The Edward Show Schema Is Useless for LLMs: The SEO Lie That’s Finally Exposed
12 snips
Sep 18, 2025 Discover why the hype around schema for AI and LLMs is unfounded. Viral experiments reveal that structured data is ignored during tokenization, making it ineffective for boosting visibility in AI search. Instead, learn actionable strategies like optimizing visible text and creating rich content that appeals to AI tools. Stop wasting resources on snake oil SEO tactics and embrace methods that genuinely drive engagement and sales in the digital landscape.
AI Snips
Chapters
Transcript
Episode notes
Schema Gets Destroyed By Tokenization
- Large LLMs tokenize text and break schema into ordinary tokens, so schema structure gets lost during training.
- This means schema rarely provides a distinct signal to the core LLM training process.
Niche Schema Is Costly And Often Unnecessary
- Niche or granular schema implementation is time-consuming and costly compared with standard schema.
- Edward argues that because schema doesn't help LLM visibility, that cost often isn't justified.
Real Test: Visible Text Beats Schema-Only Pages
- Julio C ran experiments with product pages where one page showed visible text and the other only had structured data.
- Both ChatGPT and Gemini returned data only from pages with visible text, not from pages with just schema.