LLMs.txt File SEO: What It Does and Doesn't Do in 2026

An llms.txt file is a plain Markdown file at a site's root that lists its key pages for AI systems to read. For llms.txt file SEO purposes, the honest 2026 answer is that it does not move rankings or AI citations: Google has confirmed it ignores the file, and independent studies of thousands of domains found no measurable citation lift from having one.

Key takeaways

• An llms.txt file is a Markdown summary and link list at yoursite.com/llms.txt, proposed by developer Jeremy Howard in September 2024.
• Google's Gary Illyes confirmed Google does not read or act on llms.txt for Search, and John Mueller compared it to the old keywords meta tag.
• SE Ranking's May 2026 study of nearly 300,000 domains found only 10.13% had adopted llms.txt, and 0% of the top 1,000 sites by traffic did.
• MaxAEO's matched-pair study of 2,400 domains found a 0.2 percentage point citation gap between adopters and non-adopters, well inside normal noise.
• Its clearest real use is developer tooling: AI coding assistants like Cursor and Claude Code can use it to fetch documentation faster.
• Content quality, schema markup, and topical authority remain the signals with an actual documented link to AI citations.

What is an llms.txt file?

An llms.txt file is a plain Markdown document that a website places at its root, so it loads at a URL like yoursite.com/llms.txt, to summarize the site's purpose and point AI systems to its most important pages. Developer Jeremy Howard of Answer.AI proposed the format in September 2024 through the reference site llmstxt.org, arguing that AI agents are "best served by concise, expert-level information gathered in a single, accessible location" rather than forced to parse full HTML pages built for human navigation, ads, and layout. A second version of the specification, informed by roughly two years of real-world adoption, was published in August 2026.

The idea borrows the spirit of robots.txt: a small, predictable file in a fixed location that a machine can check first. Where robots.txt tells crawlers what they may access, llms.txt tries to tell AI systems what matters most and hand them a shortcut past the rest of the page.

The llms.txt format

The specification defines a simple structure:

• An H1 heading with the site or product name (required).
• A blockquote with a one to two sentence summary.
• Optional paragraphs with extra context.
• One or more H2 sections, each listing key pages as Markdown links with short descriptions.

Some sites also publish an llms-full.txt, an expanded version with fuller page content inlined, aimed at agents that want more than links. Both files are meant to be plain text, never HTML or JavaScript, so any language model can parse them without rendering a page.

How do you create an llms.txt file?

Building one is a short technical task, not a content project:

1. Write the H1. Use your site or product's actual name.
2. Add the blockquote summary. One or two plain sentences describing what the site is and who it serves.
3. Add context if useful. A short paragraph on scope, audience, or how to use the linked resources.
4. List key pages under H2 sections. Group by topic, such as "Docs" or "Guides," and link each page with a one-line description.
5. Save it as /llms.txt. Plain text, at your domain root, no build step required.
6. Keep it current. Update the list when pages move or new priority content ships, the same discipline an XML sitemap needs.

Adoption of this simple process is real but far from universal. SE Ranking's May 2026 analysis of nearly 300,000 domains found that only 10.13% had a working llms.txt file in place, roughly even across low, mid, and high traffic sites, and 0% among the top 1,000 domains by traffic. That is a long way from the near-universal adoption of robots.txt or XML sitemaps, which suggests most site owners and SEOs are not yet treating it as a default.

Does an llms.txt file actually help SEO or AI search rankings?

No credible evidence shows that an llms.txt file improves search rankings or AI citations as of August 2026. Google's Gary Illyes confirmed in mid-2025 that Google does not read or act on llms.txt and has no plans to, and independent research from SE Ranking and MaxAEO, covering more than two thousand domains combined, found no statistically meaningful difference in AI citation rates between sites with and without the file.

MaxAEO ran the most direct controlled test. Between February and April 2026, the firm tracked 2,400 domains across eight AI platforms, including ChatGPT, Gemini, Perplexity, Claude, Copilot, Grok, and Google's AI Mode and AI Overviews, and built 240 matched pairs of similar sites, one with llms.txt and one without, to isolate the file's effect. The adopters averaged an 11.8% AI citation rate against 11.6% for the matched non-adopters, a 0.2 percentage point gap the researchers described as within their own week-to-week noise band. When 58 domains added llms.txt mid-study, their citation rate moved by a median of just 0.1 percentage points relative to controls, essentially no change.

SE Ranking's separate 300,000-domain analysis reached the same conclusion from a different angle: when the researchers removed llms.txt as a variable from their citation-prediction model, the model's accuracy improved, meaning the file was adding noise rather than signal once site authority, schema density, and content recency were already accounted for. Two independent studies, using different methods and different domain sets, landed on the same result: no measurable lift.

What do OpenAI, Google, and Anthropic say about llms.txt?

As of Q1 2026, no major AI provider, including OpenAI, Google, Anthropic, Meta, or Mistral, had publicly confirmed that its production systems read or prioritize llms.txt files. Google has been the most explicit. Gary Illyes said at a Google Search Central event that Google does not support llms.txt and is not planning to, and John Mueller separately compared it to the long-abandoned keywords meta tag, writing that none of the AI services have said they use it and that server logs show most AI crawlers do not even request it.

Mueller's framing is worth quoting directly because it captures the core objection: llms.txt is "what a site-owner claims their site is about," and Google would rather read the site itself than trust a self-reported summary, the same reasoning that made the keywords meta tag useless two decades ago. In a small, telling incident, an llms.txt file briefly appeared on Google's own Developer Docs site in December 2025 through an internal content system, then was pulled the same day, which industry coverage treated as accidental rather than a signal of new support.

OtterlyAI's independent server log study backs up Mueller's "crawlers don't check for it" observation with numbers. Over a 90-day window, the site tracked more than 62,100 AI bot visits and found only 84 of them, about 0.1%, requested llms.txt, roughly three times worse engagement than an average content page received. AI crawlers are overwhelmingly reading standard web pages, not the summary file built for them.

Does llms.txt help with AI coding tools like Cursor and Claude Code?

The clearest practical use for llms.txt today is developer tooling, not search visibility. AI coding assistants such as Cursor, GitHub Copilot, and Claude Code often fetch documentation directly while a developer works, and a well-structured llms.txt can point them to the right reference pages faster and cut down on wasted tokens compared to crawling an entire docs site. That is a real, narrow benefit, distinct from SEO or AI Overview citations, and it is the exact use case the original llmstxt.org proposal was built around.

If your site is developer-facing, a software product, an API, or an SDK with reference docs, this is the strongest case for adding llms.txt. If your site is a local service business, an e-commerce store, or a content publisher with no developer audience, the format has little to offer beyond a small, low-cost experiment.

llms.txt vs robots.txt vs sitemap vs schema markup

It helps to see llms.txt next to the technical files it gets compared to, since they solve different problems and have very different track records.

Should you still create an llms.txt file for your site?

Adding an llms.txt file is low cost, usually a 15 to 30 minute task, and there is no evidence it causes harm, so it is a reasonable low-priority item on a technical SEO checklist, particularly for developer-facing products with documentation that AI coding tools might fetch. It should never come before the work that has a documented connection to rankings and citations: crawlable architecture, accurate content, schema markup, and topical depth. Treat it the way you would treat a nice-to-have meta tag, not a launch blocker.

Skip it entirely if you are already stretched thin. The opportunity cost of an SEO team's time spent maintaining an unread file, instead of fixing crawl errors or writing answer-first content, is the real risk, not the file itself.

What actually helps you get cited by AI search engines?

The signals that SE Ranking and MaxAEO found do correlate with AI citation are the same ones that have always mattered for search: site authority, schema density, content recency, and answer-first structure. Our own AEO strategy framework runs on exactly those four levers, auditing AI visibility, mapping buyer questions, writing answer-first content, and building the entity and third-party authority models weigh before they name a source.

We have seen this play out directly with clients. When we rebuilt LiveHelpNow's content around answer-first structure and stronger schema rather than any file-based shortcut, the site gained roughly 3,000 additional monthly visits and started getting cited in AI Overviews, the kind of result the llms.txt studies above say the file alone cannot produce. For the full playbook behind that kind of work, see our answer engine optimization guide, and for a broader reference on where AI search is heading, our AI search manual covers the discipline end to end. If you want to see how the wider toolchain fits together, our roundup of AEO tools and our explainer on what AEO stands for are good next reads.

Common llms.txt mistakes to avoid

• Treating it as an SEO hack. No search engine or AI provider has confirmed a ranking or citation benefit, so budgeting real optimization time around it is misplaced.
• Skipping the fundamentals to add it. Content quality, schema, and crawlability all have a proven link to AI citation; llms.txt does not replace any of them.
• Auto-generating an inaccurate file. A summary that misrepresents your content can hurt trust with the small number of tools that do read it.
• Letting it go stale. Linking to moved or deleted pages defeats the file's one job, even for the developer tools that use it.
• Contradicting your own robots.txt. Listing pages in llms.txt that are blocked from crawling elsewhere sends mixed signals to any system that does check both.
• Expecting a fast ranking change. Both major 2026 studies found essentially no measurable movement, so treat any claimed llms.txt success story with skepticism until it names its data.

Frequently asked questions

What is an llms.txt file? An llms.txt file is a plain Markdown document placed at a website's root, such as yoursite.com/llms.txt, that summarizes the site and links to its most important pages for AI systems to read. Developer Jeremy Howard proposed the format in September 2024 through llmstxt.org as a way to give AI agents concise, expert-level information in one place instead of forcing them to parse full HTML pages built for humans.

Does llms.txt improve SEO rankings? No credible evidence shows llms.txt improves search rankings. Google's Gary Illyes confirmed in mid-2025 that Google does not read or act on llms.txt for Search and has no plans to, and it has never been documented as a ranking input for any major search engine.

Do ChatGPT, Google, and Claude actually read llms.txt files? As of Q1 2026, no major AI provider, including OpenAI, Google, Anthropic, Meta, or Mistral, has publicly confirmed that its production systems read or prioritize llms.txt. OtterlyAI's 90-day server log study found only 84 of over 62,100 AI bot visits touched a site's llms.txt file, about 0.1% of AI bot traffic.

How do you create an llms.txt file? Create a plain text file named llms.txt with an H1 title naming your site, a one-sentence blockquote summary, optional context paragraphs, and H2 sections listing your key pages as Markdown links with short descriptions. Save it at your domain root so it loads at yoursite.com/llms.txt, with no HTML or JavaScript.

Is llms.txt the same as robots.txt or a sitemap? No. robots.txt controls which paths crawlers may access and is honored by essentially every major crawler. An XML sitemap lists indexable URLs for search engines to discover. llms.txt is a curated summary aimed at AI agents, and unlike the other two, no major search engine or AI provider has confirmed it reads or acts on it.

Does llms.txt help with AI Overviews or Google's AI Mode? No. Google has stated on the record that llms.txt plays no role in AI Overviews or AI Mode. John Mueller compared it to the old keywords meta tag, since it is a page owner's own claim about their content rather than something Google verifies by reading the site directly.

Is llms.txt useful for anything? Its clearest practical use is developer tooling rather than search visibility. AI coding assistants such as Cursor, GitHub Copilot, and Claude Code can fetch documentation directly while a developer works, and a well-structured llms.txt can point them to the right reference pages and reduce wasted tokens, which is the use case llmstxt.org was originally built around.

Should every website add an llms.txt file? It is low cost and unlikely to cause harm, so adding one is a reasonable low-priority task, especially for developer-facing products with API or SDK documentation. It should never come before content quality, schema markup, crawlability, and topical authority, which have a proven connection to both search rankings and AI citations.

What actually helps you get cited by AI search engines? Answer-first content that states a fact clearly in the opening sentences, valid schema markup such as FAQPage and Article, demonstrated topical authority across a site, and third-party citations from sources AI models already trust. These are the same signals SE Ranking and MaxAEO found actually correlate with AI citation, unlike llms.txt.

What to do next

Spend your next hour on the things with proven impact: check that your priority pages answer their target question in the first sentences, verify your FAQPage and Article schema validate, and confirm your important content is actually crawlable. Add an llms.txt file afterward if you have five spare minutes and a developer audience, not as a substitute. If you want a clear picture of where your AI visibility gaps actually are, request a free SEO audit from Rankite and we will show you which signals are worth your time.