World of Marketing Mix Modeling experts

May 14, 2021 | Chris Kervinen, Veli-Pekka Julkunen

World of Marketing Mix Modeling experts

In our series of ”World of Marketing Mix Modeling”

we’re entering the Matrix and diving into sea of data to bring you the real life superstars of marketing mix modeling.


Veli-Pekka Julkunen


- Master’s degree Economics/ Econometrics from the University of Helsinki

- +12yrs. of experience of utilizing Data Science & AI (Artificial Intelligence) in solving complex business decisions

- Co-Founder of VisualMind & GameRefinery

- Currently leading Performance Marketing at Fiskars Group

Current ventures:

- Driving data and AI/Machine Learning based initiatives forward to support current businesses and identify new opportunities

- Building and coaching cross-functional teams of experts to solve complex business challenges

- Working of various other projects related to marketing efficiency improvement

How did you get into Marketing Mix Modeling (MMM) in the first place?

I started in 2006 at Marketing Clinic when I took the responsibility of conducting Marketing Mix Models for our clients. I think I have conducted 50-100 MMM projects during my career.

Which method would you say works best to prove the value of marketing?

Really broad question. I believe that a combination of different methods works the best. Marketing Mix Modeling is good in many cases but it has it's drawbacks.

For example, it doesn't take into account the order of which individual consumers have been exposed to different marketing activities which on the other hand can be very important in digital marketing.

What would you say are the biggest challenges with MMM for large advertisers today?

There exists two profound challenges that need to be solved:

1) Getting the required data in the right format in a scalable way

The quantity of data is hardly a problem for most companies nowadays.It’s the quality that creates challenges when it comes to advanced modeling projects.

Sales data is often in standardized form, whereas marketing data seems to have as many forms as there are companies. As the algorithms require the data to be in specific form, having varying year-on-year data either renders the analysis useless or forces the analytics team to go through laborious data cleaning exercise.

To streamline the MMM, companies should understand early on in what format different data sets provide the highest value, ensure supporting guidelines and aim to automate the data pulls for best possible efficiency. Because not only does getting the required data in the right format in a scalable way accelerate the process but it can also enable far more granular analyses. Which relates to the second challenge,

2) Making the modeling at granular level enough

Broader, more generic analyses are great indicators for depicting the overall company performance. But they provide poor lever for any profit optimisation efforts, as they’re not able to extract the effects of the smaller investments or individual product-level activities.

More granular modeling capabilities are especially important when a company has large product portfolio, as the optimal media mix and promotional strategy can vary a lot across product categories and brands. Without decent transparency into the marketing effectiveness, best practices might be applied blindly in situations that would require different approach, which in turn leads to weaker business results.

So, in order for the modeling to provide meaningful, impactful and applicable insights for profit optimisation, the modeling need to be on granular level enough. This means going from total sales levels to product category, store and even individual product level.

What’s your advice to succeed in MMM?

1. Define clearly the questions you want to get answers to.

As in any research, the objectives define the methods and tools that can be utilised in finding out the results. If you want to find out the overall Marketing ROI, summed metrics might be sufficient. But if you want to know how different product group perform and compare against each other, you have to go on more granular level.

2. Think also beforehand what decisions / changes you are ready to make in case of different results.

For example, if TV seems to be not working based on the analysis, are you ready to reduce investments to it or what you would do? Analysis is pointless if you are not going to change anything based on the results :)

How marketers can survive in current situation?

- Pay close attention to your own channels and well as your online media mix.

I believe that importance of own and digital channels will grow now that people are spending more time in virtual forums.

- Have a good plan on what to do in each channel.

Each channel has its own unique characteristics, capabilities and core audiences. Use them to your advantage when setting up new campaign and media plans.

- Have the right people executing the plans.

Build a team of (current and future) marketing superstars that are able to handle the pressure and deliver superior value for your customers.

- Be prepared to test and fail.

You can’t trust blindly that what you knew before will apply for the current situation. When resources become scarcer, most consumers have to re-evaluate their consumption behavior and available product assortment. Re-examine the markets through trial and error.

And don't forget to measure the impact of your activities.


Curious to learn more? Book a demo.

Marketing Mix Modeling

Related articles

Read more posts

No items found!