llms.txt is a small Markdown file at your domain root that hands AI tools a clean, curated map of your most important content. It takes about 20 minutes to write, it costs nothing, and it's worth doing — but not for the reason most guides claim. Here's the honest version.
Cited by AI or invisible? llms.txt is one piece of the puzzle. Our AI visibility audit checklist scores all 12 signals that decide whether ChatGPT, Perplexity, and Gemini quote your business.
What is llms.txt?
llms.txt is a plain-text Markdown file served at yoursite.com/llms.txt. Proposed by Anthropic in late 2024, it gives language models a structured summary of your site: who you are, what you do, and which pages matter most — organized in sections with short descriptions per link.
Think of it as a sitemap written for models instead of crawlers. Where an XML sitemap lists every URL for indexing, llms.txt curates: it says "if you're going to read anything here, read these ten pages." For AI tools with limited context windows, that curation is the point — it raises the odds that the content a model actually ingests is your best material, not a random tag archive.
The format is deliberately simple:
## Your Brand Name
> One paragraph: what you do, who you serve, where you operate.
## Services
- [Service one](https://yoursite.com/services/one): one-line description of who it's for.
- [Service two](https://yoursite.com/services/two): one-line description of who it's for.
## Locations
- [City page](https://yoursite.com/locations/city): what the page covers.
## Guides
- [Your best guide](https://yoursite.com/blog/best-guide): the question it answers.
## Contact
- Email, phone, and a link to your contact page.
That's it. An H1 with your brand, a blockquote summary, H2 sections, and links with descriptions. No JSON, no build step, no plugin required. How models discover and use the file is part of the wider AI visibility audit; for the citation strategy around it, read how to get ChatGPT to recommend your business.
Do you actually need an llms.txt file?
Here's the honest answer most guides skip: for Google's AI features, no. Google has said it may crawl llms.txt but gives it no special treatment — an AI Overview won't cite you just because the file exists. Some non-Google AI tools do consult it, and it's nearly free to add, so there's little reason to skip it. But treat it as hygiene, not strategy.
The real value is indirect and almost nobody talks about it: writing the file forces a content-and-crawler audit. To write a good llms.txt, you have to decide which pages actually deserve to be cited — which usually reveals you have no canonical service page, three competing versions of it, or a robots.txt quietly blocking GPTBot. The file is the artifact; the audit is the value.
Context on why this matters now: BrightEdge tracking shows AI Overviews triggering on roughly 46–48% of Google searches (up 58% year over year), and Google's AI Mode alone reached 75 million daily active users. Gartner projected 25% of search traffic would come from non-traditional search by 2026. When half of searches involve an AI answer, making your best content trivially easy for models to find is basic maintenance. Run the full 12-point AI visibility audit to see where the file sits among your other gaps — a technical SEO audit will surface the crawler blocks that writing it tends to reveal.
How do you write an llms.txt file?
Follow this five-step method:
1. Write the summary as if explaining to a stranger. One paragraph: what the business does, who it's for, where it operates. Skip marketing adjectives; models (and readers) need facts. "WebStackRank builds SEO-first websites and runs no-retainer SEO for US service businesses" beats "cutting-edge digital solutions."
2. Link only pages you'd want cited. Five to fifteen links, tops. Every link should point to a page with a clear, quotable answer — service pages, location pages, your best guides. If a page is thin or outdated, fix the page first; don't launder it through llms.txt.
3. Group by topic, not by URL. Sections like Services, Locations, Guides, Pricing, and Contact mirror how buyers think — and how answer engines slice queries. A flat list of 40 links is a directory, not a map.
4. Write factual one-line descriptions. Each link gets a short note about what the page covers and who it's for. Avoid keyword stuffing; the descriptions exist to help a model pick the right page, not to game anything.
5. Serve it as plain text at the root. yoursite.com/llms.txt should return the Markdown with a text/plain (or text/markdown) content type. Add it to your sitemap index or link it from your HTML head — discoverability helps the tools that do look for it.
Keep the file current. Every time you publish a flagship page or retire a service, update the file. An llms.txt pointing at dead or redirected URLs actively misleads the tools it's meant to help. Teams scaling this across many pages automate it with custom GPT and AI builds; our services overview maps where the file fits inside a full AI-visibility engagement.
What mistakes break llms.txt files?
Three failures account for nearly all of them:
Treating it as a ranking hack. It isn't one. No AI engine ranks you higher for having the file. If someone sells you "llms.txt optimization" as a standalone service, they're selling the envelope, not the letter.
Letting it go stale. An outdated llms.txt is worse than none — it steers models toward retired pages and wrong contact details. Date the file (a comment at the bottom works) and tie updates to your content calendar.
Blocking the crawlers anyway. This is the killer. llms.txt is pointless if robots.txt blocks GPTBot, OAI-SearchBot, PerplexityBot, or ClaudeBot — a block there makes the file moot. Always check both together. Our technical SEO audit catches this class of silent misconfiguration routinely.
A fourth, subtler mistake: linking pages that can't be cited. A page that's slow, buried behind JavaScript rendering, or written as a wall of marketing copy won't survive contact with an answer engine. Pair the file with genuinely answer-first content — the kind we build through custom GPT and AI automation work for clients scaling their content.
Where does llms.txt fit in the bigger picture?
llms.txt is the smallest piece of AI visibility — the technical floor, not the building. Consider the numbers: only 17% of AI Overview citations come from pages ranking in the organic top 10 (BrightEdge), which means citation is won on answer quality, original data, and entity corroboration — not file presence. And AI Mode queries run about 3× longer than traditional searches (Alphabet Q4 2025 earnings), which rewards pages that answer complex, multi-part questions rather than single keywords.
The hierarchy that actually gets you cited:
- Answer-first, fact-rich content — quotable passages with real numbers. See how to get ChatGPT to recommend your business.
- Entity clarity — Organization schema, consistent NAP, sameAs links.
- Independent corroboration — reviews, roundups, community mentions.
- Technical floor — open crawlers, fast semantic pages, and yes, llms.txt.
If you want the full method implemented rather than DIY'd, our no-retainer SEO service covers it end to end — or run the 12-point AI visibility audit yourself first and see where you land.
Frequently asked questions
Does llms.txt help with ChatGPT or Google AI Overviews?
Partially. Google gives llms.txt no special treatment for AI features — it may crawl the file but won't rank or cite you because of it. Some non-Google AI tools do use it as a discovery shortcut. Its biggest proven benefit is forcing the content-and-crawlability audit that writing it requires.
Where exactly do I put the llms.txt file?
At your domain root, served as plain text at yoursite.com/llms.txt. Make sure it's reachable over HTTPS, returns a 200 status, and isn't blocked by robots.txt. Optionally reference it from your XML sitemap for extra discoverability.
Is llms.txt the same as robots.txt?
No. robots.txt controls which crawlers may access which parts of your site; llms.txt suggests which content matters most. robots.txt is the gatekeeper and must be correct first — a crawler block there makes llms.txt completely moot.
How often should I update llms.txt?
Every time you publish a flagship page, retire a service, or change contact details. At minimum, review it quarterly alongside your AI visibility audit. Add a "last updated" date at the bottom of the file so staleness is visible.
Can llms.txt hurt my SEO?
Not directly — it's a tiny, optional file. The only real risk is pointing it at thin, outdated, or broken pages, which can mislead the AI tools that read it. Curate ruthlessly: five excellent links beat twenty mediocre ones.
Should small businesses bother with llms.txt?
Yes, because it's 20 minutes of work with no downside. But don't stop there — for a local business, consistent entity data and reviews will move AI citations far more than the file alone. Our local SEO ranking system covers the higher-leverage local work.
What should I do right after publishing llms.txt?
Verify it loads at the root URL, confirm robots.txt allows AI crawlers, check Bing Webmaster Tools indexing (ChatGPT's search leans on Bing), and then move up the hierarchy: answer-first content, entity schema, and independent mentions. The file is step one of twelve.
Twenty minutes, zero cost, no downside. Publish your llms.txt today — then talk to our team about the other eleven checks that actually get you cited.
