# How AI-ready are independent hotels and restaurants? 150 US sites (2026-11-21) · EarnCitations

> We audited 150 independent US hotel and restaurant websites drawn from Wikidata. 44 no longer load or no longer show the business; of the rest, 17 of 39 hotels answer

Source: https://earncitations.com/research-local.html

> Every page on this site: https://earncitations.com/llms.txt (the whole site as one file: https://earncitations.com/llms-full.txt)

Research 2026-11-21

## How AI-ready are independent hotels and restaurants?

When a guest asks ChatGPT what time check-in is, or whether a restaurant is open late, the answer comes from what an AI crawler could read. We audited 150 independent US hotel and restaurant websites, drawn from Wikidata so nobody picked them, to see what that is.

### What we found

- 39% of the official websites gave our crawler nothing to read about the business. 36 didn't load for it (a domain that no longer resolves, a certificate that doesn't match, a refused connection, a timeout), 14 answered with a bot challenge (AI crawlers may or may not be let through), and 8 loaded but no longer name the business. Wikidata still lists all of them as the official website, and it is a public source that search engines and AI models can draw on to tell which business is which.

- "What time is check-in?" We asked our [question doctor](https://earncitations.com/question.html) for each of the 39 hotels we could read. 17 answer it plainly: a heading or FAQ entry for it, with the time in the sentence below. 12 mention check-in but don't answer, for example a "Check-in" section about ID requirements, or the time only in structured data. 10 have no page that covers it, so an assistant has to find the time somewhere else, such as a booking site, or go without.

- "What are your hours?" Of 32 restaurants, 16 answer plainly, 15 weakly (hours in a footer with no heading, under a heading whose first line is something else, or loaded by a widget so the server HTML has no times), and 1 only after JavaScript runs.

- JavaScript-only pages: 10 of 42 restaurant sites and 4 of 50 hotel sites send almost no text without JavaScript, and most AI crawlers don't run it.

- Blocking AI isn't the problem: 0 of 92 readable sites block an AI search crawler in robots.txt.

### Can an AI crawler read the site at all?

| | Sampled | Didn't load | Bot challenge | Not the business any more | Readable | Median score |
| --- | --- | --- | --- | --- | --- | --- |
| Hotels | 75 | 16 | 7 | 2 | 50 | 74 |
| Restaurants | 75 | 20 | 7 | 6 | 42 | 65.5 |

Median score is over readable sites, out of 100, from the same [audit](https://earncitations.com/#audit) anyone can run.

### Does the site answer what guests ask?

One question per site, asked of every readable site our audit recognised as a hotel (39) or a local restaurant (32). The question doctor reads the homepage, robots.txt, sitemap and up to six pages most likely to answer, the way an AI crawler does, and gives one verdict.

| Question | Sites | Answered plainly | Weak page | No page covers it | Needs JavaScript |
| --- | --- | --- | --- | --- | --- |
| Hotels: "what time is check-in?" | 39 | 17 | 12 | 10 | 0 |
| Restaurants: "what are your hours?" | 32 | 16 | 15 | 0 | 1 |

Answered plainly: a heading or FAQ entry that names the question, and the time in the first sentence under it. Weak: the page covers the topic but the time is missing from that sentence, only in structured data, or there's no heading for it.

### The rest of the audit

Over readable sites. Hours rows count sites our audit recognised as a hotel (39) or a local restaurant (32); the audit reads fewer pages than the question doctor, so its counts are lower bounds.

| Check-in times (hotels) or opening hours (restaurants) as text or schema, on the pages the audit reads | | | |
| --- | --- | --- | --- |
| Hotels | | 21 of 39 | |
| Restaurants | | 29 of 32 | |
| …in both text and schema | | | |
| Hotels | | 7 of 39 | |
| Restaurants | | 8 of 32 | |
| A rooms page (hotels) or menu or price list (restaurants) linked from the homepage | | | |
| Hotels | | 31 of 50 | |
| Restaurants | | 25 of 42 | |
| Business markup (Hotel, Restaurant, LocalBusiness or Organization JSON-LD) | | | |
| Hotels | | 37 of 50 | |
| Restaurants | | 21 of 42 | |
| Name, address and phone in markup | | | |
| Hotels | | 19 of 45 | |
| Restaurants | | 9 of 32 | |
| Among sites recognised as local businesses. | | | |
| Prices in markup (priceRange, menu or Offer) | | | |
| Hotels | | 7 of 50 | |
| Restaurants | | 7 of 42 | |
| Text only after JavaScript runs | | | |
| Hotels | | 4 of 50 | |
| Restaurants | | 10 of 42 | |
| Publishes /llms.txt | | | |
| Hotels | | 16 of 50 | |
| Restaurants | | 5 of 42 | |

### Sites that no longer show the business

Each is the official website Wikidata lists for the business in brackets, and none of them names that business: thegouldhotel.com (The Hotel Clarence); mcmenamins.com (Kennedy School); moonriverbrewing.com (Moon River Brewing Company); theoryatomsi.com (Theory); ardoisesf.com (L'Ardoise Bistro); benjys.com (Benjy's); kermitsbg.com (Kermit's); historicpalace.com (The Palace Restaurant and Saloon). If one is yours, update Wikidata's official website (P856) or its name, and check your Google Business Profile links to the right site.

### What to do if you run a hotel or restaurant

- Check your domain still works and that Wikidata, Google Business Profile and the booking platforms link to it.

- Put the guest facts in plain text: check-in and check-out times, opening hours, parking, pets, reservations. Then add them to your Hotel or Restaurant JSON-LD. The [fix kit](https://earncitations.com/developers.html#fix) drafts the markup from what your pages already say.

- Link a rooms or menu page from the homepage, as HTML rather than an image or PDF only.

- Ask the questions guests ask in the [question doctor](https://earncitations.com/question.html) to find the weak page.

### Method

- Sample. Wikidata items that are a hotel (Q27686) or restaurant (Q11707), or a subclass of either, in the United States, with an official website (P856), and with no dissolution (P576) or closure (P3999) date. The website had to be a homepage on the business's own domain. We left out domains listed for two or more items, and items Wikidata classes as a restaurant chain, chain store or hotel chain (312 items). The rest were ordered by SHA-256 of the seed earncitations-local-2026-11 and the Wikidata ID, and we took the first 75 of each kind. The queries, their results and the code are in the repository (tools/research/local-sample.mjs, tools/research/runs/local-2026-11-22/).

- Crawl. One audit per site between 2026-11-21 17:58 and 2026-11-21 18:01 UTC, with our identified user agent ([EarnCitationsAudit](https://earncitations.com/bot.html)), honouring robots.txt, without running JavaScript: the homepage, robots.txt, sitemap, llms.txt and up to six key pages.

- Guest questions. One run of the [question doctor](https://earncitations.com/question.html) per readable hotel ("what time is check-in") and restaurant ("what are your hours"), on 2026-11-21, with the same user agent. Verdicts and our own findings only are published, no page text.

- Still the business? Compared the Wikidata English label's distinctive words with the site's own name, title and description (never its domain, which a new owner keeps).

- Limits. One crawl from one place at one time. Wikidata favours notable places. Our checks read what a non-JavaScript crawler reads; some sites show more in a browser. "Didn't load" includes certificate errors a browser might also refuse.

Data: [research-local.json](https://earncitations.com/research-local.json) (hosts, Wikidata IDs, check results; no page text). Our software study: [124 software companies](https://earncitations.com/research.html).

### Questions

#### Which businesses are in the sample?

Hotels and restaurants in the United States that Wikidata lists with an official website and doesn't mark as closed: 1545 hotels and 1804 restaurants when we queried it. We left out chains (a domain listed for two or more businesses, or a business Wikidata classes as a chain), ordered the rest by a SHA-256 hash of a fixed seed and the Wikidata ID, and took the first 75 of each. Nobody picked the sites.

#### Why Wikidata?

It is public and reproducible, and it is a public source that search engines and AI models can draw on to tell which business is which. It also leans toward notable places (historic hotels, well-known restaurants), so these are likely better-resourced than the average independent business.

#### What does "doesn't show the business any more" mean?

The site loaded, but what it calls itself (its schema or og name, title and description) shares fewer than half of the distinctive words in the business's Wikidata name or any of its aliases. When we looked, most were lapsed domains now showing something unrelated (gambling sites, a sportsbook, a shop); two were a restaurant group's homepage rather than the restaurant's, and one hotel used a different name from Wikidata's. We list them below so you can check.

#### Do the hours numbers mean a restaurant has no opening hours on its site?

They mean our crawler found none on the pages it read, without running JavaScript: the homepage, contact, about, FAQ and a few others. Hours shown only in an image, a script-loaded widget or a PDF menu are invisible to most AI crawlers too, which is the point.

#### Can I check my own hotel or restaurant?

Yes: the [free audit](https://earncitations.com/#audit) runs the same checks, and the [question doctor](https://earncitations.com/question.html) checks one guest question ("what time is check-in?", "are you open late?").
