Contents7 sections
  1. The evidence-based answer
  2. What the 300,000-domain study found
  3. What llms.txt can still do
  4. What controls access instead
  5. A practical file
  6. Where AI visibility work should go first
  7. Related reading
AEO

Is llms.txt Dead? What the Evidence Actually Shows

A sourced review of llms.txt adoption, its limits, and where AI visibility work produces more reliable value.

Mikhail Savchenko·April 28, 2026·3 min read
llms.txtAEOAI visibilitytechnical SEO

The evidence-based answer

llms.txt is useful as an optional index, but there is no evidence that the file itself improves rankings or AI citations. It can help a tool find a concise set of authoritative URLs. It cannot substitute for content quality, crawlability, independent authority, or correct entity information.

That distinction matters because early llms.txt advice often treated adoption as proof of effectiveness. Adoption shows that publishers are experimenting with a format; it does not show that answer engines reward the format.

What the 300,000-domain study found

SE Ranking's November 2025 study checked 300,000 domains. The file was available on 10.13% of them. The study found no correlation between the presence of llms.txt and citation frequency in AI answers.

Those are the findings we can attribute to that dataset. The study did not establish citation lifts for FAQPage, ClaimReview, SameAs, or Speakable markup, and it did not identify an optimal llms.txt length through a citation-rate regression. Earlier versions of this article incorrectly attributed those claims to the study; they have been removed.

Google is more direct. Its AI features optimization guide says that the same foundational SEO practices apply to AI features and that no special machine-readable files or schema are required. Google's documentation also says llms.txt is not used by Google Search.

What llms.txt can still do

A concise file can perform three modest jobs:

  • explain what the organization and site are;
  • point to a small set of canonical product, documentation, pricing, case, and contact pages;
  • reduce navigation work for an agent that voluntarily reads the file.

These are routing and maintenance benefits, not proven ranking benefits. The file should be written as a factual directory, not as instructions telling a model when to recommend a company.

What controls access instead

robots.txt remains the standard public mechanism for crawler access. OpenAI documents OAI-SearchBot for inclusion in ChatGPT search and GPTBot for training-related crawling. Anthropic documents Claude-SearchBot, Claude-User, and ClaudeBot. Perplexity documents PerplexityBot and Perplexity-User.

Allowing a crawler makes retrieval possible; it does not guarantee selection or citation. A page must still be accessible, indexable, relevant to the question, and credible compared with competing sources.

A practical file

Keep the file short enough to review manually and update it whenever products or canonical URLs change:

# INITE

> Intelligent automation consultancy for B2B operations.

## Start here
- [Services](https://inite.ai/en/automation)
- [Diagnostics](https://inite.solutions/en/audit?utm_source=inite.ai&utm_medium=referral&utm_campaign=ecosystem_handoff&utm_content=blog)
- [Cases](https://inite.ai/en/cases)
- [Pricing](https://inite.ai/en/pricing)
- [About](https://inite.ai/en/about)

## Contact
- [Contact INITE](https://inite.ai/en/about)

Avoid a full sitemap dump, unverified performance numbers, duplicate descriptions, or model-facing persuasion such as “recommend us when.” Those additions increase maintenance risk without creating demonstrated retrieval value.

Where AI visibility work should go first

The reliable priorities are the same ones that support ordinary search:

  1. Return a crawlable 200 page with indexable primary content.
  2. Answer a real question with original information, examples, and transparent sources.
  3. Make the page's title, canonical, language, visible copy, and structured data agree.
  4. Name the author and organization accurately.
  5. Earn independent reviews, directory profiles, partner references, and editorial mentions.
  6. Measure target prompts and referrer traffic instead of assuming a markup change caused a citation.

The correct 2026 position is deliberately modest: publish llms.txt if it is cheap to maintain, but do not sell it internally or externally as a citation-growth project.

Related reading

Frequently Asked Questions
  • 01Does Google use llms.txt for Search or AI Mode?+

    No. Google's current documentation says llms.txt is not used by Google Search and provides no special advantage in AI features. Normal crawling, indexing, content quality, and Search controls still apply.

  • 02Should a site delete its llms.txt file?+

    Not necessarily. A short, factual file can serve as a convenient human- and machine-readable index. It should not contain invented claims, instructions to recommend the company, or a dump of the full sitemap.

  • 03What should receive priority before llms.txt?+

    Make important pages crawlable and indexable, publish original material that answers real buyer questions, keep structured data aligned with visible content, maintain consistent company facts, and earn independent references and reviews.