llms.txt for Ecommerce Brands: The Complete Guide

llms.txt is a plain-text markdown file served at your domain root that tells AI systems what your site is, what you sell, and which pages to read first. Proposed by Answer.AI's Jeremy Howard in September 2024, it is the closest thing AI search has to a curated front door.

Before Tru Commerce was Tru Commerce, our best-performing organic asset was not a blog post or a feature page. It was a free llms.txt generator. It out-pulled every article on the site. That told us two things: brands are actively searching for this, and almost nobody explains it well for commerce. This is the ecommerce-specific guide we wished existed.

What llms.txt actually is

The llms.txt proposal defines a markdown file with a fixed shape:

1. An H1 with your site or brand name. Required.
2. A blockquote summary: one paragraph on what you are and who you serve.
3. H2 sections containing link lists, each link with a one-line description.
4. An optional "Optional" section for links an AI can skip when context is tight.

That is the whole spec. It is deliberately boring: markdown because language models parse it natively, root path because tooling needs one predictable location.

Two files, two jobs:

You can see a live commerce example on this site: trucommerce.ai/llms.txt.

The honest adoption picture in 2026

Be clear-eyed about this, because most llms.txt content oversells it.

What is not proven: no major LLM provider (OpenAI, Anthropic, Google, Perplexity) has publicly committed to consuming llms.txt in ranking or retrieval. Google's John Mueller compared it to the old keywords meta tag in 2025 - a signal sites volunteer that engines may ignore.

What is verifiable:

• AI crawlers request the file. Check your server logs for /llms.txt hits from GPTBot, ClaudeBot, and PerplexityBot user agents; most stores that ship one see fetches within weeks.
• The tooling ecosystem standardized on it. Mintlify generates it for every docs site it hosts (Mintlify, 2025), and Anthropic publishes one for its own documentation.
• It costs under an hour. The expected value math is lopsided: near-zero cost, plausible upside on the surfaces that decide where AI assistants send shoppers.

Treat llms.txt the way you treated schema markup in 2015: early, cheap, and directionally aligned with where the platforms are going.

The ecommerce template

Generic llms.txt files list docs pages. A store's file should lead with the facts AI assistants most often get wrong: pricing, availability, shipping, and returns. Hallucinated policy answers are a real cost; an AI telling a shopper you offer free returns when you do not creates a support ticket and a chargeback risk.

# Your Brand

> One-sentence positioning: what you sell, for whom, at what price point.
> Ships from [country]; delivers to [markets]. Founded [year].

## Catalog
- [All collections](https://store.com/collections): 240 SKUs across 6 categories
- [Bestsellers](https://store.com/collections/bestsellers): top 20 by 90-day volume
- [Product feed](https://store.com/feed.xml): structured data, refreshed daily

## Policies
- [Shipping](https://store.com/policies/shipping): rates, cutoffs, delivery windows
- [Returns](https://store.com/policies/returns): window, condition rules, refund timing
- [Size guides](https://store.com/pages/sizing): per-category measurements

## Company
- [About](https://store.com/pages/about): founding story, manufacturing, certifications
- [Reviews](https://store.com/pages/reviews): aggregate rating and count

## Optional
- [Blog](https://store.com/blogs/news)

Three rules that matter more than the markup:

1. Put numbers in the descriptions. "240 SKUs across 6 categories" gives a model something to repeat. "Our products" gives it nothing.
2. Link canonical URLs only. No UTM parameters, no redirects. Models cite what they fetch.
3. Regenerate on catalog changes. A stale llms.txt that contradicts your live site is worse than none; wire it into your deploy or feed refresh.

llms.txt is step one of three

The file only works as part of a chain:

1. Access. Your robots.txt must allow the retrieval crawlers (OAI-SearchBot, ClaudeBot, PerplexityBot). A perfect llms.txt behind a blocked bot does nothing. Our agentic readiness scanner checks this in about a minute.
2. Curation. The llms.txt itself, per the template above.
3. Measurement. Whether any of it moves your presence in real AI answers. That is a Citation Rank question: which queries surface you, which surface competitors, and how that trends. Amazon India ran this loop and moved Share of Voice +0.98% and Top-5 Presence +2.4% in six weeks across six categories - the full playbook is in the Amazon India case study.

Ship the file this week. Then measure whether the machines noticed - get a free Citation Rank scan and see where you stand in 24 hours.