- [ Home ](/) / - llms.txt and SEO # llms.txt and SEO Google does not use llms.txt for Search rankings or its generative Search features. The file has a different, narrower purpose: helping compatible agents find curated, machine-readable material. Last updated: August 12, 2026 ## The confirmed answer **Publishing llms.txt will not improve a page's Google ranking.** Google says Search does not use AI text files or special Markdown files, including for its generative features. Google may crawl or index such a file, but that does not give it special treatment. ⚠ No ranking promise Claims that llms.txt boosts rankings, AI Overviews, or Google citations are unsupported. This page treats llms.txt as an agent-discovery convention, not an SEO shortcut. ## What Google actually says Google's published advice is conventional SEO: keep pages crawlable, link important content internally, provide a good page experience, expose important information as text, and make structured data match visible content. Eligibility for AI Overviews and AI Mode still starts with being indexed and eligible for a normal Search snippet. Google also warns against producing extensive automated content primarily to attract search visits. Original reporting, reproducible research, clear sourcing, and demonstrable expertise are more defensible than another generic explanation of the same proposal. ## Where llms.txt can still be useful The v2 proposal addresses agents that deliberately fetch a site map for context. It supports a root file, more specific files under paths such as `/docs/llms.txt`, clean Markdown page alternates, and HTML or HTTP discovery relations. An agent must implement those conventions before they have any effect. That makes the strongest use case technical documentation and other text-heavy resources. A curated file can reduce the work required by a compatible coding agent, but the proposal does not grant access, control crawling, or force an assistant to cite a page. ## What the evidence does not prove Our fixed panel found 113 valid root files among 218 reachable hosts on 2026-08-03. That measures publication, not consumption. A company publishing a file does not prove that its crawler reads the convention or that the file changes recommendations. Ahrefs separately reported that most observed files received no request during its measurement window. The useful conclusion is modest: measure server requests and downstream referrals on your own property rather than inferring adoption from file counts. ## SEO priorities that matter - **Serve the search intent completely.** Answer the actual question before discussing the proposal. - **Publish evidence others cannot reproduce from memory.** Experiments, dated datasets, and transparent methods create information gain. - **Consolidate overlapping pages.** Give each query family one primary URL and link supporting research back to it. - **Improve titles and snippets.** Use concise, distinct wording that accurately reflects the visible page. - **Earn editorial references.** Publish citable datasets and tools instead of manufacturing mentions. - **Keep technical foundations clean.** Indexability, canonicals, hreflang, internal links, performance, and stable deployments remain prerequisites. ## How to measure the right outcomes - Use Search Console for queries, pages, clicks, impressions, CTR, and position. - Use server or edge logs to identify verified requests for `/llms.txt` and Markdown pages. - Track AI referral traffic separately, without attributing it to llms.txt unless an experiment supports that conclusion. - Compare updated pages against a fixed pre-change period and keep annotations for releases. Continue with the [v2 best-practice guide](/best-practices/), inspect [verified production files](/examples/), or test a file with the [llms.txt validator](/validator/). ## Sources - [ Google Search Central, generative AI optimization guide ](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [ Google Search Central, helpful people-first content ](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [ llmstxt.org, llms.txt v2 proposal ](https://llmstxt.org/) - [ Ahrefs, 137,000-domain llms.txt study ](https://ahrefs.com/blog/what-is-llms-txt/) On this page - [ The confirmed answer ](#answer) - [ What Google actually says ](#google) - [ Where llms.txt can still be useful ](#agents) - [ What the evidence does not prove ](#evidence) - [ SEO priorities that matter ](#priorities) - [ How to measure the right outcomes ](#measure)