The Only Thing Preventing Full Automation of MMM is Adstock – And That’s a Good Thing.

In a recent meeting with a prospective client, we were asked – “So why is it that MMM can’t be fully automated ?”

Fully automated MMM is a beautiful dream and every vendor and client has aspires for it.

But there is one stubborn obstacle standing in the way – Adstock

Adstock is not just a mere statistical transformation. It is a belief about how advertising behaves over time and it opens up many other pertinent questions like:

▪️How long does ads linger?

▪️Does performance media decay in days or weeks?

▪️Is the carryover effect geometric or weibull?

▪️Does creative fatigue alter decay?

The moment we choose an adstock function, we are making structural assumptions about reality. And we all know reality is messy.

If MMM were purely mechanical, we could standardize adstock across every brand and be done with it.

But sadly we can’t because:

▪️Categories differ

▪️Purchase cycles differ

▪️Creative wear-out differs

▪️Media quality differs

▪️Competitive pressure differs (not many think about this, but competition does influence the shape of your adstock)

At Aryma Labs, we have seen this repeatedly:
Two brands in the same industry almost similar spend levels but totally completely different carryover and saturation structures.

Why?

Because brand equity, creative quality, and consumer psychology are not constants.

If you want to automate something, the underlying phenomenon must be as static as possible.

Fortunately or Unfortunately that is not the thing with Marketing.

📌 Modeling without Adstock

One can build a MMM without adstock (I guess then it can just be called a Linear regression then). But we won’t have answers to questions like :

– Is this channel driving instant response or building future demand?
– What is my current operating level? Can I keep spending more ?
– Where do I hit the saturation point and my next dollar spend stop making sense?

One of the things that makes MMM useful/ forward looking and not just be curtailed to being a ‘tell me what happned in the past’ is – Adstock

At Aryma Labs we make sure our models are handcrafted and built with statistical rigor. Yet we just take 2 weeks of time to deliver the final outputs post all data collection !!

Sometimes if you want to know the causal truth behind your marketing spends and KPI, a bit of a wait is ok I suppose 🙂

P.S : If you are curious about the image, check the link in comments.
Other interesting articles on adstock in comments too.