Since the rise of AI assistants like ChatGPT and Gemini, a new practice has spread fast across SEO communities: creating an llms.txt file at the root of a website, modeled after robots.txt, to “guide” AI models and hopefully get cited more often in their answers. Dozens of articles and WordPress plugins pushed this idea as the next big SEO opportunity. The problem: Google just settled the debate, and the data confirms this practice does almost nothing for classic search rankings.
What is llms.txt and why everyone started talking about it
llms.txt is a text file format proposed in late 2024 by an independent developer, meant to summarize a site’s content in a format readable by large language models (LLMs) — essentially a simplified Markdown sitemap for AI. The appealing pitch: instead of letting an AI “guess” what your site contains by crawling complex pages, you hand it a clean, structured summary directly.
Drawn by the promise of better visibility in AI-generated answers, countless site owners, agencies, and plugin developers adopted the format within months, without waiting for any official confirmation from search engines.
An analogy, not a technical confirmation
Much of llms.txt’s success rests on a comparison with robots.txt, a real, well-established file search engines have used since the 1990s. But the analogy stops there: unlike robots.txt, no major search engine has ever confirmed reading or using llms.txt to rank or understand content.
What Google officially said in 2026
Google’s AI optimization guide, updated on June 15, 2026, is unambiguous: site owners don’t need to create new machine-readable files, AI text files, markup, or Markdown to appear in Google Search, including its generative features like AI Overviews or AI Mode. Google Search, the guide states plainly, simply doesn’t use files of this kind.
Google spokespeople doubled down publicly. Gary Illyes said Google does not support llms.txt and has no plans to. John Mueller called the idea purely speculative, comparing it to the old meta keywords tag — a format that existed, that plenty of sites invested time into, but that search engines never actually used for ranking.
Mueller also clarified an interesting detail: when llms.txt occasionally shows up on some Google properties, it’s only because an internal content management system added it automatically — not because Google Search itself makes any use of it.
The numbers confirming the lack of impact
Beyond the official statements, field data points the same way. A large-scale study by Ahrefs covering 137,000 sites found that 97% of the llms.txt files analyzed generated zero traffic in May 2026. That number speaks for itself for every site owner who spent time writing and maintaining this file expecting an AI visibility boost.
This fits a recurring pattern in SEO: every new technology brings its share of unverified “miracle fixes” that pull attention away from the levers that actually move rankings.
Why the confusion persists anyway
Part of the confusion comes from mixing up two very different use cases:
- llms.txt for classic SEO: useless according to Google, with no measurable effect on search rankings or AI-generated search features.
- llms.txt for AI agents and APIs: Anthropic explicitly recommends this format in its “Writing for Agents” documentation, and OpenAI maintains llms.txt files for its Agents SDK and its agentic commerce protocol.
These two use cases have nothing to do with each other: one concerns organic visibility in a consumer search engine, the other concerns software agents interacting directly with technical documentation or an API. Conflating the two fueled a good part of the noise around this topic in 2026.
Where to actually invest your time in 2026
Rather than spending hours on a file with no proven effect, site owners are far better off focusing on levers with documented impact:
- Content that is genuinely crawlable, with nothing hidden or rendered only client-side without a clean fallback.
- Up-to-date Schema.org structured markup, still the real language that both Google and generative AI systems natively understand.
- Core Web Vitals within recommended thresholds (LCP under 2.5s, INP under 200ms, CLS under 0.1), which directly affect experience and indexing.
- A coherent internal linking structure, which helps both crawlers and generative AI understand your site’s structure and authority.
- Demonstrated expertise and trust (E-E-A-T), the factor that genuinely weighs in whether AI Overviews cite you.
Conclusion
llms.txt is a textbook case of the risk in following an SEO trend without checking its source: an appealing idea, massively amplified, but never confirmed by the search engines it was supposedly built for. In 2026, the winning strategy remains the same as before generative AI arrived: a technically flawless site, structured and trustworthy content, and authority built over time. To find out where your site actually stands on these fundamentals, the seoforge.io team offers a full SEO audit and hands-on guidance to help you prioritize the actions that truly move the needle for your visibility.
