Amanda Drury - one of the worlds most experienced marketing analyst

How did you get into MMM?

I have always liked math and science. When I studied, I was very focused on math and science and didn't even know marketing analytics existed. At the same time, my jobs when studying were bartending, waitressing, customer service – interacting with people. Understanding people and how they think has always interested me. Eventually, I ended up with an international MBA program focused on international marketing and finance, where I went into a Corporate Finance position.

The best thing about being in a financial planning analysis role is that I get to really understand how a company works and all the levers you could pull for growth in a company and margin. With Marketing Mix Modeling I get to combine my mathematic background, marketing background, and my consumer behavioral interest!

MMM triggered my intellectual curiosity. It’s like trying to fit a puzzle of what's going on and why. Why is it one campaign doing so much better than the others? Is the creative that much better at communicating value to a consumer? Can we look at the targeting? I fell in love with it and today I have been in analytics for eight years.

What led you to working with MMM at Bel?

I started out working for consultancy agencies like Nielsen and IRI. I specialized in marketing mix modeling on the consultancy side for about four and a half years before I went into a client insights position to work with clients across all analytic solutions. In 2015 I worked on the marketing mix for Bel and when my contact had to relocate, Bel reached out to me. They were looking for someone to join their team to focus analytics on marketing effectiveness. They wanted to get more technical with their planning. For me, this was a great opportunity to get a better understanding of the manufacturer's side of analytics.

“First do the regular marketing mix and then add consumer panel data”

At Bel, I got a lot of freedom to come up with a plan for our first marketing mix and I looked into many vendors. One of the things that I felt was so valuable for clients our size is to first do the regular marketing mix and then add consumer panel data on to that analysis. So we did our regular marketing mix for Babybel and had the learnings, but then we added consumer panel data to understand what are the best tactics for getting new buyers? And what are the best tactics to keep retained buyers? It was great to do this and really understand how we could apply the insights to our business and have our marketing work harder for us.

Can you share a success story, using MMM?

I think part of my success story is having worked on so many different cases, for many different clients and both on the consultancy and brand side. I had one convenience store retailer as a customer that I worked with consistently for four years and we did multiple models, in the beginning, to test city/media/messaging. We were able to, shift their marketing spend so that they doubled their returns within those four years.

“They doubled their returns within those four years”

Another client I worked with was a green category organization. They spend a lot of money in their community. Like Twitter party with Moms and PR. As you know, those activities tend to be kind of difficult to read in a mixed model because they are not really driving sales per se. They are driving awareness and engagement. When we were modeling these activities with the traditional IRI grocery models, we weren't having any indicators that those programs worked at all. But when we were able to get whole foods data and amazon.com data, all of a sudden we saw them pop. It was one of those, ”yes moments”. We’ve had a hunch that this was working and could finally verify it.

Do you see other challenges with MMM?

Well, the situation I just described is one of the big challenges working with analytics - how to account for tactics that are meant to drive engagement? You're not really supposed to see sales lift on it. And how to make sure management is not cutting all these valuable activities because they’re not seeing good returns on it.

Another big challenge is to measure the effect of marketing long-term and brand equity. There is a risk of focusing on what’s easier to measure and look at the marketing mix from a short term perspective. Then you optimize your tactics but might miss other activities needed to make your brand relevant longterm.

How will you develop MMM for Bel in the future?

Well, it goes back to being able to communicate on all different levels. Working for IRI, I was used to delivering things to analytic people. Now I had to shift my whole thinking and the presentation to a bigger picture. It’s a challenge stepping away from being technical and instead focusing on what it means for the business.

Bel has really embracing utilizing data to drive insights. We have added different dimensions like pricing, assortment, discreet choice, or the marketing mix. The next step is using MMM and all of these learnings not only for marketing but for many other areas to optimize our business. These will feed into our strategic planned process

“The next step is using MMM not only for marketing but for many other areas to optimize our business.”

Do you use multi-touch attribution as well today?

Honestly, I don’t think multi-touch attribution is valid for CPG companies. I think it's very, very hard to prove out because I've done some of that. For instance, you have messages coming at you from every different angle in CPG, so how do you really get that multi-touch attribution to be accurate?

Best advice to succeed with MMM?

I think the biggest part that determines the success is, before going into modeling and data collection, to really lay out what your questions are. Make sure that the modelers understand all aspects of the questions so they can tweak the model in the right way. I always stress the importance of having a discovery period talking to the brand- and sales teams and getting as much input from everyone to really identify the most important questions. Then make sure you have the data to answer them and not stress into modeling and data collection.

If anything was possible, how would you like to improve or develop MMM?

I wish I had the budget to do MMM more often, I wish it did not take 12 weeks to get results. I also wish that it wasn't just IRI or Nielsen that had store-level data as I believe that is key to control for trade (Discounts/Features/Display),

“I wish I had the budget to do MMM more often, I wish it did not take 12 weeks”

I also wish that more companies would do the consumer mix aspect and overlay the consumer groups. I think that is learning for organizations our size that is so invaluable and frankly, many people do not even know this exists.