Context
A 2026 preprint surveys 45 studies on Generative Engine Optimization and critically synthesizes evidence about interventions intended to influence retrieval, selection, citation, prominence and visibility in generative answer systems.
What changed
The paper consolidates a fragmented GEO research literature into a critical survey and measurement-oriented framework, while identifying methodological limitations in claims about durable discoverability effects.
Why it matters
For AI visibility measurement, the survey supports treating retrieval, downstream selection, citation/prominence and outcome measurement as distinct stages rather than collapsing them into one visibility metric.
Visibility OS interpretation
Visibility OS interprets the survey as support for stage-specific experimentation: an observed citation or inclusion effect should not automatically be generalized into a universal ranking or durable cross-platform discoverability claim.
What remains uncertain
Which GEO interventions replicate longitudinally across production systems? How stable are effects across model, retrieval stack, query class and time? Which intermediate visibility measures reliably predict user or business outcomes?