Splitting Media Variables by Year / CPM Changes Over Time

[boxx0r]

Hello,

In light of only being able to model spends (as opposed to modeling media exposure data like impressions), the recommendation from this group has been to split the media variables when spend/impression patterns are different (see #466 (comment)). I am already doing this by splitting media channels with different CPMs into separate variables.

However, another issue with modeling spends is the changing cost of media exposure over time, e.g. CPM (cost per impression), CPP (cost per point).

1. Would it be advisable to split media variables into years, e.g. TV Year 1, TV Year 2, etc. in order to control for the changes in media cost? That is, $1M of TV spend in Year 1 and 2 may yield the same contribution volume, but if Year 2 GRPs are more expensive, then the Year 2 TV actually performed more efficiently, because it must have had fewer GRPs behind it.

2. Is there a plan to be able to model non-spend for media variables, i.e. impressions? This would help solve a lot of these workarounds.

Thank you.