An llms.txt file for an e-commerce store is a short, structured guide that tells AI systems which pages matter: your best categories, your flagship products, your shipping and returns policies, and your buying guides. AI shoppers already ask ChatGPT, Claude, and Perplexity for product recommendations. When those systems can read a clean llms.txt file, your products are easier to find, describe, and cite. When they cannot, they guess from whatever pages they happen to crawl. This guide shows how to write an llms.txt file for an online store, what to include, and the mistakes that make one useless.
Start with how AI shopping actually works. A customer asks an AI assistant for a recommendation: the best running shoes under $100, a gift for a coffee lover, a laptop bag that fits a 16-inch laptop. The assistant needs facts: product names, prices, key features, availability, shipping times, and return terms. If it can fetch those facts from your site quickly, it can recommend your products with confidence. If it has to wade through navigation menus, pop-ups, and faceted category pages, it may skip your store entirely or, worse, invent details that are wrong. An llms.txt file is your way of handing the assistant a clean, prioritized map instead of leaving it to wander.
What llms.txt actually is
The llms.txt specification, published at llmstxt.org, defines a simple Markdown file that lives at the root of your site, at yoursite.com/llms.txt. It starts with an H1 containing your site name, followed by a short blockquote summary of what the site offers. Then it has H2 sections with lists of links, each link followed by a one-line description. That is the whole format. There is no permissions section, no training-policy field, and no special e-commerce extension. Its power comes from its simplicity: a small, fast file that any AI system can fetch and understand.
The spec also defines an optional companion file, llms-full.txt, which contains the full text of the linked pages in one document. For most stores, llms.txt alone is enough. A full-text file for a store with thousands of products would be enormous and stale within days. Link to your live pages instead, and let the AI system fetch the current price and availability when it needs it.
One thing llms.txt does not do: it does not control crawling. It is a guide, not a gate. Robots.txt decides which bots may visit. Llms.txt tells the bots you allow where to look. The two files work as a pair, and a store that runs both has a coherent AI policy: block the trainers, guide the answer engines.
Why e-commerce stores benefit more than most sites
Three features of online stores make llms.txt especially valuable. First, stores have deep, messy structures. A catalog with hundreds of categories, filters, and sorting options creates thousands of near-duplicate URLs. Faceted navigation alone can generate more URLs than a crawler will ever read. An llms.txt file cuts through that noise by naming the handful of category pages that actually matter.
Second, purchase decisions depend on facts that change. Prices, stock levels, shipping times, and promotions move constantly. An AI assistant that quotes a stale price from a cached page gives the customer a bad experience and your support team a ticket. Llms.txt should never contain prices or stock claims. It should point to the live product pages where the current facts live. The file stays stable while the prices change underneath it.
Third, trust decides the sale. An AI recommendation that links to your shipping policy, your returns page, and your warranty terms is far more persuasive than one that just names a product. Including your policy pages in llms.txt makes it easy for AI systems to answer the follow-up questions shoppers actually ask: how fast is shipping, what if it does not fit, and how do I return it.
What to include in a store's llms.txt file
A store llms.txt file needs seven sections. Keep each link's description to one plain sentence. Resist the urge to write marketing copy. The audience is a machine that needs facts, not a shopper you are trying to excite.
1. Store overview. The H1 and blockquote should say what you sell, who it is for, and your price range in plain terms. Example: a store selling trail-running gear for beginners and intermediate runners, with most products between $40 and $180.
2. Top categories. Link to your five to ten most important category pages. These are the pages shoppers start from: men's running shoes, women's running shoes, hydration packs, trail accessories. Skip thin categories with two products and skip filter and sort URLs entirely.
3. Flagship and best-selling products. Link to your ten to twenty best products, the ones you would show a first-time visitor. These are the products most likely to be recommended, so they deserve a place in the file. Rotate the list as your bestsellers change, at least once a quarter.
4. Buying guides and comparisons. Link to your guides: how to choose trail shoes, waterproof versus water-resistant jackets, size and fit guides. Guides are the content AI systems love to cite when a shopper asks an open-ended question. They also keep the assistant on your site instead of sending the shopper to a competitor's guide.
5. Policy pages. Shipping, returns, exchanges, warranty, and privacy. Link all of them. Shoppers ask about these constantly, and AI systems that can answer from your own policy pages give answers you control.
6. About and trust pages. Your about page, customer reviews or testimonials page, and any certifications or guarantees. AI systems use these to judge whether a store is legitimate. A store with no about page looks thin to a machine, just as it does to a human.
7. Support and contact. Your contact page, FAQ page, and size-chart or fit-help pages. When an assistant can point a shopper to a real FAQ instead of guessing, the answer is better and your support load is lower.
A worked example
Here is a compact llms.txt file for a fictional trail-running store. Notice what it leaves out: no prices, no stock claims, no filter URLs, no marketing adjectives.
# Trailhead Outfitters
> Trail-running shoes, apparel, and hydration gear for beginner and intermediate runners. Most products $40-$180. Ships from Colorado, USA.
## Categories
- [Men's Trail Running Shoes](https://example.com/mens-trail-shoes): Trail shoes for men, organized by cushion level and terrain.
- [Women's Trail Running Shoes](https://example.com/womens-trail-shoes): Trail shoes for women, organized by cushion level and terrain.
- [Hydration Packs](https://example.com/hydration-packs): Running vests and hydration packs from 5L to 15L.
- [Trail Accessories](https://example.com/trail-accessories): Headlamps, trekking poles, gaiters, and socks.
## Best Sellers
- [Ridgeline GTX Trail Shoe](https://example.com/products/ridgeline-gtx): Waterproof trail shoe for wet and muddy conditions.
- [Summit 10L Hydration Vest](https://example.com/products/summit-10l-vest): Lightweight 10-liter vest with two soft flasks included.
## Guides
- [How to Choose Trail Running Shoes](https://example.com/guides/choose-trail-shoes): Cushion, drop, grip, and fit explained for first-time buyers.
- [Trail Shoe Size and Fit Guide](https://example.com/guides/shoe-fit): How trail shoe sizing differs from road shoes, with a measurement guide.
## Policies
- [Shipping Policy](https://example.com/shipping): Shipping options, delivery times, and costs for US orders.
- [Returns and Exchanges](https://example.com/returns): 60-day returns and free size exchanges.
- [Warranty](https://example.com/warranty): Two-year warranty on manufacturing defects.
## About and Support
- [About Us](https://example.com/about): Who runs the store and how products are tested.
- [Customer Reviews](https://example.com/reviews): Verified buyer reviews by product.
- [Help and FAQ](https://example.com/help): Sizing, orders, shipping, and returns questions.
- [Contact](https://example.com/contact): Email and phone support hours.That file is under forty links and readable in seconds. A store with 5,000 products does not need 5,000 links in llms.txt. It needs the fifty links that answer ninety percent of shopper questions.
How to add llms.txt to your store
On most platforms, adding the file is simple. On Shopify, upload it as a static asset or serve it through a page template, then confirm it loads at your root domain. On WordPress with WooCommerce, place the file in your site root or use an SEO plugin that supports custom root files. On a custom-built store, add it as a static file served at /llms.txt. If you are starting from scratch, build the file with a free llms.txt generator and then edit the output to match the seven sections above.
After publishing, check three things. First, open yoursite.com/llms.txt in a browser and confirm it loads with the right content and no HTML wrapper. A file served as an HTML page instead of plain text will confuse parsers. Second, click every link. A dead link in llms.txt is worse than no link, because it sends AI systems to an error page and signals that the file is unmaintained. Third, confirm the file is not blocked in robots.txt. Some stores block all text files by accident with an over-broad disallow rule. Your llms.txt file must be crawlable, or no AI system will ever read it.
Keep it fresh: the maintenance habit
An llms.txt file rots faster on a store than on any other kind of site. Products get discontinued, categories get renamed, and seasonal lines come and go. A stale file points AI systems at dead pages and missing products, which is worse than having no file at all.
Set a quarterly review. Remove discontinued products and dead links. Add new bestsellers and new guides. Check that the policy pages still match the current terms. If your store runs seasonal collections, update the file when the season changes. The review takes under an hour for most stores, and it is the difference between a file that helps and a file that misleads.
One shortcut helps: keep the file short. A file with fifty links is easy to review. A file with five hundred links is a second catalog that nobody will maintain. Short and current beats long and stale every time.
What llms.txt does not replace
Three common misunderstandings deserve a clear answer. First, llms.txt does not replace product structured data. Google's product structured data is what powers rich results, price display, and availability in search. Llms.txt is for AI systems; structured data is for search engines. Run both.
Second, llms.txt does not replace your sitemap. A sitemap tells search engines about every page on your site. Llms.txt tells AI systems about the pages that matter most. Different audiences, different jobs.
Third, llms.txt does not guarantee AI citations. It makes your content easier to find and understand, which improves your odds. No file can force an AI system to recommend your products. The products still have to be good, the prices still have to be competitive, and the guides still have to be genuinely useful. Llms.txt removes friction. It does not create merit.
Mistakes that make a store llms.txt file useless
Listing every product. A store with thousands of products cannot fit them in a useful file. Curate the best sellers and let the category pages do the rest.
Putting prices or stock in the file. Prices change daily. A file that quotes a price becomes wrong within days, and wrong facts destroy trust. Link to live pages instead.
Including filter, sort, and search URLs. These URLs are near-duplicates that waste the reader's time. Link to clean category pages only.
Writing marketing copy instead of descriptions. "The best trail shoe ever made, loved by thousands" tells an AI system nothing. "Waterproof trail shoe for wet and muddy conditions" tells it everything.
Forgetting policy pages. Shipping and returns are the questions shoppers ask most. Leaving them out means AI systems answer from guesswork or from a competitor's policy.
Letting the file go stale. A file that still lists a discontinued product line is a sign the store does not maintain its own information. Review it quarterly.
Blocking the file in robots.txt. Check that your robots.txt allows crawling of /llms.txt. A blocked file helps nobody.
FAQ
Does llms.txt help my store rank on Google?
Not directly. Google has said llms.txt is not a ranking factor. Its value is with AI systems: it helps them find, understand, and cite your products. Treat it as an AI-discovery tool, not an SEO ranking tool.
Should a small store with 20 products use llms.txt?
Yes. Small stores benefit most, because the file is quick to write and easy to keep fresh. A twenty-product store can list every product in its llms.txt file and be done in an afternoon.
Should I put prices in my llms.txt file?
No. Prices change too often. Link to the live product pages where the current price and availability are shown. Keep the file free of facts that go stale.
Does llms.txt replace my XML sitemap?
No. A sitemap is for search engines and lists every page. Llms.txt is for AI systems and lists the pages that matter most. Run both, and keep both current.
Can llms.txt stop AI crawlers from training on my product descriptions?
No. Llms.txt is a guide, not a gate. To opt out of AI training, use robots.txt rules for training crawlers, and check each platform's own opt-out controls. Llms.txt tells allowed bots where to look; it cannot block anything.
How often should I update my store's llms.txt file?
Review it at least once a quarter, and update it whenever you add a new bestseller, discontinue a product line, or change your shipping or returns terms. A short file makes the review fast.
