MMM puts media, promotions, seasonality, and total outcomes into one portfolio view. It is not an attribution replacement or an automatic verdict.
Key moments
- 0:00 Each ad platform claims the same order, but the cash register shows one sale.
- 0:23 Geo tests show what causes what; modeling adds inferences, including for channels that can’t be tested.
- 0:44 What goes into a model: spend and impressions by channel, promotions, and seasonality.
- 1:23 Weaknesses: flat spend, channels moving in lockstep, and a wrong decay assumption.
- 2:04 Geo tests act as pins on the map that narrow the uncertainty for untested channels.
- 3:05 Use the model to find where each channel starts to get worse and set spend up to that point.
What this video covers
An MMM is not a verdict or an attribution replacement. Weak variation, correlated channels, and carryover assumptions can limit the answer. A useful model makes those limits visible and identifies where a calibration test would be the better next move.
Read the related guide: Marketing Mix Modeling for DTC Brands