Why the MMM – Experimentation – MTA Triangle is kinda wrong
Around a decade ago, I was in between jobs and took on a NLP consulting gig. The task was to develop a Topic Modeling system. It is here that I first learnt about Dirichlet distribution.
It is a beautiful concept.
During a recent client pitch meeting, we walked through our familiar MMM-Experimentation – Causality triangulation slide and something clicked.
Our triangulation diagram reminded me of the Dirichlet allocation.
And after thinking about it for days, I realized something uncomfortable.
Almost every vendor (including us) draws this as a perfect equilateral triangle. And this might be conceptually wrong !!
📌 What the equilateral triangle quietly implies:
When we draw an equilateral triangle, we implicitly suggest:
▪️Equal reliability
▪️Equal credibility
▪️Equal ability to converge to “ground truth reality” (ROI, mROI, incrementality etc)
While this as a marketing speak sounds good, mathematically it is a bit misleading.
The Dirichlet simplex (triangle) if taken as a mental model, is not just a geometric representation. It is a probability simplex – meaning that each point inside is a ‘belief allocation’.
How much do I trust MMM, MTA or Experimentation (causal ones too) to represent ground truth reality?
📌 The colour significance
The colours matter.
Red near a corner means the system strongly prefers that method as the closest proxy to truth.
Blue areas indicate low credibility for that allocation.
Red near MMM means MMM is the dominant lens on truth.
Red near Causal Experiments means causal tests are honing in on the truth.
📌 The catch
You cannot be at all three corners at once nor permanently in the dead center.
Because the center implies:
Every combination of MMM, MTA, and Experimentation is equally likely to be correct !!
This is equivalent to saying “All methods are always equally credible.”
No serious marketing measurement vendor actually believes this.
And empirically too it is never true.
📌 So what is the key takeaway:
I am going to instruct my website team to first make changes to our triangulation and replace it with a Dirichlet simplex representation.
And for each use case or objective, we will show you which corner of the triangle glows red.
For example:
✅ If your goal is holistic marketing measurement, then the corner of MMM should glow red.
✅ If you want to test the efficacy of short term campaigns, Pmax or ASC etc. then it should be Experimentations / causal experiments.
Most importantly, not all the methods will hone in on a single ROI and the expectation to do that also is wrong.
This is why I have also been saying that MMM should not be calibrated with experimentation results. Yes talking about calibration not validation. Check the research paper in comments.
Also link to the ‘State of Marketing Measurement 2026’ fireside chat where we discuss ‘Is there a order to do ‘MMM-Experimentation-MTA’? in comments.
