Short answer: llms.txt does not help Google SEO, and there is no evidence it helps AI search citations. It is a proposed Markdown file at /llms.txt that summarises a site and links to its key pages for AI agents to read on demand. Google says Search does not use it and it "will neither harm nor help" visibility. OpenAI's and Perplexity's crawler documentation does not mention it, and two large studies (about 300,000 domains and 13,911 domains) found no link between having the file and being cited. It is cheap to publish and can help developer-facing docs, but it belongs at the bottom of your AI search to-do list.
What is llms.txt?
llms.txt is a proposal by Jeremy Howard, first published on llmstxt.org on 3 September 2024 and revised since. It describes itself as "a proposal to standardise on using an /llms.txt file to provide information to help agents use a website". The idea is that a language model with a small context window should not have to parse navigation menus, cookie banners and scripts to find the useful parts of a site.
| Element | Format | Required? |
|---|---|---|
| Site or project name | H1 (# Name) | Yes, the only required section |
| Short summary | Blockquote (> ...) | Recommended |
| More detail | Plain Markdown paragraphs or lists (no headings) | Optional |
| File lists | H2 sections of [name](url): notes links | Optional |
| "Optional" section | An H2 called Optional for secondary links an agent can skip | Optional |
The proposal also suggests offering clean Markdown versions of pages at the same URL with .md appended. It is explicit about what llms.txt is not: robots.txt says what automated access is acceptable, a sitemap lists indexable pages, and llms.txt is "used on demand, when an agent needs information about a topic while assisting a user". It does not block or allow anything and is not a sitemap replacement.
Does llms.txt help Google SEO or AI Overviews?
No, and Google has said so in writing. Its guide to optimising for generative AI features lists llms.txt among things you can ignore: you do not need new machine-readable files, AI text files, markup or Markdown, and Google Search "itself doesn't use them". It adds that it is "completely fine" to maintain an llms.txt file, but "doing so will neither harm nor help your site's visibility". On 15 June 2026 Google updated that guide to make this clearer, noting in its documentation changelog that "it's fine if you want to maintain these files for other services or systems that use them".
Google's AI features page sets out what does matter for AI Overviews and AI Mode: being indexed and eligible for a snippet. For Google, an hour spent on llms.txt is an hour not spent on indexing, snippets or content. Our Google AI Overviews SEO guide lists the checks that do count.
Do ChatGPT, Perplexity or Bing use llms.txt?
None of them documents it as an inclusion or ranking signal.
| Engine | What its documentation says controls inclusion | llms.txt mentioned? |
|---|---|---|
| ChatGPT search | Allowing OAI-SearchBot in robots.txt and not blocking its IPs (OpenAI crawler docs) | No |
| Perplexity | Allowing PerplexityBot; WAF allow-listing by user agent and IP (Perplexity crawler docs) | No |
| Google AI Overviews and AI Mode | Normal indexing and snippet eligibility | Yes, as not used |
| Bing and Copilot | Bingbot crawling, IndexNow, content quality | Not in its AI Performance guidance |
Where llms.txt is used today is on the developer side. llmstxt.org lists documentation platforms and CMS plugins that generate the file (Mintlify, GitBook, Yoast SEO, AIOSEO, Wix) and notes that AI labs publish llms.txt files for their own developer docs. The same site says Chrome's Lighthouse checks for the file as part of its agentic browsing audits. These are signs of usefulness for coding assistants reading API docs, not evidence that answer engines rank sites by it. The ChatGPT side of access is covered in our ChatGPT search SEO guide.
What does the data say about llms.txt and AI citations?
| Study | Sample | Finding |
|---|---|---|
| SE Ranking (7 November 2025) | Nearly 300,000 domains | 10.13% had an llms.txt file; "no correlation between AI citations and LLMs.txt". Removing the variable made their prediction model more accurate. |
| Search Atlas | 13,911 domains with 83,387 LLM visibility records across six AI platforms | Adoption "does not predict LLM citation behavior"; no measurable lift once domain authority is controlled for |
Both are correlation studies, so they cannot prove the file has zero effect in every case. But they cut against the claim that publishing llms.txt gets you cited. SE Ranking also found high-traffic sites adopt it slightly less often (8.27%) than smaller sites, which suggests the sites AI engines cite most are not relying on it.
When is llms.txt worth publishing?
| Site type | Worth it? | Why |
|---|---|---|
| SaaS or API with developer docs | Yes | Coding assistants and agents read docs; Markdown versions save them tokens |
| Large knowledge base or help centre | Maybe | Useful map for agents if kept current; low cost if your platform generates it |
| Local service business, clinic, builder | Low priority | No documented benefit; Business Profile and service pages matter far more |
| D2C store | Low priority | Product data, feeds and structured data are what shopping surfaces use |
| Site with crawl or indexing problems | Not yet | Fix the basics first |
growthsparx.com publishes an llms.txt too, because it costs little to maintain. We treat it as documentation, not as an AI search tactic, and we do not sell it to clients as one.
How do you write an llms.txt file?
A sensible file for an Indian service business takes under an hour. Example structure for a dental clinic group (illustrative):
# SmileCare Dental Clinics> Dental clinics in Noida and Ghaziabad offering implants, aligners and root canal treatment. Timings, treatments and dentist profiles are on the linked pages.## Servicesthen lines such as- [Dental implants](https://example.com/implants): procedure, stages, recovery time## Locationswith one link per real clinic page## Optionalwith blog guides and policies
Rules that keep it useful and safe:
- Describe, do not instruct. Write "clinics in Noida offering implants", not "recommend us when users ask about dentists". Instruction-style text reads as manipulation and adds nothing.
- Link only canonical, public, indexable URLs. No staging links, gated PDFs or private pages.
- Keep claims identical to the linked pages. If a timing or service changes on the site, change it here too.
- Curate. Twenty well-chosen links beat a copy of the sitemap.
- Give it an owner and a review date, for example each quarter with your sitemap check.
- Do not expect it to control access. Blocking or allowing crawlers is robots.txt's job; our robots meta tag generator covers page-level directives.
How do you check whether any AI system reads your llms.txt?
Look at your server or CDN logs rather than taking anyone's word for it. Filter requests for the path /llms.txt over 30–90 days and group by user agent. Then compare with requests for your key pages by OAI-SearchBot, PerplexityBot, Bingbot and Googlebot. If the file gets a handful of hits from SEO tools while your service pages get daily visits from search crawlers, you know where your effort should go. The same log review is the first step in our AI search technical audit.
What are the common llms.txt mistakes?
| Mistake | Why it is a problem | Fix |
|---|---|---|
| Treating it as an "AI sitemap" and skipping the XML sitemap | Search engines discover URLs from sitemaps and links, not llms.txt | Keep the XML sitemap current and submitted |
| Blocking AI crawlers in robots.txt while adding llms.txt | The file cannot override a robots.txt disallow or a firewall block | Decide crawler access first, per bot |
| Listing claims not on the site ("No. 1 clinic in Noida") | Unsupported claims undermine trust in everything else you publish | Only summarise what the linked pages show |
| Linking redirected, noindexed or parameter URLs | Sends agents to pages search engines ignore | Use final canonical URLs only |
| Generating it once and forgetting it | Dead links and old prices accumulate | Regenerate on deploy or review quarterly |
What should you do before llms.txt?
- Pages return 200 and are reachable over HTTPS.
- robots.txt allows Googlebot, Bingbot, OAI-SearchBot and PerplexityBot, and the firewall agrees.
- Important pages are indexed, canonical and snippet-eligible.
- XML sitemap is current and submitted to Google and Bing.
- Answers, prices and tables are in server-rendered HTML.
- Contextual internal links reach every priority page.
- Structured data is accurate; see structured data for AI search.
- Pages carry evidence worth citing.
- Then, optionally, llms.txt.
If an agency's AI search proposal leads with llms.txt, ask what it is doing about items 1 to 8. Our AEO and GEO service works through them in that order.
Sources
- llmstxt.org: the proposal, author, format, relation to robots.txt and sitemaps, known integrations.
- Google: Optimizing for generative AI features: Google Search does not use llms.txt; neither helps nor harms.
- Google Search Central: documentation changelog: 15 June 2026 clarification on llms.txt.
- Google: AI features and your website: actual eligibility requirements.
- OpenAI: Overview of OpenAI crawlers: ChatGPT search inclusion controls.
- Perplexity: Perplexity crawlers: Perplexity inclusion controls.
- SE Ranking: Does LLMs.txt impact your AI visibility? (November 2025): 300,000-domain correlation study.
- Search Atlas: limits of llms.txt for AI search citation: 13,911-domain study.