Incrementality Testing at Scale: Geo-Lift Experiments and Synthetic Control Methods

Attribution models answer 'who gets credit?' — incrementality testing answers 'would the sale have happened anyway?' This paper presents a comprehensive incrementality testing framework combining geo-lift randomized experiments with synthetic control methods (Difference-in-Differences, Synthetic Control Method, Bayesian Structural Time Series). The framework provides unbiased causal lift estimates with formal confidence intervals, enabling budget reallocation decisions grounded in experimental evidence. Provides ground-truth validation for Robinson (2026a) through (2026b).