New GEO survey reveals SEO tactics don't work for AI search engines

A new critical survey of Generative Engine Optimization (GEO) research from 2023 to 2026 casts serious doubt on the effectiveness of current SEO strategies for AI-powered search engines. Authored by Olivier Martinez and published on arXiv, the paper reviews 45 studies and argues that GEO is not a single ranking task but a complex, stochastic pipeline spanning search activation, crawling, retrieval, reranking, citation, and user behavior. The widely cited gains from the foundational GEO paper are shown to be valid only within a narrow experimental setting—where content was already present in a fixed context—and do not demonstrate organic discoverability or durable traffic effects.

Key findings include that topical relevance and context position are the most reproducible levers, while generic heuristics transfer poorly across domains. Competition can erode individual gains, and citation-oriented rewrites can actually impair retrieval performance. Commercial audits further reveal low source overlap, substantial run-to-run variability, and persistent fidelity gaps. The survey proposes a multistage formal model, a visibility vector separating discoverability, citation, absorption, and economic outcomes, and a reproducible protocol based on repeated measurements, paraphrases, controls, human validation, and multi-actor interference. The conclusion is sobering: no reviewed technique shows a stable, longitudinal, cross-platform causal effect on organic discoverability or downstream behavior.