llms.txt adoption is low in our sample: 15 of the 86 sites we could analyse, or 17.4%, served an llms.txt file on 29 October 2026.
What we did
For each of 100 domains, our script requested /llms.txt and kept the answer only if it was plain text, not an HTML page. Fourteen sites could not be analysed because their robots.txt was not readable, and they are left out of the percentage. The sample is described on the audit page and the file list is on the data page.
What the number means
- The file exists on roughly one site in six. It is not a norm.
- We did not test whether any engine reads it. The result says how many sites publish it, not whether it helps.
- Presence is binary here: we did not read the contents or judge quality.
What it does not show
A site with no llms.txt can still be cited, and a site with one can be ignored. Treat it as a cheap file to publish if you want to, not as a ranking lever. We will repeat the check and show the change over time.