Move 37 Won’t Come From the Core of MMM but the Periphery

In 2016, AlphaGo vs Lee Sedol match, AlphaGo played Move 37.

A move so unconventional that even the best human players thought it was a mistake.

But it wasn’t. It was a move which eventually proved decisive for the victory against Sedol.

For the fist time, it was a glimpse of intelligence beyond human intuition.

📌 So why we haven’t seen Move 37 in Marketing Measurement yet?

Well I would say we are looking at the wrong place firstly. The Move 37 will not happen at the core of MMM building.

This nucleus is about:

– Variable selection
– Functional form decisions
– Adstock specification
– Saturation curves
– Bias diagnostics
– Causal validity checks

All of these can’t be done reliably by any AI agents still. The moment we can we would have solved causality and hence achieved AGI

📌 Move 37 will happen in Peripheral Layer

My strongest belief is that Move 37 will happen in peripheral layer and Aryma Labs will be the first to crack it 😎.

AI today is strongest in:

Scenario simulation
Budget allocation interfaces
Insight surfacing
Pattern recall
Rapid querying of past learnings

This is where something interesting happens
Speed + breadth + recall start exceeding human limits

And that’s the exact setup where non obvious decisions emerge.

Guess what Aryma Labs already pioneered some ground breaking research and product development since 2024 on exactly the above.

📌 What a Peripheral Layer Move 37 looks like?

It will definitely not be a better model (that is still human domain). But a peripheral layer move 37 will result in better decision not envisioned before.

Something like:

– Recommending a budget optimization combination that looks wrong but could prove beneficial in future. Something like cut budget on the hero channel by 20% because it is already saturated or will definitely
saturate.

– Highlighting a strategy pattern buried across multiple past MMM cycles

– Surfacing a budget and spend pattern from Covid times now because ‘now’ has some similarities to the past

And eventually the human choosing to act on the above. Because the biggest bottleneck still would be an Agent – A human. We have agency in truest sense but that is also why refrain from taking some decisions.

📌 Exciting Times Ahead

We will be launching Aryma AI (our dedicated AI vertical) soon. This vertical focused on Marketing Measurement will boast of cutting edge AI research and innovative products never seen before.

“Can AI build the MMM model?” is a wrong question.

The right question is “Can AI help us see decisions we would never thought of and consider?”

Because Move 37 was not about building a better board. It was about seeing a move no one else could see.

MMM doesn’t need fully agentic models to have Move 37 moment.
It just needs humans willing to act on AI surfaced insights.

We are working towards making this AI surfaced insights reliable, accurate and actionable. More updates soon 😎.