NotionCue
AI Visibility Platform
All systems live
Sign in →
AEO Guidellms.txt GeneratorRobots.txtBLUF TemplatesBlogChangelogAbout
← Blog
TechnicalJun 22, 2026·8 min read

Your llms.txt File Probably Isn't Doing What You Think

Three major studies, 300,000+ domains analysed, and one finding: llms.txt does not directly move AI citation rates. Here is what it is actually good for.

SS
Sudhir Singh
Senior SEO & AEO Specialist · NotionCue
📄

There is a version of this file that does something useful. There is also a version that takes thirty minutes to build, sits at your domain root untouched, and accomplishes nothing. Most sites have the second version.

llms.txt got a lot of attention in 2024 as the AEO community started treating it like a shortcut to AI citation visibility. The pitch was simple: write a plain text file pointing AI crawlers to your best pages, and they start citing you more. That pitch was wrong, or at least wrong about the causal direction.

What the File Actually Is

llms.txt is a plain Markdown file you place at the root of your domain at yourdomain.com/llms.txt. Its job is to give AI systems a clean, curated summary of your site without them having to parse JavaScript, navigate menus, or guess which pages matter.

The spec at llmstxt.org defines a specific structure. Every valid file starts with an H1 containing the site or project name. Immediately after, a blockquote summary explains what the site does. Then optional H2 sections organise links to key pages, each with a short description.

# NotionCue

> AEO tracking platform that monitors brand citations across ChatGPT, Perplexity, Claude, and Google AI Overviews, with an llms.txt generator and prompt-level visibility dashboard.

## Core Tools

- [AI Citation Tracker](/tracker): Track how often and where your brand is cited across five AI engines.
- [llms.txt Generator](/llms-generator): Build a spec-compliant llms.txt file based on your actual crawl and citation data.
- [AI Crawler Audit](/crawler-audit): See which AI bots are fetching your pages and which are being blocked.

## Optional

- [Changelog](/changelog): Recent product updates and new features.

The Evidence on Whether It Works

Several independent studies now exist on this question, and the consensus is fairly clear.

  • SE Ranking analysed 300,000 domains and found no correlation between having an llms.txt file and AI citation frequency.
  • Generix Marketing ran 2,139 prompt checks across ChatGPT, Claude, and Perplexity against 2,500 of the world's most visited websites and found only a 1.27x over-representation, likely explained by stronger overall technical SEO on sites that bother to publish the file.
  • Ahrefs published a study in June 2026 across 137,000 sites and found that 97% of llms.txt files received zero requests in May. No bots. No humans. Nothing.

Google stated explicitly in its May 2026 generative AI optimisation guide that no special files, markup, or Markdown are needed to appear in Google Search or AI Overviews.

Where It Actually Has Value

Two reasons hold up. First, some AI-powered coding tools and browser agents do parse the file, Cursor, Claude Code, and similar tools. If your site serves developers or technical users, the file is actually read by a meaningful audience. Second, the file costs very little if built correctly. A clean, accurate llms.txt takes an afternoon and requires no maintenance beyond updating it when your site structure changes materially.

The mistake is treating llms.txt as a citation lever. It is not. It is infrastructure, and low-priority infrastructure at that. Build your file once the fundamentals are solid, crawler access, schema, content freshness, entity signals.

The Five Mistakes That Make the File Useless

  • The H1 is a tagline, not a name. "The world's most powerful AEO platform" is not a name. "NotionCue" is. Positioning copy belongs in the blockquote.
  • The blockquote is vague. "We help businesses grow online" could describe ten thousand companies. Write it for a system that is categorising your site, not a human skimming a homepage.
  • Pages are listed that should not be there. llms.txt is a curated directory, not a sitemap. Thin pages, archived content, and old campaign landing pages should not be in it.
  • The file is outdated. An llms.txt with last year's product names or discontinued features is worse than no file, it creates a misinformation problem.
  • It is blocked by the CDN. Verify your file returns an HTTP 200. curl -I https://yourdomain.com/llms.txt takes five seconds and confirms it.
Share this post
Check your AEO score
Scan your domain free — get your AI visibility score across 5 LLMs in 30 seconds.
Scan my site →
SS
Sudhir Singh
Senior SEO & AEO Specialist · NotionCue

Senior SEO and AEO specialist with 12+ years across e-commerce, global education, and healthcare. Building Notion Cue to track brand citations across ChatGPT, Perplexity, Gemini, and AI Overviews.

View all →
📖
Technical11 min read

The AEO Glossary: 25 Terms Every AI Search Practitioner Should Know

This is the first entry in a new reference series. Rather than another explainer, this is a working glossary — short, direct definitions for the terms that come up constantly across AI search, AEO, and GEO discussions, each linked to a deeper guide where one exists on this blog.

SS
Jul 27, 2026
🗺️
Technical9 min read

Sitemap Strategy for AI Crawlers: The Technical Layer Most AEO Audits Skip

Server log analysis across a meaningful sample of sites found GPTBot referencing sitemaps on roughly three out of four initial domain visits, compared to just over half for Googlebot. PerplexityBot showed an even stronger dependency, crawling sitemap-listed URLs several times more often than equivalent pages it found only through links. Most sitemaps are still built entirely around Google.

SS
Jul 22, 2026
📅
Technical8 min read

Event Schema for AEO: Why Your Conference or Webinar Needs Structured Data Now

Google's own list of schema types still worth investing in after its 2026 rich-result cleanup names Event alongside JobPosting as one of the categories that maps to genuine, ongoing business use rather than SERP-gaming. Most companies running webinars and conferences have never implemented it.

SS
Jul 20, 2026
🕷️
Technical7 min read

Google-CloudVertexBot and the New AI Crawlers Your robots.txt Was Never Configured For

Google added Google-CloudVertexBot to its official list of crawlers in 2026 — a new user-agent specifically for sites owners that build Vertex AI Agents and want that agent crawling their own site on their behalf. It is one of a growing list of newer, purpose-specific AI crawlers most robots.txt files have simply never been updated to address at all.

SS
Jul 17, 2026
Get AEO updates weekly.

Citation shifts, algorithm changes, and what's actually working.