GEO & AI Search
Does llms.txt work?
No - published measurement shows llms.txt files are almost never requested, and Google has stated it does not use them, so the file is currently hygiene rather than a growth lever.
What the file is
A proposed convention: a markdown file at the root of a site telling language models which pages matter and summarising what the site is about. The analogy to robots.txt is obvious and appealing, which is why the idea spread through agency proposals faster than anyone measured it.
What the measurement found
Ahrefs studied roughly 137,000 domains that had implemented the file. About 97% of those files received zero requests. Among the small remainder that saw any traffic, retrieval bots accounted for roughly one percent. Google has separately said publicly that it does not use llms.txt. A convention only works when consumers adopt it, and adoption is currently near zero.
Should you add one anyway
It costs ten minutes and does no harm, and adoption could improve. Treat it like a well-formed humans.txt: hygiene. What is not defensible is a line item, a slide, or a monthly fee attached to it. If an agency proposes it as a deliverable, ask them for request logs.
One failure mode to check
Framework routing defaults sometimes make the llms.txt path return a server error rather than a file. An error at a well-known path is marginally worse than having no file at all. If you add one, confirm it actually returns content, then stop thinking about it and spend the attention on crawler access and third-party placements instead.
We build automation systems for small teams. Free audit call to map your specific workflows - no pitch, just a plan.