Comment by stephen_greet

I don't think you need to get the attribution exactly correct. Rather, you need to be able to defend how you arrived at a metric.

You can walk through the assumptions you made then build up the components of those assumptions to arrive at an estimated impact.