Context
Kato, Honma, and Kato propose Generative Marketing Mix Modeling (GMMM), combining repeated generated answers, query counts, generative-system shares, and notice probabilities to construct GEO exposure inputs for marketing mix modeling.
What changed
GEO measurement is extended beyond raw mention, citation, or referral counts toward expected noticed occurrences that can be modeled with carryover, saturation, treatment sequences, and uncertainty.
Why it matters
AI visibility programs increasingly need to connect answer-level exposure to business outcomes. The framework clarifies the additional measurements and identification assumptions required before treating visibility as a causal business channel.
Visibility OS interpretation
Visibility OS treats GMMM as a measurement framework, not evidence that GEO universally drives incremental revenue. The empirical answer collection provides occurrence estimates; simulated outcomes evaluate the causal estimation procedure.
What remains uncertain
How well does GMMM identify incremental effects in live field experiments? How reliably can organizations estimate relevant-question volumes, platform shares, and notice probabilities? How sensitive are estimates to rapid model and retrieval changes?