Study Reveals Key Factors in AI Answer Engine Citation

A recent study published on arXiv (2605.25517) explores the dynamics of competitive Generative Engine Optimization (GEO), examining the factors that influence which source is cited first when two options compete in AI answer engines. The researchers developed a controlled retrieval-augmented generation (RAG) testbed, incorporating two candidate sources into the model context and tracking which source received the initial citation marker. They conducted 252,000 trials across six LLMs, assessing 18 content factors while employing brand anonymization and counterbalanced source order to differentiate content effects from positional bias. Findings from mixed-effects models indicated that both topical relevance and list position significantly impact which source is cited first, emphasizing the importance of GEO for content creators.