We spent 18 months studying how Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot decide what to cite. We tested thousands of queries. We mapped the patterns. We built a repeatable framework for getting businesses into those answers consistently.