llms.txt for Online Stores: Place Products in AI Recommendations
When a customer asks ChatGPT "Which espresso machine under $500 has good reviews?", the answer may name your store – or a marketplace that delivers its product data cleanly. Whether you are even a candidate depends less on the file extension than on the quality of your product data: price, availability, brand, article number, features – structured and machine-readable. This article shows what online stores actually need to do, where the typical gaps are, and how to tell whether it is working.
🏪 Who is this article for?
For operators of online stores on WooCommerce, Shopware, Gambio, Shopify, Magento or PrestaShop who want to improve their visibility in ChatGPT, Claude and Perplexity.
Why llms.txt is Especially Important for Stores
Online stores have a decisive advantage over blogs or service providers: structured product data. Modern store systems automatically generate Schema.org markup with price, rating, availability and brand. The product drill-down uses exactly this data – and passes it to AI assistants in machine-readable form.
The result: AI can not only say "check your store" but specifically: "Model X costs $249, has 4.8 stars from 213 reviews and is in stock." This difference determines whether a AI answer turns into a click.
What an AI System Needs to Know About a Product
The most common reason stores are missing from AI answers is not missing files but incomplete product data. Before generating an llms.txt, check whether your product pages deliver these details in structured form (Schema.org Product/Offer):
- Price with currency and, for promotions, the regular price – a price without a currency is a number without meaning to a language model.
- Availability as
InStock,PreOrderorOutOfStock– AI systems are reluctant to recommend something that cannot be delivered. - GTIN/EAN and brand – only with these can a system match your offer to the same product it knows from other sources and compare prices.
- Variants (size, colour, version) as separate offers or clearly named options – "not available in blue" is information that comes up in buying advice.
- Description with features instead of marketing copy. "For portafilter machines, 15 bar, 1.8 l tank, 58 mm basket" answers buyer questions; "enjoyment at the highest level" answers none.
- Reviews with count and average – one of the few signals that appear in almost every recommendation answer.
If any of this is missing, the Schema.org Checker shows the gap; most shop systems deliver the basics out of the box but lose details through themes, plugins or disabled review modules.
4 Steps to an AI-Visible llms.txt
Enter Store URL and Check CMS
Open the generator, enter your store URL. The system automatically detects your CMS (Shopware, WooCommerce, Gambio etc.) and the sitemap structure.
Configure Drill-Down
In the next step, choose the crawl depth: how many product pages should be analyzed? For a start, 50–100 products are sufficient. For complete visibility, all products are recommended (Pro plan: up to 2,000 URLs).
Download and Upload Files
Download the ZIP. The files llms.txt and llms-full.txt belong in the root directory of your store (next to robots.txt and sitemap.xml).
Update robots.txt
Insert the included robots-llms.txt snippet into your robots.txt – so AI crawlers like GPTBot and ClaudeBot know your store is open.
What the Product Drill-Down Extracts
The drill-down reads the Schema.org data for each product page. These are typically:
- Product name and description (for semantic matching)
- Price and currency (for price comparison queries)
- Rating average and number of reviews
- Availability: InStock, PreOrder, OutOfStock
- Brand and manufacturer
- EAN/GTIN (for unique product identification)
- Product category and breadcrumb
- Product image URL (AI shows your image, not a stock photo)
Automatic Comparison Tables
A unique feature for stores: the generator automatically creates comparison tables for your product categories. These tables are delivered exclusively in llms-data.json (not in llms.txt or llms-full.txt) and require Schema.org Product markup with price. Available from Pro Solo. When ChatGPT is asked about "Best espresso machines under $500", it can retrieve a structured comparison of prices, ratings and availability from your llms-data.json.
This significantly increases the probability that your store appears by name in the answer – instead of generic recommendations for Amazon or well-known brand stores.
📈 Pro Tip: Regular Updates
Prices and availability change. For stores with changing assortments, we recommend monthly updates of the llms.txt. The generator only needs 2–3 minutes for this. Pro users can automate generation via the API.
🛒 Analyze Your Store Now
Enter URL – CMS, products and prices are automatically detected. Free, no account needed.
Create Store llms.txt →Category Pages: The Underrated Lever
Buyer questions to AI systems are rarely product-specific. "Which espresso machine for beginners?" is a category question, and the best answer to it is a well-described category page – not a single product. In most stores, however, category pages are just product grids without text. For the llms.txt this means: every category needs a description that says who the products are for, what distinguishes them and which price range they fall into. Two to four sentences are enough; the generator carries them into the category section. Without that text, the file only says "espresso machines" – and no system can start a buying consultation from that.
How to Tell Whether It Is Working
Honestly: there is no report that shows you how often ChatGPT has mentioned your store. What exists are three layers that together give a reliable picture:
- Fetches. Do GPTBot, ClaudeBot, PerplexityBot and OAI-SearchBot fetch your llms.txt and product pages? You only see that in the server logs or in AI Bot Tracking – your web analytics does not see these crawlers because they do not execute JavaScript. A fetch by
ChatGPT-Usermeans: a human has just clicked on you in an answer. - Sample. Ask ten buyer questions from your category to ChatGPT, Perplexity and Claude – logged out, so personalisation does not distort the picture – and note whether your store appears, in which position, and who else is named. Repeat monthly. It is not a statistic, but it is the most honest signal there is.
- Referrers. Visitors arriving from
chat.openai.com,perplexity.aiorclaude.aiare real clicks from AI answers. The number is small and underestimates visibility – most citations bring no click – but it is the one value directly linked to revenue.
More in the guide Does llms.txt work? What can be measured.
Frequently Asked Questions for Online Stores
WooCommerce, Shopware (5 & 6), Gambio, Shopify, Magento and PrestaShop are fully automatically detected and analyzed. Prerequisite: Schema.org Product markup is active, which is the default with these systems.
For frequently changing prices or inventory, we recommend weekly to monthly updates. The generator only needs 2–3 minutes. Pro users can automate via API.
The generator still creates a working llms.txt from sitemap and metadata. The drill-down won't deliver prices/ratings without Schema.org though. Our Schema Score tool shows how to add Schema.org markup.