Short answer
Incrementality testing measures the causal effect of marketing by comparing a treated group (exposed to the ad) with a controlled holdout. Because it is a real experiment, it produces the cleanest possible measurement: and it calibrates MMM.
Incrementality is the sales you would not have got without the ad. Everything else: brand fans, existing demand, seasonality: is base. The whole point of measurement is to isolate incrementality.
Geo-experiments split regions into test and control, run the campaign in test, and compare outcomes. Ghost-bid, PSA and audience holdouts do the same on digital platforms. Well-designed tests give unambiguous causal evidence.
Tests are expensive, slow and channel-specific. MMM is fast, cross-channel and cheap once built. The winning play: use experiments to calibrate MMM priors and validate its coefficients. Best of both worlds.
Both. Test to establish ground truth on key channels; model to see the whole picture and plan budget.
One or two per major channel per year is a healthy cadence for calibrating an MMM.
Bayesian MMM, calibrated with experiments, refreshed monthly, delivered as a decision system.
Explore our solutions