The Citation dimension systematically monitors which domains, pages, and specific URLs AI search engines retrieve and cite when generating grounded responses across major platforms like Google Gemini, OpenAI ChatGPT, and Anthropic Claude. Using dedicated citation parsers, the system dissects grounded API responses down to chunk-level supports, text annotations, and character offsets. This structure allows Auxy to classify references as selected versus unselected citations, differentiating between sources actively cited in the generated answer and those merely browsed and discarded during the search process.

In addition to tracking destination URLs and resolving search redirect links, citation mining captures intermediate fanout queries to reveal how models deconstruct user intent behind the scenes. Stored ownership rules and competitor domain mappings then aggregate citation share across target brands and topic categories. These diagnostic trails feed directly into the Content Optimizer, which executes iterative optimization loops to rewrite snippets or page content to increase extraction probability. Concurrently, the Veracity engine assesses source fidelity by scoring how many source facts survived, were lost, or were distorted in the model's generated text.