Guide
Can you add an llms.txt to Substack? (2026 guide)
Updated · July 10, 2026 — Joffrey
What is GEO?→Substack is a closed platform: you can neither drop in an llms.txt file nor add valid schema.org markup. Here's exactly what's possible on Substack — and the realistic path to making your content citable by AI answer engines.
Why Substack makes AI citation hard
AI answer engines (ChatGPT, Perplexity, Gemini, Claude) cite only a handful of sources per question, and they favor pages they can read cleanly. An llms.txt hands them a clear, structured version of your site; Q&A schema.org markup makes your answers quotable word for word. Without it, even ranking well on Google, your Substack site is unlikely to appear in an AI answer — not for lack of quality, but because nothing is machine-readable.
What you can (and can't) do on Substack
- Substack allows no custom code: no editable
<head>, no root file, norobots.txtof your own, no JSON-LD. The only AI setting is the “Block AI training” toggle. - In other words, nothing can be added on Substack to make it citable — even a custom domain unlocks no file or markup access.
- The realistic path: move your key content to a site you control (blog, static site, WordPress…) and generate your llms.txt + Q&A markup there with Citeable. That's where AI engines will cite you.
Verify it's live
As long as your content lives only on Substack, AI engines have nothing structured to read. Publish your key pages on a site you control, generate the llms.txt + Q&A markup there, then verify by opening your-site.com/llms.txt and via Google's Rich Results Test.