Google’s John Mueller Rejects Markdown-Only Pages for AI Bots

Google Search Advocate John Mueller has criticised the idea of serving Markdown-only versions of web pages to AI crawlers, calling it “a stupid idea” on Bluesky. His comments respond to experiments where developers detect AI bots and deliver raw Markdown instead of full HTML to reduce token usage. What’s the idea behind it? Some developers believe Markdown is easier for large language models to parse than HTML and claim it can reduce token usage by up to 95%. The theory is that this could help AI systems ingest more content for retrieval-augmented generation (RAG). Why Mueller disagrees: Mueller questioned whether AI crawlers even treat Markdown as a proper web document or reliably follow links inside it. More importantly, there’s no proof that AI platforms reward lighter formats with better visibility, citations, or rankings. He has consistently advised publishers to avoid bot-only content formats unless platforms officially support them. What the data shows: Studies, including large-scale domain analysis, have found no relationship between bot-specific formats (like Markdown or llms.txt) and increased citations in AI answers. So far, no major AI platform has published guidelines asking for Markdown versions of pages. Best practice going forward Until official specs exist, the recommendation is clear: keep HTML clean, reduce unnecessary JavaScript, and use structured data where supported. This approach works across search engines and AI systems alike. For learners and professionals at a digital marketing institution in Calicut, or anyone enrolled in a digital marketing academy in Calicut offering a digital marketing course in Calicut, this debate reinforces a key lesson: sustainable SEO and AI visibility come from solid technical foundations—not experimental shortcuts.