Blogs

MMM First? Or Experimentation First? In our recently conducted Fireside chat (ICYMI – Link in comments), I had asked the panelists: “Is there an ideal order of Measurement Technique to pursue? What should be carried first – MMM? Experimentation? Causal Experiments?” The panelists gave really brilliant answers (watch the session to know what they were 🙂). However in this post, I want to share my two cents on this topic. 📌Start with the eagle’s eye view I firmly believe that every brand big or small should start with Marketing Mix Modeling (MMM). Why? Because MMM gives you the big picture first: ▪️Which channels worked, by how much and in what

Why “Peeking” in Experiments Can Break Your MMM In our recent Marketing Mix – Unmixed podcast with Premjeet Sodhi, Premjeet mentioned a very interesting point. “The beauty of MMM is that it is non intrusive. You can let the world happen as it is, You can let life happen and then you can pick up the data about what measures it and gain insight into it”. An excellent point. While MMM is non intrusive, a lot of things can intrude in it and perhaps make MMM infeasible !! One such intrusion is Peeking. 📌 What is Peeking? In Experimentation, Peeking is the act of taking a a peek at your

Why timing is crucial for validating MMM through Experiments. MMM (Marketing Mix Models) by virtue of being a larger and more complete representation of marketing reality, can be used to contextualize and sanity check experimentation. However, this is only valid in a limited time window immediately after the model build. At Aryma Labs, we call this window the umbra – typically the next 3 months post MMM. But a experiment validation can still go wrong. How? Imagine you model a MMM for past 2 years (2024-25), Lets suppose your google ad spends have been most effective in the months Oct -Dec every year. However, your model right now has the

Why SCM analysts deny existence of real control When I was in 5th grade, my friends and I desperately wanted to ride a roller coaster. All of us were a little short when measured against the signage. We wanted to tear down the height measuring sign. This sign made us look bad. Years later it dawned to us that in fact the sign was there as a safety. One can’t wish a away reality even if it makes us look bad. SCM practitioners and Bayesians have one thing in common. They deny the existence of reality. SCM analysts deny the existence of a ‘real control market’ while Bayesian would deny

We need to rethink the concept of Adstock for Gen AI Ads OpenAI AI and Perplexity have been experimenting placing of ads in their platforms. I think it is only a matter of few months when the ads will be fully rolled out. With ad placements comes the question of efficacy of the ads and this is where Marketing measurement techniques like MMM comes into the picture. However MMM can’t be directly used to model the effectiveness of ads in Gen AI. Why? Gen AI Ads break one of MMM’s Oldest Assumptions: Adstock Almost every adstock function we use today is built on a assumption: ▪️ Advertising creates awareness from

When Mathematics Makes Us Forget People: A Blind Spot in Synthetic Control Method (SCM) Happy New Year Folks, All of you know I am a stickler for statistical rigor. But over my decade of practice in Marketing measurement, one thing I have realized is – “Don’t forget that all models are just an abstraction of reality”. Mathematics should not make you forget what you are modeling. One thing I increasingly notice with Synthetic Control Method (SCM) practitioners is how quickly the conversation drifts from markets to mathematics. Weights, Optimization, Loss functions, Pre period MSPE. Everything becomes beautifully mathematical. And somewhere along the way, something important gets lost. A control market

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.

The Future is still Narrow AI I can see a deluge of Linkedin Posts about GPT 5.2. Of course it is a great progress, don’t get me wrong. But IMHO, the future will still be Narrow AI and most industries in any domain will benefit from Narrow AI rather than General AI. If you are wondering what is Narrow AI, see my post from a few days ago – https://lnkd.in/guGcjU9K Frontier model progress often cascades downstream. Distilled models from likes of GPT-5.2 or Gemini-3 will drive more accurate and practical real world results. However these frontier models by themselves will lack few things. 📌 General Intelligence ≠ Domain Mastery GPT-5.2

MCP (Model Context Protocol) solves the connectivity layer. But MMM adoption problems were never purely about connectivity. However, many marketing measurement vendors are now exposing their MMM models through MCP servers, so tools like Claude or ChatGPT can directly interact with MMM platforms. But two big question still remains: ▪️Do marketers and analysts actually want to use AI chat interfaces for MMM workflows? ▪️And whether conversational interfaces are actually the best medium for: – Contribution analysis – Scenario planning – Budget optimization – Saturation Curve analysis – ROAS / MROAS analysis – Stakeholder storytelling Would love to understand where the industry is leaning.

Solving the hard problems in MMM through AI Everyone is suddenly racing to add AI into MMM. But I increasingly feel many vendors are solving the easy problems. Creating MCP servers to expose MMM outputs to Claude is not some massive breakthrough. You can vibe code it nowadays. The real problems in MMM were never about “connecting to AI”. The real problems are: • Why do organizations forget MMM learnings every year? • Why do MMM outputs live in silos? • Why do budget optimizers recommend mathematically optimal but business impossible plans? These are the problems we have been obsessed with solving at Aryma Labs. 📌 MMM Synapse One of