MMM baseline detection: are there enough periods of advertising variation?

When a model tries to measure each channel’s effect, it needs to see what happens when there is variation in the channel or that channel isn’t running. These “off” periods help define the baseline – the level of sales that would have happened without any advertising.

This matters for Out of Home advertising because without clear gaps, the model has no contrast to work with. It can’t match sales changes to OOH, because OOH is always present. The model will also rely on tiny timing differences between OOH and other channels to decide on the ‘credit’ to give each channel.