AI search

Does llms.txt work? The studies, side by side

Claims about llms.txt come mostly from vendors. This page lists the measurements and statements that exist, with exact numbers from the source, what each one can and cannot show, and a plain verdict at the end.

By Adem ÇakalPublished 4 sources6 min read
Short answer

No published measurement shows that llms.txt raises AI citations. SE Ranking found no effect on about 300,000 domains. Search Atlas found no measurable gain on 13,911. Ahrefs found that 97% of the llms.txt files it tracked received zero requests in May 2026. Google says its Search does not use the file. The honest reading: no proven benefit, no proven harm, and a few non-search readers such as coding agents.

The evidence, rated

SourceTypeSample and dateFindingLimits
Google AI optimization guideOfficialCurrent page"You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search." Google Search itself does not use them.Covers Google Search only, not other assistants
Ahrefs (June 15, 2026)Measurement, server logs137,210 domains with traffic in May 2026; 28% published llms.txt (about 38,000)97% of files got zero requests. AI retrieval bots made up 1.1% of requests to the file.Technical customer base, so 28% is an upper bound. A fetch does not prove the content was used. Did not check files against the spec.
SE Ranking (page dated Oct 15, 2026)Measurement, statistical modelAbout 300,000 domains; 10.13% had the fileNo effect on citation frequency. Removing the llms.txt variable improved model accuracy on test data.Authors call results context-dependent. The page does not name the LLMs sampled or the time window.
Search Atlas (updated June 19, 2026)Measurement, cross-sectional13,911 matched domains; 5,997 had the file; 180-day windowMedian difference +0.42 on a 0 to 100 visibility scale, r about -0.02. Sign flipped between 90-day and 180-day cohorts.Not a random sample. Visibility score, mentions and share of voice, not direct citations. Authority check covered only 1,041 domains.

We did not include other studies we could not open at the source. A vendor claim without a linked method is not in this table on purpose.

What each study can and cannot show

Two kinds of question: do AI answers cite sites with llms.txt more, and do bots fetch llms.txt at all Does it change citations? SE Ranking: no effect found Search Atlas: no measurable gain Cross-sectional. Cannot show cause, small effects or future change. Do bots fetch it? Ahrefs: 97% of files, zero requests Coding agents the top AI group Request counts are a ceiling: a fetch is not proof of use.
The studies answer two different questions. Ahrefs' log data gives a plausible reason for the null results in the other two, but none of them proves it.

How to check it on your own site

Your server log is the only evidence that matters for your domain. Count who requests the file:

# Linux, combined log format: who asked for /llms.txt, and how many times
grep "GET /llms.txt" access.log | awk -F'"' '{print $6}' | sort | uniq -c | sort -rn | head

Expect mostly SEO tools and generic bots, as Ahrefs found. To validate the format, use the llms.txt checker. To make sure the bots that matter can reach your pages at all, run the AI crawler checker.

Verdict

  • For rankings in Google or AI Overviews: no. Google says so directly.
  • For ChatGPT or other assistants citing you: unproven. Neither large study found a lift. See the ChatGPT checklist for steps that are documented.
  • For developer docs read by coding agents: plausible. The Ahrefs log data shows agents fetching the file.
  • Setup cost: low. A WordPress guide exists if you want one.

This will change if a platform documents that it reads the file. Until then, treat llms.txt as optional housekeeping, ranked below crawl access, indexing and content quality.

Questions people ask

So should I delete my llms.txt?

There is no evidence it harms you, and Google says it neither helps nor harms Search visibility. If it is accurate and cheap to keep, leaving it is fine. Do not spend a week on it.

Why would a file that AI tools do not read still be worth anything?

Ahrefs found requests for llms.txt came mostly from non-AI bots, with coding agents the largest AI group. If you publish developer documentation, that is the audience that may read it.

Do these studies prove llms.txt has no effect?

No. Two of them are correlational and cross-sectional, so they cannot rule out small or future effects. They show that large measurable effects are absent in the samples tested.

Is a vendor saying llms.txt boosts citations evidence?

Only if they publish the data and method. A claim without sample size, date and method is marketing. Use the table below as the bar.

Sources

  1. Ahrefs: llms.txt study (June 15, 2026), accessed 2026-11-12.
  2. SE Ranking: LLMs.txt analysis, accessed 2026-11-12.
  3. Search Atlas: The limits of llms.txt for AI search citation, accessed 2026-11-12.
  4. Google Search Central: AI optimization guide, accessed 2026-11-12.