Misguided investments in LLMs.txt are failing to translate into AI-driven traffic.

Website owners prioritizing LLMs.txt as a mechanism for AI discovery are chasing a non-existent utility. Google has clarified that these files were never designed for ranking or search visibility, rendering widespread efforts to implement them largely ineffective for growth.

What Happened

Google’s John Mueller recently reiterated that LLMs.txt was intended to assist AI agents in site-specific content utilization at inference time, not as a signal for discovery or ranking. Despite this, many marketing teams have implemented the standard under the mistaken belief it acts as a meta-tag for AI search. Data from Ahrefs shows 97% of these files receive zero requests, confirming the lack of adoption by major AI crawlers for indexation purposes.

Why It Matters

First-order: The SEO industry is currently misallocating technical resources toward a file format that major AI search providers (Google, OpenAI, Anthropic) do not utilize for indexing. This creates a false sense of security regarding AI-search presence.

Second-order: AI models rely on independent relevance judgments—inherently untrustworthy signals like self-reported LLMs.txt files are being deprioritized. This mirrors the abandonment of meta-keywords in the 2000s; reliance on manual tagging for ranking is a structural error in an era of automated indexation.

Third-order: The focus must shift back to technical SEO fundamentals—structured data, high-quality content, and performant architecture. These remain the only verifiable vectors for improving visibility in generative search environments.

The Numbers

  • 97% of LLMs.txt files receive zero traffic (Source: Ahrefs)
  • Zero correlation exists between LLMs.txt adoption and AI citation frequency (Source: SE Ranking)

What To Watch

  • Continued pushback from major search engines against proprietary “AI signaling” files that attempt to manipulate indexing.
  • Acceleration of structured data adoption (Schema.org) as the primary bridge for machine-readable web content.
  • Increased scrutiny of “Generative Engine Optimization” (GEO) agencies that charge fees for implementing non-functional standards.