It takes two to tango – You cannot build an accurate MMM model without the client’s inputs.
Just early this week we onboarded a new client. The client asked “Ridhima, so how much support does your team need from my marketing team?”
My answer : “Well.. a little bit 😊”
MMM is often perceived as a purely statistical exercise : collect data, run models and generate insights.
In reality, MMM sits at the intersection of data, business context and decision history. If you remove any one of these, accuracy suffers.
Here’s why client input is so crucial.
📌 Data never tells the full story
At Aryma Labs, we have an internal joke “The MMM is not clairvoyant”. It can’t magically know things that are not there in the data. At best, a well built MMM can only surface all the hidden insights from the data. But it can’t conjure up new insights outside of the data.
The MMM model is an explanatory model not a prediction model. If there are all the data ‘that explains’ the sales for a particular time, a well built MMM will unravel it.
But if there is huge jump or drop in sales and there is no data about why that happened (in the form of increase/decrease of spends or mention of some external shock) then MMM won’t capture it.
Spikes and drops in performance often have reasons that never exist in datasets. Examples include:
▪️Strategy change
▪️Distribution / Supply issues
▪️Creative changes
▪️Competitor activity
Without this context, models are forced to “explain” marketing reality incorrectly.
📌 Media variables are rarely clean
Campaign structures change. Objectives change. Platforms optimize differently over time. Only the client knows when strategy changed versus when performance genuinely changed.
If this is not incorporated, MMM attributes effects to the wrong channels.
📌 Business reality defines model constraints
Client inputs help prevent mathematically correct but commercially impossible outcomes.
MMM is a collaboration, not outsourcing or just call a open source library over a API.
The best MMM projects feel less like vendor delivery and more like joint problem solving.
When marketing and modeling teams work together, the model reflects reality.
We often get asked – “Are you worried about open source MMM libraries and players who just put a beautiful UI on top of them?”
Actually we are not 😎. Our confidence comes from the fact that – MMM is not just model building. It is about understanding the business. Talking with client stakeholders and problem solving together.
We have done this many times for clients across 5 continents over the past 7 years. This institutional knowledge is our MOAT.
Models don’t fail only because of statistics.
They also fail because context was missing.
From our side we always ensure our models have all the statistical capabilities to surface the truth from data. But it also takes a client’s input to complete the model.
It always takes two to tango 💃🕺
