unintentionally block AI crawlers: studies show that a relevant proportion of hotel websites have set up their robots.txt access file so that AI systems such as ChatGPT or Perplexity are not even allowed to crawl the page — without those responsible being aware of it.
- Schema markup has long since become more than an SEO detail: Google, ChatGPT and Perplexity now use this structured data as a primary source of information for their answers — not merely for attractive search result previews.
What does this mean for your hotel?
1. Facts instead of atmosphere: how to make your texts AI-readable
Supplement flowery descriptions with concrete, verifiable facts: How many square metres does the standard room have? How far is the station, in walking minutes? What amenities are included in the price? An AI can work with «21 m², 8 minutes' walk to the station, free WiFi and breakfast buffet included» — not with «our cosy room in a central location».
2. Schema markup: the «labels» in the code that no one sees
Schema markup is a technical standard that stores information such as price, star category, address or amenities in the background of your website — invisible to the visitor, but machine-readable for Google and AI systems. Ask your web developer specifically about «Hotel schema», «FAQPage schema» for frequently asked guest questions, and «Review schema» for your ratings. This can be checked with common testing tools such as the Google Rich Results Test — a look that is worthwhile before you invest in further measures.
3. Check whether you are accidentally locking out AI systems
Your website's robots.txt file determines which automated programs are allowed to crawl it. Have it checked whether your website explicitly allows search engine crawlers such as those of ChatGPT or Perplexity. A single incorrectly set entry can mean that your hotel remains completely invisible to these systems — regardless of how good the rest of the content is.
Concrete practical example
Starting point: A city hotel in Lucerne with 60 rooms found, as part of the self-test from Part 1, that it barely appeared for general guest questions. The website was modern in design, with high-quality photos and appealing texts — but had not been touched technically since its last overhaul four years earlier.
Measure: The web developer commissioned checked the website with the Google Rich Results Test and found that no schema markup was in place at all. He added Hotel schema with address, star category and amenity features, as well as FAQPage schema for the ten most common guest questions (check-in times, parking options, pet policy). At the same time, the robots.txt file was corrected: until then it had blocked all automated access, including that of ChatGPT and Perplexity.
Expectation: After implementation, the hotel will be re-checked with the self-test from Part 1 after four to six weeks. The expectation: since AI systems now have both access to the page and clear, structured facts, the hit rate on the test questions should improve noticeably.
Common mistakes
- Nice language instead of clear facts. Promotional texts may appeal to guests emotionally, but for an AI they are worthless as long as no verifiable information stands behind them.
- Never checking the robots.txt file. Many hotel websites unknowingly block AI crawlers — often because an old security setting was never adjusted.
- Setting up schema markup once and forgetting about it. Outdated or incorrect structured data (for example, an old room price) does more harm than having none at all, because it creates contradictions with the visible content.
Conclusion
An appealing website alone is no longer enough to be visible to AI systems. What matters is what happens in the background: clear facts instead of marketing language, correctly implemented schema markup, and a robots.txt file that does not accidentally deny access to AI crawlers. The good news: these technical building blocks can be retrofitted with manageable effort, and the vast majority of the industry has not yet implemented them — a clear advantage for anyone who acts now. The next part of the series looks at how to write content that an AI actually cites as an answer.
Sources
- Nicolas Sitter: «Hotel Schema.org Adoption Study 2026» – analysis of 121'425 hotel homepages in 7 countries
- Google Search Central: «Schema markup for vacation rentals»
- Digital Fox: «Hotel Schema Markup: The 2026 Implementation Cheatsheet»
More articles on this topic
- How visible is your hotel really (Part 1 – back-link)
- SEO Hotel: Why a slow website costs you bookings (technical interface)
- Why ChatGPT recommends your hotel – FAQ page (concrete implementation)
- Content clusters instead of single articles (structure recommendation)
- Radisson now bookable via ChatGPT (practical example, large corporation)
