Generative Engine Optimization (GEO) is the practice of structuring content so AI engines retrieve it and include it in the context window used to generate their answer. Where Answer Engine Optimization (AEO) targets the final citation surface, GEO targets the retrieval pipeline one layer earlier — if your content does not survive the retrieval step, no AEO investment can save it.
The term was coined in the Aggarwal et al. paper "GEO: Generative Engine Optimization" (arXiv 2311.09735, presented at KDD 2024). The researchers showed that nine specific content modifications could lift a source's inclusion rate by up to 40% in controlled experiments.
The strongest GEO signals are statistics with specific numbers, quotations attributed to named experts, citations to authoritative external sources (DOI, .gov, .edu, recent news), easy-to-quote summary blocks, topic clusters with semantic depth, multi-format content (table plus paragraph plus bullet), and embeddings-friendly chunking via short paragraphs with clear topic sentences.
In practice GEO is run together with AEO. Many signals overlap — adding statistics, citations, and clean chunking lifts both inclusion (GEO) and citation likelihood (AEO). The disciplines differ in where they intervene, not in what they reward.