Data study · Machine-readable doors for AI agents

Only 0.12% of Top Homepages Link to a Markdown Version for AI Agents, and 77.2% of Those Sites Also Publish llms.txt

KGWritten by Kriszta GrenyoPublished October 20263 min read

951 of 772,412 top homepages point to a markdown copy of the page with a link rel="alternate" tag. 2,889 link to an llms.txt file from the page head.

Analysis of 772,412 top-1M homepages · HTTP Archive, October 2026 crawl

Key findings

  1. 1

    951 of 772,412 top-1M homepages (0.12%) include a link rel="alternate" tag with a markdown content type.

  2. 2

    734 of those 951 sites (77.2%) also serve llms.txt, compared with 14.3% of all homepages.

  3. 3

    2,889 homepages (0.37%) link to an llms.txt file from a link tag in the page head, and 2,739 of them (94.8%) actually serve the file.

  4. 4

    No top-1,000 homepage offers a markdown alternate; 863 of the 951 sit in the 100K to 1M ranks.

  5. 5

    TYPO3 (2.2%) and Astro (1.5%) sites are the most likely to offer markdown versions.

Summary

Picture a software company that notices AI coding assistants quoting its documentation with broken formatting, menu text mixed into the answers. A developer suggests publishing a clean markdown copy of every page and telling machines where to find it.

That is the idea behind markdown alternates. A page adds a link tag with rel="alternate" and type="text/markdown" that points to a plain markdown version. AI agents that know to look can fetch that instead of parsing HTML. It is a convention some developer tools and llms.txt advocates promote, not a formal standard from a search engine.

Few sites do it yet. 951 of the top 1 million homepages offer a markdown alternate, 0.12%. The sites that do are usually already invested in AI files, with 77.2% also publishing llms.txt.

What we measured

We used the HTTP Archive October 2026 mobile crawl (httparchive.crawl.pages) for top-1M homepages and read every link tag from the almanac link-nodes metric. A site counts when a link has rel containing alternate and a type containing markdown. We also counted link tags whose address points to llms.txt.

We joined these flags with llms.txt results from the same crawl. Platform comes from HTTP Archive's technology detection.

Markdown alternates are a long-tail experiment so far

0.12%of top-1M homepages link to a markdown version

Counts by popularity rank.

Markdown alternate links by rank bucket
Rank bucketHomepagesWith markdown alternate% of homepages
Top 1K73800
1K-5K2,92050.17
5K-10K3,58670.2
10K-50K28,831330.11
50K-100K36,937430.12
100K-500K306,0483700.12
500K-1M393,3524930.13

Source: aisearchoptimization.net analysis of 772,412 top-1M homepages · HTTP Archive, October 2026 crawl

No bucket goes above 0.2% of homepages, and the top 1,000 has none.

The same sites tend to adopt several AI files

77.2%of markdown-alternate sites also serve llms.txt

How markdown alternates and llms.txt links overlap with llms.txt adoption.

llms.txt adoption among sites using AI link tags
SitesAlso serve llms.txt
Link to a markdown alternate
951
734
Link to llms.txt in the page head
2,889
2,739
All top-1M homepages
772,412
110,188

% serving llms.txt

GroupSitesAlso serve llms.txt%
Link to a markdown alternate95173477.2
Link to llms.txt in the page head2,8892,73994.8
All top-1M homepages772,412110,18814.3

Source: aisearchoptimization.net analysis of 772,412 top-1M homepages · HTTP Archive, October 2026 crawl

A small group of AI-minded site owners is layering several signals. Most of the web has none of them.

Developer-focused platforms lead

2.2%of TYPO3 homepages offer a markdown alternate

Platforms named for adoption only.

Markdown alternates by detected platform
PlatformHomepagesWith markdown alternate%
TYPO3 CMS2,522552.18
Astro2,940451.53
Contentful1,94280.41
Sanity93180.86
Joomla3,557220.62
Drupal11,424560.49
Next.js36,8991480.4
WordPress107,0571610.15
Shopify46,231300.06
Other or none detected496,6643520.07

Source: aisearchoptimization.net analysis of 772,412 top-1M homepages · HTTP Archive, October 2026 crawl

No platform reaches 2.5%. Where adoption is higher, it likely comes from an extension or theme feature on that platform rather than a default.

What this means

For most business sites, a markdown alternate is optional. AI tools still read normal HTML pages, so clean structure and clear headings matter far more. If you run documentation, a knowledge base or a product catalog that AI assistants quote often, a markdown version can make those quotes cleaner.

If you add one, keep it in sync with the HTML page so AI tools do not read outdated facts. Want to know whether AI assistants can find and quote your business before you build extra files? Our free AI visibility audit answers that, and so should any AEO agency you talk to.

Limits

We read link tags on homepages only; sites may offer markdown versions of inner pages without linking them from the homepage.

Some sites serve markdown when a client asks for it through HTTP content negotiation, with no link tag. Those are not counted.

Link tags added only by JavaScript after load may be missed.

No growth trend: link tags sit in a large metric we chose not to rescan across past crawls.

Embed this research

Add the charts from this study to your page. The embed links back here.

<iframe src="https://aisearchoptimization.net/embed/data-studies/markdown-page-versions-for-ai-agents" title="0.12% of Top Sites Offer Markdown Page Versions for AI" width="100%" height="720" style="border:0" loading="lazy"></iframe>

Cite this study

APA

Grenyo, K. (2026). 0.12% of Top Sites Offer Markdown Page Versions for AI. AI Search Optimization. https://aisearchoptimization.net/resources/data-studies/markdown-page-versions-for-ai-agents

Plain link

0.12% of Top Sites Offer Markdown Page Versions for AI - AI Search Optimization - https://aisearchoptimization.net/resources/data-studies/markdown-page-versions-for-ai-agents

Download the data (CSV)

Frequently asked questions

What is a markdown alternate link?

A link tag such as <link rel="alternate" type="text/markdown" href="page.md"> that points AI tools to a markdown copy of the page.

How many websites offer markdown versions for AI?

951 of 772,412 top-1M homepages (0.12%) link to one from the homepage.

Should I link llms.txt from my pages?

It does no harm. 2,889 homepages do it, and 94.8% of them actually serve the file, so check yours resolves.

Do AI search engines need markdown pages?

No. They read HTML. A markdown copy can help some agents get cleaner text, but it is an extra, not a requirement.

See where AI search leaves you out

Run the free AI visibility audit. If you want the fixes done for you, the plan is a flat $299/month.