Marketing Mix Modeling • 5 min reading time

4 elemental questions to ask regarding Marketing Mix Modeling

Brian Cusick - published on July 1, 2022

New choices for Marketing Mix Modeling implementation

For the last few decades the process of implementing marketing mix models (MMM) has been pretty much the same, write an RFP, meet with vendors, and select a specialized firm. As technology has changed nearly everything in life, it has also brought us new options for MMM.

If you are responsible for implementing a Marketing Mix Model you should ask yourself these questions before deciding on the best approach:

  • Who knows the business better, my team or 3rd party consultants?

  • Who will be using the output and how much do they need to know about the work?

  • How important is flexibility? Will the questions change as people see results or is the work pretty much cut and dry?

  • Do we have a history of MMM results to guide us or do we need to test and explore in order to find the best path?

marketing mix modeling process

Considering these factors will help you make the decision between outsourcing vs. in-sourcing. With today’s software packages in-sourcing marketing mix is a feasible option for any data science or general “analytics” team.

Outsourcing vs. In-sourcing MMM

Careful consideration: ScanmarQED Marketing Mix Modeling experts can guide you through the process (early startup as well as mature MMM)

Outsourcing your Marketing Mix Modeling

Nearly every high-quality “mix shop” will follow these steps:

  1. Talk through your business questions and help you flush out what you want to learn, how MMM results might impact your planning process, and identify the required data

  2. Help you actually find your data, then review the data and make sure it measures what you think it measures- presenting it back to you in a “data review”

  3. Build the statistical model(s) and interpret the results on your behalf

  4. Run some sort of forecast, simulation, or optimization

You are likely to think that you “outsourced it” but as the project gets going client’s often realize that it still requires a lot of their time- check-in meetings, asking vendors for data, checking data, explain how your business works, explain again how your business works, and finally, explain how your business works. Frankly, if you aren’t asked a lot of questions then you should be very concerned about the quality of work being done. MMM without business context is more dangerous that it is helpful.

In-sourcing MMM to your team

Your steps here will be surprisingly similar to outsourcing:

  1. Talk through your business questions with your stakeholders (same)

  2. Get your raw data, (again, same as outsourcing, but for your own use, less worry about confidentiality, less need to explain)

  3. Build your statistical model- this used to be hard, but with the right software you don’t need to be a statistician, the software simplifies model building because it combines the complex mathematics with the tricks of the trade that only seasoned MMM specialists tend to know

  4. Interpret the results- this does take some getting used to, but the leading software packages also help a lot with this… all the business metrics should be calculated and visualized for you, instantly. If they aren’t then you haven’t chosen a solid software vendor. You might even get a day or two of consulting help here, but you’re still in control and leaning.

  5. Run forecasts, simulations, and optimizations: you guessed it, a good modern MMM package will take your model(s) and walk you through this in a matter of hours


What should you choose for Marketing Mix Modeling implementation?

So let’s get back to our questions from the beginning:

  • You and your team probably know your business better than anyone that you could hire, rather than explain things in triplicate, use your knowledge to make the best MMM decisions

  • The one thing you know with certainty is that someone on the leadership team is going to throw a curve ball at the end of the project, so you definitely need the kind of flexibility that only comes from an internal capability.

  • If your business leaders make data driven decisions then Marketing Mix Modeling results will impact budgets, which impact people’s jobs, and hence are inherently political and sensitive, they aren’t the sort of thing that should be hidden inside black box 3rd party consulting firms.

  • …however, if you can’t commit any people resources at all, if your team just wants an answer and doesn’t really require an explanation, then you should probably stick to an MMM specialist shop. Modern MMM tools allow good (generalized) analysts to learn and use Marketing Mix, but they still require effort and a commitment to learn the process.

Picture of Brian Cusick

Brian Cusick

Brian Cusick leads the North American business unit for ScanmarQED, ensuring that both new and existing clients get the most value from our services and products. His expertise is in quantitative marketing research (previously worked at AC Nielsen, PepsiCo, Kraft Foods), with a focus on data driven business planning and marketing mix based optimization.