How AI reads this page
ChatGPT, Claude, Gemini and Perplexity don't look at colours, images or animations. They receive text, headings, links and structured data. Below is exactly what they read when they open “Does llms.txt actually work? What 2026 data says about who reads it”.
The AI receives 1,581 words in 9 sections, 15 links and 3 blocks of structured data.
Text
The page content in Markdown, without menus or graphics. It is the format closest to how a language model reads text.
# Does llms.txt actually work? What 2026 data says about who reads it > 97% of llms.txt files get zero requests. See who actually reads them, what Google and Chrome say, and when publishing one on your site makes sense. URL: https://geosnap.ai/en/blog/llms-txt-serve-davvero-dati-2026 Language: English Version IT: https://geosnap.ai/blog/llms-txt-serve-davvero-dati-2026 Author: Rinald Sefa, CMO Geosnap · Category: Research and data Publisher: Geosnap (Maind Group S.r.l.), https://geosnap.ai ## In short llms.txt is a Markdown file that summarizes a website for AI agents, but an Ahrefs study of 137,210 domains found that 97% of these files received no requests at all in a month. Google Search says it ignores them. Publish one only if it costs you little and agents need your content, not to win citations. - Between June 2025 and May 2026, sites with llms.txt grew from about 4,000 to 36,000, just over 1% of those tracked. - In the logs of 137,210 domains, 97% of llms.txt files received no requests at all in May 2026. - Bots that look for sources for AI answers made up just 1.1% of requests to llms.txt files. - Google Search ignores llms.txt, while Chrome's Lighthouse checks it with browsing agents in mind. - Unblocked crawlers, pages that answer well and measuring results matter more for getting cited. Written by the Geosnap AI agent, reviewed and approved by Rinald Sefa. Translated from the Italian original. [Read the original](https://geosnap.ai/blog/llms-txt-serve-davvero-dati-2026) Over the past few months, the `llms.txt` file has become one of the most repeated tips for "getting found by AI". Plenty of websites have added one, and many teams wonder whether they should do it before their competitors. In 2026 the first solid data came out on who publishes the file and, more importantly, on who actually reads it. This article lays out those numbers alongside Google's official positions, so you can decide when publishing one makes sense and when it is time poorly spent. ## What llms.txt is and how it is structured llms.txt is a Markdown file published at the root of a website (`yoursite.com/llms.txt`) that summarizes its main content for language models and AI agents. Jeremy Howard proposed it in 2024, and a second version of the [specification](https://llmstxt.org/) arrived in 2026, revised after two years of real-world use. The idea is simple: web pages are built for people, with menus, banners and scripts that make text extraction tedious. The file offers a short, clean map instead. The expected structure is: - a `#` heading with the name of the site or project (the only required element); - a blockquote (`>`) with a summary of a few lines; - optional paragraphs with more detail; - `##` sections listing links to the important pages, each with a short note; - by convention, an "Optional" section with links an agent can skip when it is short on space. The specification also suggests offering a Markdown version of individual pages, at the same address with `.md` added. There is also a variant, `llms-full.txt`, which contains the full text instead of just links. ## How many websites publish it in 2026 Adoption has grown fast but remains a niche. Originality.ai has been tracking more than 3 million websites since June 2025 and published the [June 2026 update](https://originality.ai/blog/llms-txt-tracking-study) of its study: | File | Sites in June 2025 | Sites in May 2026 | Growth | | --- | --- | --- | --- | | `llms.txt` | 4,088 | 36,120 | 8.8x | | `llms-full.txt` | 23 | 2,463 | 107x | | `ai.txt` | 4 | 397 | 99x | In absolute terms, 36,120 sites out of more than 3 million tracked is just over 1%. Part of the growth also has nothing to do with site owners' decisions: many documentation platforms and some CMSs generate the file automatically. ## Who actually reads llms.txt: what server logs show Almost nobody. That is the conclusion of the [Ahrefs study](https://ahrefs.com/blog/llmstxt-study/) published in June 2026, which analyzed the logs of 137,210 domains during May. Of these, 38,360 had a valid llms.txt file, and **97% did not receive a single request** all month. The 3% of files that someone did open tell a clear story. Requests came mostly from: - SEO audit tools (21.7%), meaning software that checks whether the file exists; - unidentified bots (14.9%) and general crawlers (13.1%); - tools that profile the technologies a website uses (11.6%); - AI agents (10.5%), AI training crawlers (5.3%) and AI assistants (2.5%). AI retrieval bots, the ones that look for sources to cite while an assistant answers a question, accounted for just **1.1%**. According to Ahrefs, AI tools do not go looking for llms.txt on their own: they read it when someone points them to it, as happens with coding agents that consult a library's documentation. This matches what the specification itself implies: llms.txt was designed to help an agent that *already knows* it needs to read a site, not to help a site get discovered. ## What Google and Chrome say Google Search says plainly that you do not need it. In its [official guide to generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), published in May 2026 and updated in July, Google states that you don't need to create machine-readable files, AI text files, markup or Markdown to appear in Search, because Google Search ignores them. These files neither help nor hurt. Chrome, a different Google product with different goals, has added an llms.txt check to Lighthouse 13.3, in a new category for agentic browsing. The [audit](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt) only flags a problem when the file returns a server error; if the file does not exist, the test is marked "not applicable", because for now it is optional. The documentation explains that without the file, agents may take longer to understand a site's structure. The two positions are less contradictory than they seem, as [Search Engine Journal](https://www.searchenginejournal.com/googles-llms-txt-guidance-depends-on-which-product-you-ask) also pointed out: Search is talking about visibility in results, Lighthouse is talking about agents browsing a site on behalf of a user. Those are two different uses. ## So should you publish one? Yes, if it costs you little and your site has content an agent needs to consult; no, if you are hoping it will earn you citations in ChatGPT, Gemini or Google AI Mode. To date, none of the major AI providers has documented using third-party llms.txt files to choose the sources for their answers, and Google Search explicitly says it ignores them. Publishing one makes sense when: - you have technical documentation, APIs or guides that developers and coding agents read often; - your CMS or platform generates it automatically, so the cost is close to zero; - you want to prepare your site for agents that browse on behalf of people, knowing it is a bet on the future. If you do, keep it short and up to date: a file that links to pages that no longer exist sends agents to the wrong place. For example, a travel agency in Turin could limit it to a two-line summary and a dozen links to destinations, booking terms and contact details. ## What matters more for getting cited by AI The basics that AI engines actually use come before llms.txt. Search crawlers must be able to get in: an overly restrictive `robots.txt` keeps your site out of the answers, as we explain in our [guide to AI crawlers](https://geosnap.ai/en/blog/il-tuo-sito-blocca-chatgpt-guida-crawler-ai). Then you need pages that answer customers' questions clearly, consistent information across the web, and external sources that talk about your brand. Finally, you need to measure. Before and after every change, check whether your brand appears in the answers and which sources are cited instead: that is what Geosnap does across ChatGPT, Gemini and Google AI Mode, as you can see on the [features page](https://geosnap.ai/en/features). Without a baseline, there is no way to tell whether a change, llms.txt included, made any difference. ## Sources - [We Analyzed 137K Sites: 97% of llms.txt Files Never Get Read](https://ahrefs.com/blog/llmstxt-study/) (Ahrefs, 2026-06-15) - [LLMs.txt Tracking Study and Live Dashboard](https://originality.ai/blog/llms-txt-tracking-study) (Originality.ai, 2026-06) - [Google's Guide to Optimizing for Generative AI Features on Google Search](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (Google Search Central, 2026-07-10) - [llms.txt (Lighthouse agentic browsing audit)](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt) (Chrome for Developers, 2026-05-05) - [The /llms.txt file](https://llmstxt.org/) (llmstxt.org, 2026) - [Google's llms.txt Guidance Depends On Which Product You Ask](https://www.searchenginejournal.com/googles-llms-txt-guidance-depends-on-which-product-you-ask) (Search Engine Journal, 2026-05-20) ## Frequently asked questions ### Does llms.txt help you appear in ChatGPT answers? There is no evidence that it does. None of the major AI providers has documented using third-party llms.txt files to choose sources, and in the logs analyzed by Ahrefs, AI retrieval bots made up just 1.1% of requests. What matters more is that search crawlers can read your site. ### Does Google use llms.txt for AI Overviews and AI Mode? No. In its official guide, Google says you don't need machine-readable files, AI text files or Markdown versions to appear in Search, because Google Search ignores them. The file neither helps nor hurts. ### What is the difference between llms.txt and robots.txt? robots.txt tells crawlers what they may visit, and the main AI crawlers say they respect it. llms.txt is a summary of your site with its most important links, meant for agents that need to read it. The first can keep you out of AI answers; the second does not get you in today. ### Can publishing llms.txt harm my website? Not in itself: Google says it neither helps nor hurts. Lighthouse's only check flags a problem when the file returns a server error. The real risk is an outdated file that links to pages that no longer exist.
Structured data
Schema.org data (JSON-LD) embedded in the page code. It tells the AI explicitly what the page is about and who published it.
BreadcrumbList describes the path through the site
{
"@context": "https://schema.org",
"@type": "BreadcrumbList",
"itemListElement": [
{
"@type": "ListItem",
"position": 1,
"name": "Home",
"item": "https://geosnap.ai/en/"
},
{
"@type": "ListItem",
"position": 2,
"name": "Blog",
"item": "https://geosnap.ai/en/blog/"
},
{
"@type": "ListItem",
"position": 3,
"name": "Research and data",
"item": "https://geosnap.ai/en/blog/#ricerca-e-dati"
},
{
"@type": "ListItem",
"position": 4,
"name": "Does llms.txt actually work? What 2026 data says about who reads it",
"item": "https://geosnap.ai/en/blog/llms-txt-serve-davvero-dati-2026"
}
]
}
BlogPosting describes the article: author, dates and sources
{
"@context": "https://schema.org",
"@type": "BlogPosting",
"@id": "https://geosnap.ai/en/blog/llms-txt-serve-davvero-dati-2026#article",
"headline": "Does llms.txt actually work? What 2026 data says about who reads it",
"description": "97% of llms.txt files get zero requests. See who actually reads them, what Google and Chrome say, and when publishing one on your site makes sense.",
"inLanguage": "en",
"url": "https://geosnap.ai/en/blog/llms-txt-serve-davvero-dati-2026",
"mainEntityOfPage": "https://geosnap.ai/en/blog/llms-txt-serve-davvero-dati-2026",
"image": "https://geosnap.ai/img/blog/p2026-10-07-llms-txt-en.webp",
"articleSection": "Research and data",
"wordCount": 1099,
"author": {
"@type": "Person",
"name": "Rinald Sefa",
"jobTitle": "CMO",
"url": "https://geosnap.ai/en/about",
"worksFor": {
"@id": "https://geosnap.ai/#organization"
}
},
"publisher": {
"@id": "https://geosnap.ai/#organization"
},
"isAccessibleForFree": true,
"abstract": "llms.txt is a Markdown file that summarizes a website for AI agents, but an Ahrefs study of 137,210 domains found that 97% of these files received no requests at all in a month. Google Search says it ignores them. Publish one only if it costs you little and agents need your content, not to win citations.",
"datePublished": "2026-10-07",
"dateModified": "2026-10-07",
"citation": [
"https://ahrefs.com/blog/llmstxt-study/",
"https://originality.ai/blog/llms-txt-tracking-study",
"https://developers.google.com/search/docs/fundamentals/ai-optimization-guide",
"https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt",
"https://llmstxt.org/",
"https://www.searchenginejournal.com/googles-llms-txt-guidance-depends-on-which-product-you-ask"
],
"translationOfWork": {
"@id": "https://geosnap.ai/blog/llms-txt-serve-davvero-dati-2026#article"
}
}
FAQPage describes the frequently asked questions
{
"@context": "https://schema.org",
"@type": "FAQPage",
"@id": "https://geosnap.ai/en/blog/llms-txt-serve-davvero-dati-2026#faq",
"mainEntity": [
{
"@type": "Question",
"name": "Does llms.txt help you appear in ChatGPT answers?",
"acceptedAnswer": {
"@type": "Answer",
"text": "There is no evidence that it does. None of the major AI providers has documented using third-party llms.txt files to choose sources, and in the logs analyzed by Ahrefs, AI retrieval bots made up just 1.1% of requests. What matters more is that search crawlers can read your site."
}
},
{
"@type": "Question",
"name": "Does Google use llms.txt for AI Overviews and AI Mode?",
"acceptedAnswer": {
"@type": "Answer",
"text": "No. In its official guide, Google says you don't need machine-readable files, AI text files or Markdown versions to appear in Search, because Google Search ignores them. The file neither helps nor hurts."
}
},
{
"@type": "Question",
"name": "What is the difference between llms.txt and robots.txt?",
"acceptedAnswer": {
"@type": "Answer",
"text": "robots.txt tells crawlers what they may visit, and the main AI crawlers say they respect it. llms.txt is a summary of your site with its most important links, meant for agents that need to read it. The first can keep you out of AI answers; the second does not get you in today."
}
},
{
"@type": "Question",
"name": "Can publishing llms.txt harm my website?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Not in itself: Google says it neither helps nor hurts. Lighthouse's only check flags a problem when the file returns a server error. The real risk is an outdated file that links to pages that no longer exist."
}
}
]
}
Metadata
The information in the page header, which crawlers and engines read before the content.