7.20Intermediate9 min

llms.txt: 844,000 Implementations, Zero Confirmed AI Usage

Lucas Blochberger·7 April 2026·Updated 11 June 2026

Definition

llms.txt is a Markdown file proposed by Jeremy Howard (September 2024) providing LLMs with a curated map of a site's most important content. Despite 844,000+ implementations, no major AI provider has confirmed using llms.txt, and SE Ranking found no correlation with AI citation rates.

Key Takeaways

✓844,000+ implementations according to BuiltWith
✓No AI provider has confirmed usage
✓Google's Gary Illyes: Google does not support llms.txt and has no plans to
✓John Mueller compares it to the keywords meta tag
✓SE Ranking: No correlation with AI citation rates (300,000 domains)
✓Server log analysis: AI crawlers do not request llms.txt
✓Yoast SEO integrated automatic generation (June 2025)

llms.txt is the most controversial technical SEO topic: high adoption meets zero confirmed effect.

The Specification

Jeremy Howard proposed the llms.txt specification on September 3, 2024 — a Markdown file that provides LLMs with a curated map of a website's key content. BuiltWith tracks over 844,000 implementations, concentrated in developer tools and SaaS (Anthropic, Cloudflare, Stripe, Vercel, Hugging Face, Zapier).

The Evidence Against

No major AI provider has confirmed its use. Google's Gary Illyes stated at Search Central Live in July 2025: Google does not support llms.txt and has no plans to. John Mueller compared it to the Keywords meta tag and pointed out that bots already download full pages. SE Ranking found no correlation between llms.txt presence and AI citation rates across 300,000 domains. Server log analyses consistently show that major AI crawlers do not request llms.txt files.

Google briefly — and apparently accidentally — added llms.txt to some of its developer docs in December 2025 before removing it again.

The Recommendation

Current evidence supports llms.txt as an optional future-proofing measure with negligible current effect. Schema markup is the far higher priority. The investment calculus is clear: implement structured data first, then llms.txt as a low-cost supplement.

Data & Statistics

Veröffentlicht am 3. September 2024 von Jeremy Howard; einziger zwingender Bestandteil ist eine H1-Überschrift mit dem Projekt- oder Site-Namen ("This is the only required section")

Answer.AI - Jeremy Howard, "The /llms.txt file" (Originalvorschlag) (2024)

"no AI system is currently using the LLMS.txt file" (John Mueller/Google); für AI Overviews gilt "use normal SEO practices"; Gary Illyes: Google unterstützt llms.txt nicht und plant es nicht

Search Engine Land (2025)

84 von über 62.100 KI-Bot-Anfragen (rund 0,1 Prozent) über 90 Tage zielten auf /llms.txt; 84 Besuche gegenüber etwa 265 für eine Durchschnittsseite (rund dreimal seltener)

OtterlyAI - The llms.txt Experiment (2025)

Rund 300.000 Domains analysiert; nur 10,13 Prozent mit llms.txt; kein Effekt von llms.txt auf die KI-Zitierhäufigkeit; ML-Modell verbesserte sich beim Entfernen der Variable

SE Ranking Blog (2025)

Bis zu plus 40 Prozent Sichtbarkeit in generativen Antworten; Statistiken, Zitate und Quellenangaben als Top-Methoden (30-40 Prozent PAWC, 15-30 Prozent Subjective Impression); GEO-bench mit 10.000 Anfragen

arXiv (Aggarwal, Murahari, Narasimhan, Deshpande et al.), "GEO: Generative Engine Optimization", KDD '24 (2024)

Markenerwähnungen im Web korrelieren mit r=0,664 deutlich stärker mit AI-Overview-Sichtbarkeit als Backlinks mit r=0,218 (75.000 Marken untersucht), Verhältnis rund 3 zu 1

Ahrefs Blog - An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied) (2025)

Klickrate 8 Prozent mit KI-Zusammenfassung gegenüber 15 Prozent ohne; nur 1 Prozent klickte eine Quelle in der KI-Zusammenfassung; 26 Prozent beendeten die Sitzung nach einer Seite mit KI-Zusammenfassung gegenüber 16 Prozent ohne (US-Daten)

Pew Research Center (2025)

Die Nutzung generativer KI-Tools wie ChatGPT liegt in der österreichischen Gesamtbevölkerung bei etwa 30 Prozent; 31 Prozent gaben an, ein solches Tool schon mindestens einmal genutzt zu haben (Erhebung IKT-Einsatz in Haushalten 2024)

Statistik Austria - Künstliche Intelligenz: Nutzung und Einstellung in Österreich (2025)

FAQ

What is llms.txt in simple terms?

llms.txt is a Markdown file in the root directory of a website designed to provide AI systems with a curated map of the most important content. It was proposed on September 3, 2024, by Jeremy Howard, co-founder of Answer.AI. The only mandatory component is an H1 heading with the project or site name; all other sections are optional.

Do ChatGPT, Google, or Perplexity use the llms.txt file?

According to current evidence, no. John Mueller from Google stated that currently no AI system uses the llms.txt file, and Gary Illyes clarified that Google does not support it and has no plans to do so. For AI Overviews, Google recommends normal SEO practices. No major AI provider has confirmed active usage.

Does llms.txt lead to better rankings or more AI citations?

There is no documented correlation. SE Ranking analyzed around 300,000 domains and found no effect of llms.txt on the frequency of AI citations, either statistically or through machine learning. When the variable was removed from the model, predictions actually improved.

What is the difference between llms.txt, robots.txt, and sitemap.xml?

robots.txt controls what crawlers are allowed to do and what not. sitemap.xml lists all indexable URLs in machine-readable format. llms.txt is a curated Markdown table of contents intended to highlight only the most important content. robots.txt and sitemap.xml are established standards evaluated by search engines, while llms.txt is a proposal without official provider support.

Should DACH B2B companies implement llms.txt?

Only if the file is generated automatically via CMS or plugin, as effort and risk are then minimal. A measurable effect on AI search visibility is not to be expected. With purely manual maintenance and no agentic or documentation use case, the maintenance effort outweighs the proven benefit. Limited GEO budget belongs in measures with evidence.

What actually drives AI visibility instead?

Primarily brand mentions on the web, which according to an Ahrefs analysis of over 75,000 brands correlate significantly more strongly with AI Overview visibility at r=0.664 than backlinks at r=0.218. Additionally, front-loading of answers as well as the use of statistics and source citations directly in content, which according to the GEO study from the KDD '24 proceedings can increase visibility in generative answers by up to 40 percent.

How do you measure AI visibility and AI referral traffic correctly?

Two-pronged approach: In GA4, isolate AI sources like ChatGPT, Perplexity, or Gemini as separate referral sources and evaluate their engagement and conversions separately. Additionally, analyze server logs to see which pages the AI bots actually retrieve. Classic click metrics underestimate the value, as according to Pew Research, only 8 percent instead of 15 percent click on a search result with AI summaries.

How does your website perform?

Get a free, AI-powered SEO report of your website by email – technical SEO, on-page, keywords & competitors. No obligation.

Get a free SEO audit →

Previous← Agentic Commerce: AI Agents as Purchase Mediators