How AI systems choose their sources.
Google's AI Overviews are grounded in Google's own index: the system generates an answer, then attaches supporting pages that rank well for related queries. A page that cannot rank in the classic sense rarely appears as a citation. ChatGPT and Copilot lean on Bing's index when they browse, and Perplexity runs its own retrieval layer with similar authority weighting.
Underneath retrieval sits training data. Large language models learn entity associations from web crawls where high-authority publishers are heavily overrepresented. A brand that appears in Forbes, The Guardian, and Business Insider gets encoded as an entity connected to its topic. A brand that only appears on its own website does not.