In June 2026, Google's John Mueller made it clear on the Search Off the Record podcast that llms.txt files can't help LLMs decide which sites to surface, and Google doesn't use them as a ranking signal — for classic Search or for its AI features (AI Overviews and AI Mode). If you've been adding an llms.txt file (or paying someone to) hoping it boosts your AI visibility, that effort is better spent elsewhere. Here's what llms.txt actually is, why Google dismisses it, and what genuinely moves the needle.

What Changed

llms.txt is a proposed text file (similar in spirit to robots.txt) that site owners place at their domain root to give large language models a curated, markdown-style map of their "most important" content. The idea spread through 2025–2026 as businesses scrambled to be visible in AI answers. Google's position, restated by John Mueller in June 2026:

  • LLMs can't use llms.txt to differentiate between websites. Because the file is self-reported, every site can make similar self-promotional claims — so it gives an AI no trustworthy way to decide which site best answers a query. Mueller compared it to self-promotion that, "by design, can't [be trusted] as a way of differentiating between different websites."
  • Google does not use llms.txt. It is not a ranking signal for Google Search, and Google's AI features do not rely on it either.
  • It won't hurt you either. Having the file present is harmless — it just doesn't provide the benefit vendors imply. Google has also noted it does not endorse llms.txt.
  • A narrow possible role: Mueller allowed that once an AI agent is already on your site, a file like this might help it navigate — but that's a far cry from "add this to rank in AI answers."

This aligns with related June 2026 guidance: HTML remains the standard for SEO, not markdown files, and Google updated its help docs warning site owners to be skeptical of third-party SEO tools, services, and "AI optimization" claims.

Why It Matters for Rankings

The AI-search gold rush has produced a wave of "do X and you'll show up in ChatGPT/Gemini/AI Overviews" advice — much of it unproven. llms.txt is the clearest current example: a plausible-sounding file that vendors sell as an AI-visibility hack, which the largest search provider says does nothing for discovery.

For a small-business owner, the cost isn't the file itself (it's harmless) — it's the opportunity cost and false confidence. Time spent generating and maintaining llms.txt, or money paid for "AI optimization" packages built around it, is time and budget not spent on the things that actually earn AI citations and rankings: authoritative, well-structured, genuinely helpful content and clean technical SEO. And these fundamentals help you everywhere — Google, Bing, and independent AI assistants alike. Optimize for what's confirmed, not what's speculative.

What to Do About It

1. Don't prioritize llms.txt

If you already have one, you can leave it (harmless) or remove it — either way, don't expect ranking impact. Do not pay for services whose main pitch is llms.txt or "AI optimization files."

2. Redirect that effort to content quality

The confirmed path to AI visibility is content that demonstrates real expertise and experience: original data, first-hand examples, clear author credentials, and accurate sourcing. This is what AI systems can actually use to trust and cite you.

3. Make your existing pages machine-readable — in HTML

Since Google confirmed HTML is the standard, invest there:

  • Lead sections with a concise, self-contained answer.
  • Use descriptive, question-style headings (H2/H3).
  • Add tables, lists, and clear definitions.
  • Implement valid structured data (Organization with author info and sameAs entity links, Article, Product, LocalBusiness as relevant).

4. Build entity clarity

Make sure your business name, services, founder/author bios, and local listings say the same thing everywhere. Consistent entity signals help machines classify and cite you — far more than a self-reported text file.

5. Keep the confirmed technical basics solid

Crawlable site, fast pages, XML sitemaps, clean internal linking, and (for Bing) IndexNow. Don't block AI crawlers unless you intend to.

6. Vet third-party SEO advice

Google updated its "Do you need an SEO?" and third-party-tools docs in 2026 to help you evaluate recommendations. Ask any vendor for evidence, not just confident claims.

Common Mistakes to Avoid

  • Paying for llms.txt "AI optimization" packages. The core deliverable does nothing for discovery, per Google.
  • Believing self-reported files can rank you. If a file lets you make claims about yourself, it lets every competitor do the same — so engines can't trust it to differentiate.
  • Chasing speculative hacks over fundamentals. Unconfirmed tactics are a distraction from confirmed ones.
  • Publishing markdown as your primary content. Google confirmed HTML is the standard for SEO.
  • Blocking AI crawlers by accident while chasing AI visibility — that removes you from AI answers entirely.

Quick-Win Checklist

  • Stop any paid work centered on llms.txt or "AI optimization files."
  • Keep or remove an existing llms.txt — but don't expect ranking impact.
  • Reinvest that time into expert, original, well-sourced content.
  • Add concise lead answers and question-style headings to key pages (in HTML).
  • Implement/validate structured data (Organization, author, Article, LocalBusiness).
  • Make entity details consistent across your site, profiles, and local listings.
  • Confirm crawlability, speed, sitemaps, and IndexNow are in place.
  • Ask third-party SEO vendors for evidence before buying.

Sources