No. There is no evidence that llms.txt gets your Shopify store cited by AI, and no major answer engine has confirmed reading it for shopping. The things that actually move AI recommendations are your product feed, your structured data, and third-party citations. llms.txt is the fashionable answer; it’s not the correct one.
What is llms.txt supposed to do?
llms.txt is a proposed markdown file — pitched as “robots.txt for AI” — that lists your site’s key pages so language models can find them. The theory: hand the model a clean map of your catalog and it recommends you more often. It’s a reasonable-sounding idea. Shopify now even serves the file natively; per the Shopify Developer Community, you can drop a templates/llms.txt.liquid into your theme and it “executes the same way as robots.txt.” Serving the file and any engine reading it are two different claims, and only the first one is confirmed.
Is there any evidence AI engines read it?
No. The largest study to date — SE Ranking’s crawl of 300,000 domains, covered by Search Engine Journal — found “no measurable link” between having llms.txt and AI citation frequency. Two damning details: removing llms.txt from their model improved its accuracy (it was adding noise, not signal), and of the 50 most AI-cited domains, exactly one had the file. A year of hype hasn’t produced a single controlled result showing it works.
Then what’s the “40%” number people cite?
The one real, peer-reviewed number in this space comes from the Princeton/Georgia Tech GEO: Generative Engine Optimization paper (arXiv 2311.09735, KDD ‘24). It found that adding citations, quotations, and statistics to a source can boost its visibility in AI answers by up to 40%.
Be honest about what that measures: it’s about the content of a source, not about llms.txt. No study measures llms.txt’s lift because there isn’t one to measure. Treat 40% as a proxy for a general principle — credible, specific, cited content wins — not as a number you can attach to a text file.
What actually moves AI citations vs. what doesn’t
| Lever | Evidence it works | Effort |
|---|---|---|
| Product feed (OpenAI spec) | Documented ranking signals: popularity_score, return_rate | Medium |
| Structured data (Schema.org Product) | Used by Shopify Catalog to match specs to queries | Low–Medium |
| Third-party citations & reviews | GEO study: up to 40% visibility lift from cited, factual content | High |
| llms.txt | None. 300k-domain study found no effect | Low |
The pattern is clear: spend your hours on the first three rows.
Why the product feed beats the text file
When a ChatGPT shopper asks for a product, it queries Shopify Catalog — a structured product database, not your llms.txt. Shopify’s help docs confirm eligible products are surfaced through the Catalog and owned feeds, at no extra fee. OpenAI’s Agentic Commerce product feed spec goes further and asks merchants for real performance signals — popularity_score and return_rate — so the engine can rank products customers actually keep. That’s a documented ranking mechanism. llms.txt is not mentioned anywhere in it.
What should a Shopify merchant do this week?
- Fix the feed. Fill every attribute OpenAI’s spec asks for, including
popularity_scoreandreturn_rateif you have the data. This is the surface engines rank from. - Get your Product schema right. The GEO research shows specific facts (“100% GOTS organic cotton, 200 GSM”) beat marketing copy (“luxuriously soft”). Put the specs in structured data.
- Earn third-party citations. Reviews, roundups, and comparison pages that describe your product in factual terms are the measured 40% lever.
- Add llms.txt if you want — it’s a five-minute, low-risk hedge against future adoption. Just don’t count it as done work toward getting cited.
The honest bottom line
Adding llms.txt won’t hurt you, and if answer engines start reading it tomorrow, you’re covered. But if you’re doing it instead of your feed and structured data, you’ve optimized for a file nobody’s proven anyone reads while ignoring the two surfaces engines demonstrably rank from. Spend the effort where the evidence is.
That’s exactly what AI Visibility for Shopify is built to do: it audits your product feed against OpenAI’s spec, checks your Schema.org Product markup for the gaps that keep you out of AI answers, and shows where competitors are getting cited and you aren’t — the measured levers, not the fashionable one. See where your store actually stands at aeo.crosstowntech.com.