Why MMMs are implicitly causal but yet you still can’t prefix C before MMM
MMM is implicitly causal if you build it the right way.
A well-built MMM explicitly controls for all variables that affect the KPI.
When endogeneity shows up in MMM, it’s usually due to omitted variable bias (OVB). You get into this mess while trying to solve for another problem- Multicollinearity. One tends to drop the variables (not an advisable move).
Anyhow, if there is no OVB, there is no endogeneity. Of course there are solutions for endogeneity and multicollinearity like 2SLS method. But for the sake of brevity I won’t go into it in detail here.
So when I say MMM is implicitly causal, I mean an MMM that is specified correctly and free from endogeneity.
📌 But why is MMM not explicitly causal?
Well MMM ticks almost all the boxes for causality except for Counterfactual.
MMM does not answer the counterfactual by itself.
We still don’t know what would have happened in the absence of an intervention (say the non implementation of spend increase on Meta or Google as suggested by your MMM).
To make MMM explicitly causal, you still need quasi causal experiments like DiD.
That’s why you can’t prefix a “C” before MMM.
📌 MMM can be called Causal only post-hoc
MMM can be called causal only post hoc i.e. by administering a Difference-in-Differences (DiD). DiD specifically operates with a real control market unlike SCM and this helps in establishing a true counterfactual. See the details of our paper in comments.
If you implement a SCM post MMM, your MMM would still only be implicitly causal because SCM by nature operates with a ‘synthetic’ control. As I have stated in the past in many of my posts (Link to my posts in comments.), be wary of anything ‘Synthetic’. Not to mention Synthetic Control Method is fundamentally flawed and one can never get accurate casual effects.
📌 Bottom line:
MMM is implicitly causal because its estimand, structure, assumptions, and outputs are causal.
But without an explicit counterfactual, it remains implicitly, not explicitly, causal.
