Service · AI visibility
How do AI assistants see your hotel, and what do we measure?
Travelers also ask AI assistants for hotel recommendations. The visits from this path are still small; they increased in a panel's spring 2026 measurement. The assistant gives a short list and whether your hotel is in it can be measured. We carry out two measurements: whether your site is open to AI crawlers, and your visibility in assistant answers in the assessment report we write for your hotel. We do not decide what any assistant recommends; we measure and write what we see as it is.
- What we do
We measure the access of 16 AI crawlers to your site. In the assessment report we test your hotel with 8 questions across 4 assistants and write the result as one item of a 100-point breakdown.
- What changes at the hotel
You learn by measurement, not by guess, whether your site is open to AI crawlers. You see which five items the score in the report comes from, one by one.
- Possible benefit
A crawler access that was accidentally closed may be noticed and fixed. Whether and how the assistants mention your hotel becomes visible.
Request an AI findability report
How much visit comes from AI?
The two figures below show today's volume. Both are figures not from your hotel but from broad samples.
- 0.90%
The share of total sessions from AI referrals at the end of May 2026 in a panel of more than 17,000 hotel sites. Weekly AI sessions roughly doubled from mid-March to the end of May.
Nicolas Sitter, The Hotels Network panel analysis, 2026 - Fewer than 1 in 100
According to Booking Holdings, in the second quarter of 2026 the share of referrals from ChatGPT, Gemini, and similar models within 100 room-nights.
Travelers Today, Booking Holdings 2026 second quarter interview
These figures show that traffic from AI is still small today. Because it changes quickly, read the figures with their dates.
Who are the AI crawlers and what do they do?
AI companies send different crawlers to sites for different tasks. According to the publishers' documentation there are three main tasks; we look at all 16 crawlers.
- Search crawlers (for example PerplexityBot and ChatGPT's search crawler): they read your pages to show your site in the assistant's search feature.
- User request crawlers (for example ChatGPT-User and Perplexity-User): they open your page at that moment when a user asks the assistant a question. Some publishers note that robots.txt rules may not apply to these crawlers.
- Model training crawlers (GPTBot, ClaudeBot): they collect content that may be used in training the models.
How do we measure access?
The measurement shows whether your site opens the door to each crawler.
- 1
Crawler identity
We send a request to your site with the identity string that each crawler's publisher documents.
- 2
Response code
We look at whether the site responds with 200. 200 shows that the page is open to that crawler.
- 3
Reason if blocked
When we see a response other than 200 (for example 403), we look for the reason: a robots.txt rule or a security layer in front of the site may be the cause.
- 4
Re-measurement
After the fix, we run the same measurement again.
What does this measurement not show?
The response code shows whether the site opens the door to the crawler. It does not show that a crawler actually visited your site, or that an assistant used your page in an answer. We measure that separately: in the assessment report we ask questions to the assistants and look at the answers.
What do llms.txt and structured data do?
We publish an llms.txt file at the root of your hotel site. The file describes in plain text what the site is and which pages it consists of. On the pages we also write the information of the hotel, room types, and frequently asked questions with structured data. We produce both from the same input as the page text.
Google's documentation says that no new machine-readable file, AI text file, or special markup is needed to appear in AI summaries. For this reason we publish llms.txt as a regular addition, not as a condition. We cannot measure whether the assistants read the file and use it in answers, or whether structured data changes the answers.
How is the score formed in the assessment report?
The score only appears in the assessment report. The report measures your hotel on five items and writes the total out of one hundred.
If an item cannot be measured, we drop that item's score from the maximum of the total; we do not write an estimated score for an item we cannot measure.
- Appearance in AI answers (30): we ask 8 questions through your hotel's name and your area to 4 assistants. The share of answers in which your hotel's name or your site appears is converted to 30 points.
- Hotel search ranking in the city you are in (20): as your site's rank rises in hotel searches on Google, the score increases.
- Site technical audit (20): secure connection, mobile view, title and description, structured data, language pairs, llms.txt, robots.txt, sitemap, and booking link are audited; the share of passed audits is converted to a score.
- Google Business Profile (15): the profile is claimed, we have written the site address, and there is a sufficient number of photos.
- Review (15): your Google score compared with the average of hotels in your area.
Limits of measuring assistant answers
An assistant does not give the same answer to the same question every time; the answer changes by time, by user, and by the version of the assistant. The 32 answers from 8 questions across 4 assistants are a small sample, not a photograph. Part of the difference between two measurements may come from this variability.
The report also compares the information given by the assistants with the information on your Google profile; when we see conflicting information, we write it as a separate line in the report. We do not decide which hotel the assistants recommend; we do not promise ranking or being mentioned.
Sources
Frequently asked questions
- How do you measure the access of AI crawlers to our site?
- We send a request to your site with each of the 16 crawlers' identities and look at whether the site responds with 200. When we see a response other than 200, we look for the reason.
- What is the score and where does it appear?
- The score is the 100-point breakdown in the assessment report: appearance in AI answers 30, hotel search ranking in the city you are in 20, site technical audit 20, Google Business Profile 15, review 15. The score only appears in the assessment report.
- Does the llms.txt file get me into the answers?
- We cannot say that it does. We publish the file at the root of your hotel site, but Google's documentation says that no extra file is needed to appear in AI summaries and we cannot measure whether the assistants read the file.
- Do you undertake that our hotel will be recommended by the assistants?
- No. We do not decide which hotel the assistants recommend. We measure whether your hotel appears in answers and how it is described, and write it as it is in the report.
- What do you do when an item cannot be measured?
- We drop that item's score from the maximum of the total; we do not write an estimated score.