# Almost nobody reads your llms.txt > llms.txt is supposed to help AI engines find and understand your content. The logs say almost nothing reads it. Here is what Google said, what server data from 137,000 sites shows, and what to do instead. **Félix Hérault**, GEO Analyst, CrawlerPal · September 22, 2026 · Technical GEO Canonical: https://crawlerpal.com/blog/llms-txt-almost-nobody-reads-it --- The belief behind llms.txt is appealing. Put a file at the root of your site that tells AI systems which content matters and how to read it, and engines will understand you better. It is cheap to add and it sounds sensible, and a lot of GEO advice treats it as a foundation. The evidence says almost nothing reads it. ## What Google said In June 2025, Google's John Mueller [wrote on Bluesky](https://www.seroundtable.com/google-ai-llms-txt-39607.html) that “no AI system currently uses llms.txt.” His reasoning was that it is visible in server logs: the consumer chatbots fetch your pages for training and grounding, and none of them fetch the llms.txt file. ## What the logs say One statement is one thing. In June 2026 Ahrefs published [an analysis of 137,210 domains](https://ahrefs.com/blog/llmstxt-study/) in its web analytics that received traffic in May 2026, looking at every request made to their llms.txt files. 97% of those files received no requests at all that month. Among the AI bots that did request them, retrieval bots, the kind that fetch pages to answer a user's question in the moment, made up about 1.1%. The engines people hope llms.txt will influence are barely reading it. ## Why it was never likely to work The likely reason is that llms.txt asks AI systems to trust what a site says about its own content. Search engines learned long ago to be wary of self-description, because it is the easiest signal for a site to game. An engine that wants to know what a page says can simply read the page. ## What it is still good for None of this makes llms.txt harmful. It costs almost nothing to publish, and some developer tools and agents may read it. If you already have one, there is no reason to remove it. What it is not is a visibility lever. Publishing one will not make an engine cite you, and time spent perfecting it is time not spent on the things that do. ## What to do instead The pages themselves are what engines read. Make sure your commercial pages render without heavy scripting and answer the questions buyers actually ask, with clean structured data behind them. Then check that the AI crawlers you want are not blocked in robots.txt, because a page an engine cannot retrieve cannot be cited. ## The bottom line Leave your llms.txt in place if you have one. Just stop counting on it. The evidence points to the page, not the pointer. ## Frequently asked questions ### Does llms.txt improve AI visibility? The evidence says no. Google's John Mueller said in June 2025 that no AI system currently uses it, and an Ahrefs analysis of 137,210 domains found that 97% of llms.txt files received no requests in May 2026. It is not a visibility lever. ### Should I delete my llms.txt? There is no need. It costs almost nothing to keep and some developer tools and agents may read it. Just do not treat it as something that will make engines cite you, and do not spend time perfecting it. ### What do AI engines actually read? Your pages. Engines fetch the content itself, to train and to answer questions. What matters is that your important pages can be retrieved and clearly answer the questions buyers ask. ### If not llms.txt, what should we fix first? Check that robots.txt is not blocking the AI crawlers you want, then make sure your commercial pages carry clean structured data and answer real buyer questions. A page an engine cannot reach cannot be cited, whatever file points to it. --- CrawlerPal · https://crawlerpal.com