> For example: if sales fall off a cliff, it's reasonable to interpret that as "perhaps I over did it." and so on.

But (a) that's not the outcome, so it doesn't help here, and (b) with the pandemic, even if sales went to zero, you'd still need some way to know the effect of that.