New data on 515 million AI traffic events comes with a rather blunt verdict: LLMS.txt is largely ignored by the major AI bots that determine visibility in generative answers.
The news has two sides: on one hand, it debunks one of the most popular "SEO quick wins" of recent months; on the other, Microsoft launches Citation Share in Bing Webmaster Tools — the first official tool to concretely measure how much you're cited in AI results.
The Data on LLMS.txt
The analysis of 515 million AI bot events shows that the main crawlers — GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, OAI-SearchBot, and Google-Extended — visit the llms.txt file at a statistically irrelevant rate.
A separate analysis of 300,000 domains confirms: LLMS.txt has no measurable impact on how AI systems see or cite your content today.
Google went further: it openly stated that it doesn't use LLMS.txt to differentiate between sites.
This doesn't mean LLMS.txt is entirely useless in absolute terms — it could become relevant with future developments in the standard. The implementation cost is low, so those who already have it can keep it. But as a strategy for improving AI visibility in 2026, the data doesn't support the optimism that had been circulating.
What Bing Is Launching: Citation Share
The constructive news comes from Microsoft. The AI Performance dashboard in Bing Webmaster Tools is gaining four new features:
- Citation Share: the percentage of AI citations your site captures for a given query — the most direct metric ever made available
- Intents: the search intents for which you're cited
- Topics: the topics where you have visibility in AI answers
- Compare: comparison against competitors (in preview)
It's the first official tool that lets you concretely measure GEO (Generative Engine Optimization) visibility instead of guessing.
What Really Matters for Being Cited by AI
If LLMS.txt doesn't move the needle, what does? The data now available points to much more traditional factors:
Domain authority: sites with over 32,000 referring domains are 3.5 times more likely to be cited by ChatGPT than sites with fewer than 200 backlinks. The authority you've built for Google also counts in AI.
Citations on authoritative sources: brands mentioned frequently on media and recognizable industry sites have a bigger presence in AI answers. AI learns from the web ecosystem — if you're cited everywhere, you're more "real" to generative systems.
Structured content: pages that directly answer specific questions, with clear headings and verifiable data, are preferred as sources over generic pages.
Brand consistency and entity identity: a recognizable entity — consistent name, site, social media presence, Wikipedia if applicable — makes it easier for AI systems to recognize you.
The Lesson
GEO (Generative Engine Optimization) isn't separate from traditional SEO — it's an extension of it. Sites with solid authority, quality content, and external recognizability start ahead in AI visibility too.
LLMS.txt was a sensible idea — an explicit signal to AI crawlers about how to treat your content. But the data says that, for now, crawlers aren't listening to that signal. AI visibility is earned by building real authority, not with a text file.
If you're investing time in purely technical AI optimizations like LLMS.txt, the data suggests redirecting that energy toward building authority, content quality, and external presence. Those are the factors that matter — now and probably for a long time.