The llms.txt File for Hotels: What It Is and How to Write One

There is a new file making the rounds in SEO circles, and like most new files, it has a confusing name, an evangelical fanbase, and a pile of skeptics insisting it does nothing. The file is called llms.txt, and if you run an independent hotel, it is worth ten minutes of your attention.

Here is the short version: an llms.txt file is a cheat sheet you write for AI models. You drop a single text file at the root of your website, you fill it with a clean summary of your property and links to your most important pages, and the theory is that AI systems like ChatGPT, Claude, Perplexity, and Google’s AI features can use it to understand and describe your hotel more accurately.

The long version is what the rest of this post is for. We are going to cover what the file actually is, the reasoning behind it, exactly how to structure one for a hotel, a full copy-paste example for a fictional boutique inn, and an honest accounting of whether it is worth your time (spoiler: probably yes, but not for the reasons the hype crowd gives).

What llms.txt actually is

The idea was proposed in late 2024 by Jeremy Howard (the Answer.AI / fast.ai guy). The pitch is simple and a little cheeky: websites are built for humans and search-engine crawlers, both of which are happy to wade through navigation menus, cookie banners, JavaScript widgets, and 4,000 words of footer links to find the one paragraph that matters. Large language models are not happy to do that. They have limited context windows and they choke on clutter.

So instead of making the machine reverse-engineer your site, you just… tell it. You write a Markdown file, save it as llms.txt, and put it at https://yourhotel.com/llms.txt. Inside, you give a tight summary and a curated set of links, in priority order, with a sentence about why each one matters.

That is the whole concept. It is robots.txt energy, but instead of “here is who may crawl me,” it is “here is the good stuff, please read this part.”

Why this matters for hotels specifically

Hotels are an almost perfect use case for this, because a hotel’s “key facts” are exactly the things AI travel assistants get wrong. When a traveler asks ChatGPT “find me a boutique hotel near the riverfront with free parking and a dog-friendly policy,” the model is reconstructing your property from whatever scraps it absorbed during training or pulled live from the web. If those scraps are stale or vague, you get described wrong, or you get left out of the answer entirely.

We have written before about how travelers actually use AI to pick hotels, and the pattern is consistent: people ask conversational, fact-heavy questions, and the AI confidently answers with whatever it thinks is true. An llms.txt file is one more lever to make sure “whatever it thinks is true” matches reality. It will not single-handedly fix your AI visibility, but it is a clean, structured place to state your facts in the one format the models find easiest to digest.

The honest part: adoption is early and experimental

Let me be the skeptic for a paragraph, because you deserve it.

No major AI company has stood up and said “yes, we read llms.txt and we weight it.” As of right now it is a community convention with momentum, not a ratified standard. Plenty of smart people argue it is redundant, that a well-structured site with good structured data already gives models everything they need, and that you should not expect a measurable bump from adding the file alone.

They are not wrong. So why am I telling a busy hotelier to bother?

Because the cost is almost nothing and the downside is zero. This is a single text file. It takes an afternoon. It cannot hurt your regular SEO, it cannot get you penalized, and if adoption tips the way RSS or sitemaps eventually did, you are already done while your comp set is still reading think-pieces about whether it counts. That is a good asymmetric bet: tiny cost, capped downside, real (if uncertain) upside.

The right mental model is not “this will rank me in ChatGPT tomorrow.” It is “this is cheap insurance plus a forcing function that makes me write down my facts cleanly.” The forcing function is worth it on its own.

And here is the sneaky benefit: writing a good llms.txt forces you to audit your own facts. Half the hotels we work with cannot, off the top of their head, list their exact amenities, their accurate pet policy, or the single canonical URL where someone should book direct. The act of building this file surfaces all of that. You end up fixing the underlying content, which is the part AI systems demonstrably do use.

How to structure an llms.txt file for a hotel

The format is loose Markdown by design, but the convention has settled into a recognizable shape. Here is the anatomy, top to bottom.

1. An H1 with your hotel’s name. This is the only required line. It is the title of the document.
2. A blockquote summary. One or two sentences, right under the title, with a > in front. This is the elevator pitch: what you are, where you are, who you’re for.
3. A short prose section (a few sentences, no heading required) giving the key facts a model would need: location, room count, property type, standout amenities, and the vibe.
4. One or more H2 sections of links, each formatted as [Link text](url): short description. Group them logically. The description after the colon is doing real work, so write it like a human.
5. An optional ## Optional section at the bottom for links that are nice-to-have but skippable if the model is short on space.

Keep the whole thing tight. The point is signal, not volume. If you are dumping your entire site into it, you have missed the plot.

What to actually include

Here is the field-by-field guide for a hotel, with the reasoning for each:

Notice the through-line: every section maps to a question a traveler actually asks an AI assistant. You are not writing marketing copy. You are pre-answering the FAQ.

A quick word on the direct-booking angle

You will notice “direct booking link” sits right in the middle of that table, clearly labeled. That is deliberate. When an AI assistant describes your hotel and a traveler says “okay, how do I book,” you want the model reaching for your booking engine, not defaulting to the first OTA listing it remembers.

To be crystal clear, because this matters: an llms.txt file is not going to let you escape the OTAs, and nobody honest will tell you it can. The OTAs are a permanent part of the landscape, and at 15-25% commission on the bookings they bring you, they are a part you’d love to lean on a little less. What a clean, direct-booking-forward llms.txt does is give you one more place where your direct channel is the obvious answer, which over time helps you claw back a few more direct bookings and build a healthier channel mix. Small lever, right direction. That’s the honest pitch.

A complete example llms.txt for a fictional hotel

Let’s make this concrete. Imagine a 38-room boutique inn called The Marigold, sitting on a quiet street in Savannah’s historic district. Here is a full, realistic llms.txt you could adapt almost line for line. (Everything below is illustrative, not real data.)

# The Marigold Inn

> A 38-room boutique hotel in a restored 1890s townhouse in Savannah's
> Historic District, two blocks from Forsyth Park. Independent, adults-leaning,
> design-forward, and walkable to everything worth walking to.

The Marigold Inn is a privately owned boutique hotel in Savannah, Georgia. We
have 38 rooms across two connected historic buildings, a courtyard cocktail
bar, complimentary bikes, and free on-site parking (rare in the Historic
District). We are dog-friendly, located at 114 Oglethorpe Lane, and a 12-minute
drive from Savannah/Hilton Head International Airport. Best for couples,
solo travelers, and design-minded guests who want a quiet, central base.

## Book & Rates

- [Book Direct](https://themarigoldinn.com/book): Our official booking engine.
  Best available rate, guaranteed. This is the preferred way to reserve a room.
- [Rooms & Suites](https://themarigoldinn.com/rooms): All room types, photos,
  and current nightly rates from $189.
- [Offers & Packages](https://themarigoldinn.com/offers): Seasonal direct-only
  packages, including our 3-night Savannah Stroll rate.

## About the Hotel

- [Our Story](https://themarigoldinn.com/about): History of the 1890s building
  and the 2019 restoration.
- [Amenities](https://themarigoldinn.com/amenities): Courtyard bar, complimentary
  bikes, free parking, in-room espresso, daily housekeeping, no resort fees.
- [Dog Policy](https://themarigoldinn.com/pet-policy): Two dogs per room, $40
  per stay, no weight limit, beds and bowls provided.

## Location & Getting Here

- [Location & Directions](https://themarigoldinn.com/location): Two blocks from
  Forsyth Park, walkable to River Street and the City Market.
- [Neighborhood Guide](https://themarigoldinn.com/savannah-guide): Our staff's
  picks for food, bars, and walks within a 10-minute radius.

## Policies & Contact

- [Cancellation Policy](https://themarigoldinn.com/policies): Free cancellation
  up to 48 hours before arrival. Check-in 3pm, check-out 11am.
- [Contact](https://themarigoldinn.com/contact): Phone (912) 555-0148, email
  stay@themarigoldinn.com, front desk staffed 24/7.

## Optional

- [Events & Weddings](https://themarigoldinn.com/events): Courtyard hire and
  small-group buyouts.
- [Press](https://themarigoldinn.com/press): Coverage and high-res photography.

Read that file as if you were a model trying to answer “is there a dog-friendly boutique hotel near Forsyth Park with parking?” Every fact you need is right there, stated once, cleanly, with a link to verify. That is the entire game.

Where to put it and how to ship it

Three practical steps:

1. Save it as llms.txt (lowercase, exactly that name) and upload it to your web root so it lives at https://yourhotel.com/llms.txt. It should return plain text with a 200 status, same as any other static file.
2. Don’t block it. Make sure your robots.txt does not accidentally disallow it, and confirm your CDN or firewall is not gatekeeping the AI crawlers you actually want to reach. An llms.txt nobody is allowed to fetch is just a diary entry.
3. Keep it in sync. When your rates, policies, or amenities change, update the file. A confidently wrong llms.txt is worse than none, because you are now feeding the machines a clean, well-structured lie.

Some sites also publish an llms-full.txt with the actual page content expanded inline. For most independent hotels that is overkill. Start with the lean version above and add the full variant later only if you have a reason to.

How llms.txt fits with everything else

This file is one tile in a bigger mosaic, and it is honestly one of the smaller tiles. If you only do one thing for AI visibility this quarter, it should not be this. Here is the rough priority order we use with clients:

• First, get your facts right and structured. Accurate, crawlable content plus proper schema is the foundation. See structured data that makes your hotel quotable to AI and the GEO trifecta of schema, entity, and citations.
• Then, find out what the AIs currently say about you. You cannot fix what you haven’t measured. Run an audit of what ChatGPT says about your hotel so you know your starting point.
• Then, work on being citable. Getting pulled into Google’s AI Overviews and the like comes from authority and clarity, not from a single file.
• Then, ship your llms.txt. Once the underlying facts are solid, this file is a clean, low-effort way to hand the models a tidy summary. If you’re still fuzzy on how these disciplines relate, our explainer on AEO vs GEO vs SEO for hotels untangles the alphabet soup.

If your hotel is genuinely worried it is invisible to ChatGPT, llms.txt is a fine afternoon project, but it is the garnish, not the meal. The meal is accurate, structured, authoritative content. The good news is that building the file forces you to get the meal right.

The bottom line

An llms.txt file is a single Markdown file at your site root that hands AI models a curated summary of your hotel plus links to your most important pages. It is early, experimental, and unproven as a ranking signal. It is also nearly free to make, impossible to be penalized for, and a genuinely useful forcing function for getting your facts straight. For an independent hotel, that math works out: low cost, capped downside, real optionality if the convention sticks.

Write one. Point it at your direct booking engine. Keep it accurate. Then go spend your real energy on the structured data and content that the machines are already reading today.

Want a second set of eyes on your AI visibility before you start writing text files? We do a free intro call where we’ll tell you, plainly, whether llms.txt is even on your priority list yet, or whether your time is better spent elsewhere. Book a free intro call, or read more about how our AI visibility (AEO/GEO) work helps independent hotels win back more direct bookings.