Data Speaker Series: Thorsten Dietzsch on Building Data Products at Zalando

I will never stop learning is the first line of our creed. As data analysts, scientists, and engineers, we work in a fast-moving field with a variety of subtopics. There is a lot to learn from others in the industry or in academia. To create learning opportunities, we’re inviting speakers to share their insights in our new Data Speaker Series. Our first speaker was Thorsten Dietzsch. He is a product manager for personalization at the fashion ecommerce company Zalando. In his talk, we learned how data science is organized at Zalando and what product management for data-based products means.

Thorsten Dietzsch

Thorsten Dietzsch

Thorsten has experience in data and science, ranging from mathematical modeling for neuroscience and analyzing particle collision data at CERN, to working on various data science projects at Zalando.

Team setup

Zalando’s data scientists work either in the research lab or in cross-functional teams organized around products. Some product teams are responsible for user-facing data products such as search and recommendations. Other teams work on internal data science products such as A/B testing tools. Depending on what they build, data product teams consist of data scientists, data engineers, software engineers, and sometimes designers or project managers. Product teams work with product managers, like Thorsten, who define what they build and why.

What is Product Management?

At Automattic, we don’t have dedicated product managers for data science. Instead, we take turns leading data science projects and product management is thus an activity that each of us gets to do to some extent. Thorsten’s talk inspired a new mindset with which to approach product management and provided some tools we can use for it. Thorsten’s slide illustrates what product management is:

product management

Product managers marry what customers want with what’s technically possible and what generates value for the business. They drive product development from the discovery phase, through definition and design, to delivery. They help the team to understand the why behind what they’re building.

Thorsten contrasted his role to the one of a project manager: project managers help teams to be productive. At Zalando, they are experts in agile methodologies, and use daily stand-ups, retrospectives, and demo sessions. Product managers, on the other hand, set each team’s roadmap and vision. They prioritize, act as a bridge to internal stakeholders, and represent the customer’s view.

Discovery for data products

Thorsten encourages data science teams to think of themselves almost as small startups within the company. A tool that he likes is the Business Model Canvas,  as adopted in the book Value Proposition Design. It provides key questions for team positioning and product discovery such as:

  • Who are we building this product for?
  • What is our value proposition to them?
  • In which format are we going to provide our product? For data scientists, this could be an API or a report, for example.
  • Who are our key partners in building this product?

In this snippet from his talk, Thorsten explains the Business Model Canvas for a data science team building data products in collaboration with other teams in the company:

Key Learnings at Zalando

Thorsten shared some key learnings from building data products that apply not only to Zalando but to any company that puts data science into production. Among other points, he mentioned that:

  • The bottleneck for building a great data product is often getting enough labelled data, and not finding the best algorithm.
  • When working with internal stakeholders like the marketing department, explainability is often as or even more important than the accuracy of the algorithm.
  • It’s important to not only take responsibility for shipping a product but also for testing it. It’s very likely that the data scientists in your team know more about A/B testing than internal stakeholders.

Recommended reading

Thorsten recommended several books:

Many thanks to Thorsten for sharing his product management knowledge with us here at Automattic. To learn more about Zalando, check out their blog.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s