Do you need an llms.txt file? What the evidence says so far
For most business websites, no. Google added a note to its guidance in June 2026 saying an llms.txt file is not needed for Search and neither helps nor hurts visibility. Published analyses have found that the vast majority of these files receive no requests and that having one shows no meaningful correlation with AI citations. The one real use case is developer documentation that coding assistants read directly. Your time is better spent on clear pages, verifiable facts and third-party mentions.
Every few months a new file promises to fix how AI sees your website. In 2025 and 2026 that file was llms.txt: a plain-text summary of your site, placed at the root, meant to tell language models what matters.
Clients have asked us whether they need one. Here is what we know.
What it is
The proposal is simple. Put a Markdown file at /llms.txt that lists your most important pages with short descriptions, so an AI system can find the useful content without parsing navigation, scripts and layout.
It is a reasonable idea. The question is whether anyone reads it.
What Google said
In June 2026, Google added a note to its guidance on AI features, logged in its changelog as a clarification on llms.txt files. The substance: the file is not needed to appear in Google Search, including AI Overviews and AI Mode, and having one neither helps nor harms visibility or rankings.
That settles it for Google. It does not settle it for every AI system, so we looked at the other evidence.
What the studies found
Two kinds of analysis have been published, both by SEO tool vendors, so treat the exact figures with some caution:
- Traffic to the file. An analysis of server logs reported that the overwhelming majority of llms.txt files received no requests at all in the period measured. If the file is not fetched, it cannot influence anything.
- Correlation with citations. A study across hundreds of thousands of domains reported no statistically meaningful relationship between having an llms.txt file and how often a site was cited in AI answers.
Neither study proves the file can never matter. Together they make a strong case that it does not matter today.
Where it is useful
There is one real use case: developer documentation. Some AI coding assistants look for a product’s llms.txt to find the right documentation pages to read. If you sell an API, an SDK or a developer tool, a clean llms.txt pointing to current docs is worth an afternoon.
For a law firm, a restaurant group, a SaaS marketing site or an online store, it is not.
What to do instead
The things that have consistent support are less novel and more work:
State facts plainly on the page. AI systems extract what is visible in the content. A page that says “Projects start at $9,500 and take six to eight weeks” gives them something to quote. A page that says “competitive pricing and fast turnaround” does not.
Describe your company the same way everywhere. Name, category, location, services. Your site, your Google Business Profile, LinkedIn, directories and review platforms should agree. Inconsistency makes a model less confident about who you are.
Earn mentions elsewhere. Reviews, directory listings, industry press, partner pages, podcasts. AI systems corroborate across sources, and a company only described by itself is easier to leave out.
Answer the questions buyers ask. Use the wording from your sales calls, not your positioning deck.
A related claim, that schema markup is the key to AI citations, has also been tested. The results are more mixed than the advice suggests: see does schema markup get you cited by AI.
If you want to know where you stand today, the honest method is to ask the systems directly, repeatedly, with a fixed set of questions. We described how in what decides whether ChatGPT recommends you.
Common questions
Will having an llms.txt file hurt my site?
No. It is a text file that is mostly ignored. The cost is the time spent writing and maintaining it, and the risk is mistaking it for progress on AI visibility.
What is llms.txt for, then?
It was proposed as a curated, plain-text map of a site's most useful content for language models. Where it has found a use is technical documentation: some AI coding tools fetch a product's llms.txt to find the right docs pages.
Is there anything that does reliably improve AI citations?
Nothing is guaranteed, but the consistent signals are: pages that state specific, verifiable facts clearly; a consistent description of your company across your site, profiles and directories; and independent sources that mention you. These overlap heavily with good SEO.
