Experiment-calibrated MMM uses causal lift evidence, such as a geo experiment, as prior information for the model. This can ground channel estimates that aggregate observational data alone cannot identify well.
In the demo, a lift result enters the model and its illustrative estimate range narrows. This is a possible visual outcome, not a guarantee of calibration.
A test with different audiences, periods, or outcome definitions may impose the wrong constraint. Check experiment quality and transferability first.
When to use
Use it before major MMM budget decisions to ground estimates in experiment evidence.