Kyle Roof on How Google Detects AI Content & Wipes Out Entire Websites

In this episode, Kyle Roof discusses why Google’s core page understanding hasn’t changed much, arguing that ranking can still be modeled with term frequency in key on-page zones. He says word proximity and section order matter less than writing clearly on-topic, but emphasizes that above-the-fold content and limited initial crawl depth (often cited around 50KB of text) make early page clarity critical. Roof shares multi-run case studies testing LLM-generated articles, finding they often miss required contextual terms, produce dense college-level readability, and sometimes perform worse when asked to be “SEO optimized,” mainly improving headings while stripping key terms. He theorizes Google has dialed up weighting on contextual/LSI terms, contributing to sitewide  ...