Agile marketing increases speed, transparency, and customer satisfaction. But making your marketing organization responsive and personal isn’t a simple matter. Marketing and technology need to work together to enable better decisions and better customer experiences. Moritz Hahn, senior vice president at Zalando, offers perspectives on how to get marketing and technology to work together to disrupt marketing again.
Why did you change the way you do marketing at Zalando?
McKinsey: What prompted the change in how Zalando conducts marketing?
Moritz Hahn: Ten years ago, when we started Zalando, the way we did marketing was really disruptive. It made us really successful. But a lot of tech innovations means that things change on almost a daily basis. It was very risky not to change the marketing department, and we realized we needed to disrupt marketing. We figured out that we actually had to combine tech and marketing. It was actually only two years ago that we combined tech and marketing into one team.
McKinsey: What steps did you take to embed tech into marketing at Zalando?
Why did you embed the tech teams into the marketing department?
Moritz Hahn: We embedded the tech teams into marketing because we wanted algorithms to help us with our millions of day-to-day decisions, such as when and how much marketing money to invest. These algorithms then had to execute on all those millions of day-to-day decisions. We first had to automate all the marketing processes. Our second step was to rigorously A/B test. With a randomized control group and a test group, we were able to isolate the impact of each euro invested in marketing. This brought the magic of machine learning to the game, as it let us see causal inferences on a day-to-day basis in millions of decisions.
In the beginning, the machine-learning system took time to catch up with the more classical way of doing marketing. But now we can see that it brought us to a completely different growth path than we were on before. So if you put together automation and A/B testing, you have the two most important ingredients for a machine-learning system: learning and decision making.
Over time, this machine becomes better and better; it learns and learns.
As an example, in the past we had a couple of thousand creative assets on Facebook. Nowadays, we have up to millions of creative assets on Facebook at any given point in time. This is possible because we can automate not just content creation and uploading but also the routing and bidding for the placement of the different creative assets as well. As a result, we can deliver a personalized marketing experience to almost every individual customer.
Of course, on top of automation and A/B testing, we also needed to excel on the classical marketing performance topics, such as data tracking, attribution modeling, or steering.
Self-disruption in the fashion industry
McKinsey: What has your experience at Zalando taught you about the digital marketing transformation?
What were the major learnings for you?
Moritz Hahn: It’s hard work. To lead an organization through this change requires leadership. In the beginning, in particular with machine-learning systems, you don't see the benefits immediately. You have to go through a time when the performance benefits aren’t apparent. With machine-learning systems, the results come at a later stage. This makes leadership and a commitment from the top of the organization extremely important.
I think we also learned that we have to constantly disrupt ourselves from the inside. Disruption can’t be a one-time endeavor. The ability to disrupt from the inside is true agility and will help us to continuously evolve our business.