Does llms.txt Actually Boost AI Search Visibility? The Data Says No

No, llms.txt does not measurably improve AI search citations. Multiple studies published in 2026, covering over 500 million AI bot visits and 300,000 domains, found no citation lift from publishing the file, and Google has confirmed on the record that it doesn't use llms.txt at all.

For founders who've been told that adding an llms.txt file is a quick win for AI visibility, the evidence built up over the past year tells a very different story.

What is llms.txt, and why did people think it mattered?

llms.txt is a proposed standard, introduced by Jeremy Howard in September 2024, that gives AI systems a simplified, machine-readable summary of a website's content. The idea was simple: instead of making an AI crawler wade through menus, adverts and JavaScript, you hand it a clean list of your best pages.

That pitch made sense on paper, which is why agencies and SEO tools rushed to add it to their checklists. The problem is that "makes sense on paper" and "actually changes what AI answers cite" turned out to be two very different things.

What does the actual usage data show?

The clearest signal comes from server log analysis. Ahrefs analysed 137,000 sites and found that 97% of existing llms.txt files received no requests at all during the month studied.

Other independent trackers reach the same conclusion from different data sets. One tracker monitored over 500 million AI bot visits across a 90-day window and found only 408 targeted llms.txt directly. A separate study by Otterly.AI found that only 0.1% of AI crawler requests touched /llms.txt over 90 days, with the file receiving far fewer visits than an average content page.

A broader roundup pulled these findings together: studies across 300,000 domains from SE Ranking, 62,100 AI bot visits from Otterly.AI, 37,894 AI-cited domains from Trakkr, and over 500 million bot events from Limy.AI all show no measurable lift in AI search citations from llms.txt by itself.

SE Ranking went further and tested whether the file predicts citation behaviour at all. It analysed roughly 300,000 domains and found no statistically significant correlation between having an llms.txt file and how often a domain gets cited in AI answers; removing llms.txt as a variable from their predictive model actually improved its accuracy, meaning the file was noise, not signal.

Why does this matter for your business?

If you've spent budget on llms.txt implementation as a visibility tactic, this data suggests that spend didn't move the needle. Worse, it may have crowded out work that actually does affect whether AI systems cite and recommend your brand.

Google has been the most explicit about this. In July 2025, Google's Gary Illyes confirmed Google doesn't support llms.txt and isn't planning to, and John Mueller compared it to the discredited keywords meta tag. That comparison is pointed: the keywords meta tag was abandoned by search engines because it was a self-declared signal that sites could game, and search engines learned not to trust self-declared signals.

The pattern holds across the wider AI ecosystem too. Independent log studies and ecosystem analyses echo the same finding: mainstream AI search and LLM providers are not meaningfully relying on llms.txt at this time. And crucially, no major LLM provider, including OpenAI, Anthropic, Google, Meta or Mistral, has publicly committed to using llms.txt as a signal in their production search or answer surfaces.

This doesn't mean the file is entirely useless. It has a real, narrower home. Major developer platforms including OpenAI, Anthropic, Stripe, Cloudflare, Mastercard, Vercel and the Microsoft Teams SDK use llms.txt as a routing layer for AI coding agents. If you run a developer-facing product, that's a legitimate reason to have one. If you're trying to get cited in ChatGPT, Gemini or Perplexity answers about your brand, it isn't.

What should you do instead?

The evidence points firmly towards effort spent elsewhere. As one analysis summarised it, brands should focus on monitoring which queries and answers actually cite their brand rather than chasing files with no proven citation benefit.

Data on what actually drives citations is instructive here. Yext analysed 6.8 million AI citations across ChatGPT, Gemini and Perplexity in 2025 and found about 86% came from brand-controlled or brand-influenced sources: 44% from first-party websites, 42% from listings, 8% from reviews and social, and 6% from uncontrolled news or forums. As Yext's CEO put it: "When brands control their data, they control their visibility."

That means the levers that matter are the same fundamentals GEO practitioners have been pointing to all year: clean, crawlable pages that answer engines can actually render, accurate listings data, and content that clearly states what you do and who it's for. If you want to know whether your own pages are showing up when AI systems answer questions about your category, checking your AI visibility with Sited's free audit at https://sited.online will tell you more in five minutes than adding an llms.txt file will.

Robots.txt still matters for controlling what crawlers can access, separately from citation strategy. Providers now separate training and search crawlers, so you can block training agents like GPTBot, Google-Extended and ClaudeBot while allowing search agents such as OAI-SearchBot, Claude-SearchBot and PerplexityBot that make you eligible for AI-answer citations. That's a genuinely useful lever. llms.txt, based on the data available in 2026, is not.

Frequently asked questions

Does llms.txt improve my citation rate in ChatGPT or Perplexity?

The evidence says no. Crawler interest is negligible from search and answer bots, with only 408 out of over 500 million monitored AI bot visits targeting llms.txt directly over a 90-day window.

Is llms.txt the same as robots.txt?

No. llms.txt is not the same as robots.txt or sitemap.xml: robots.txt controls crawler access, sitemap.xml lists every URL you want indexed, and llms.txt curates a high-signal reading list for AI tools. Only robots.txt has documented enforcement behaviour from major crawlers.

Has any major AI company said they use llms.txt?

No. As of the most recent evidence, none of OpenAI, Google, Anthropic, Meta, Perplexity or Mistral has publicly stated that their production systems read or act on llms.txt.

Should I remove my llms.txt file if I already have one?

There's no need to remove it, but don't expect it to drive citations. Given current usage patterns, llms.txt should be seen as low risk, low cost and experimental, with little downside to having it but also little immediate upside for AI search visibility.

What should I focus on instead of llms.txt?

Prioritise first-party content accuracy, listings data and genuine crawler accessibility. Around 86% of AI citations trace back to brand-controlled or brand-influenced sources, split across first-party websites, listings, and reviews or social content. That's where the real visibility work happens.