**Publishing an llms.txt file does almost nothing for how ChatGPT, Google, or Perplexity treat your site — the traffic data says so, not opinion.** A 2026 study of 137,210 domains found 97% of published llms.txt files received zero requests in a month, and Google's own Search documentation states plainly that its systems ignore the file.
That matters because llms.txt has been marketed hard this year as an "AI SEO" essential — a short markdown file at yourdomain.com/llms.txt that's supposed to help AI engines understand and cite your business. Thousands of small-business sites, including ours, now publish one. The honest question is whether it does anything for a local business trying to show up in an AI answer. The data says mostly no, and it's worth knowing exactly why before you pay anyone to build one for you.
What Is llms.txt, and What Was It Actually Built For?
llms.txt is a proposed spec published by Jeremy Howard on September 3, 2024, with a v2 update on August 10, 2026 (llmstxt.org, 2026). It's a plain markdown index at a site's root meant to give AI systems a clean, navigation-free summary of a site's content — explicitly aimed at AI coding agents pulling software documentation, not at SEO or AI-search visibility. The spec's own framing is that "agents are best served by concise, expert-level information gathered in a single, accessible location," a problem that matters most for developer docs and API references, not for a plumbing company's homepage.
Do ChatGPT, Google, or Perplexity Actually Read llms.txt?
| Platform | Public stance on llms.txt | What it means for your site |
|---|---|---|
| Google Search / AI Overviews | Explicitly ignores it — "will neither harm nor help your site's visibility or rankings" (Google, 2026) | No ranking or citation effect, confirmed in Google's own docs |
| OpenAI / ChatGPT | Publishes its own llms.txt as a documentation index, but GPTBot and OAI-SearchBot are governed by robots.txt, not llms.txt (OpenAI, 2026) | OpenAI uses the format for its own API docs — not as an input for citing other sites |
| Anthropic / Claude | Publishes its own llms.txt but has not stated its crawlers use the standard on other sites (Ahrefs, 2026) | Same pattern: used internally, not confirmed as a ranking signal |
| Meta | No public crawler guidance and no indication of llms.txt usage (Ahrefs, 2026) | No confirmed effect either way |
Notice the pattern: OpenAI and Anthropic both publish llms.txt files for their own developer documentation — because that's exactly what the format was built for — while neither has said its crawlers rely on the file when deciding what to cite on someone else's site. That's a meaningful distinction most "add an llms.txt for AI visibility" guides skip over.
97% Zero Requests: What the Traffic Data Actually Shows
Ahrefs analyzed all 137,210 domains in its Web Analytics panel that received traffic in May 2026. 28% of those domains published a valid llms.txt file. Of the files that existed, 97% received zero requests all month (Ahrefs, 2026). Of the small sliver of traffic that did hit an llms.txt file, 96% came from bots and only 4% from humans — and of that bot traffic, 77% came from non-AI tools like SEO auditors and link-checkers, not from AI crawlers at all.
- 28% of analyzed domains publish an llms.txt file (Ahrefs, 2026).
- 97% of those files got zero requests in the month studied (Ahrefs, 2026).
- Of the requests that did land, 77% came from non-AI bots — SEO auditing tools, not AI crawlers (Ahrefs, 2026).
- Google logged a guidance update on July 10, 2026 saying Search ignores the file entirely (Google, 2026).
When llms.txt Is Actually Worth Doing
We publish an llms.txt file on our own site. It costs nothing to maintain once it's generated from existing content, and it does no harm — Google says so directly. But if an SEO consultant is charging a Whitefish or Kalispell business owner a setup fee specifically to "get AI-ready" with an llms.txt file and calling that the fix for AI visibility, the data above is the reason to ask what else is in the plan.
What Actually Gets a Local Business Cited by AI Instead?
The levers that move AI citations are less novel and more work: answer-first content that states a direct conclusion in the first sentence, real statistics and named sources instead of vague claims, FAQ and LocalBusiness schema markup, a crawlable site AI bots can actually read, and consistent business facts (name, services, location) across the web. None of that is a file you drop at the root and forget — it's the same discipline as good SEO, extended to a new set of engines.