5 Practical Tips to Build Better Data Products from Eric Weber

Welcome to the latest issue of the Product Management Learning Series - a series of live streaming events and newsletter articles to help you level up your product career! ??

?In our seventh installment, our speaker was Eric Weber, the Senior Director of Data Science - Experimentation, Metrics & Inference at Stitch Fix. His focus is enabling high quality decision making across the company, supporting and critically evaluating product strategies and leading a team of talented engineers and data scientists. He also writes a weekly newsletter called "From Data to Product" focused on data product management and thinking about data science as a product. Check out and subscribe to his newsletter: https://ericdataproduct.substack.com/??

Below are the main takeaways from my conversation with Eric:

Know the limitations and imperfections of the data to inform product decisions.

When tasked with making hard decisions with imperfect data, Eric emphasized that as a product manager you don’t suddenly go from not making a data-informed decision to making a data-informed decision. It’s not a binary process. Instead he stressed that there are degrees to which using data can actually help you make a better decision, or better understand the risks involved in making that decision. It is often assumed that data is always going to help you make a better decision, but that's not always the case. The way to get past this is to understand the limitations and imperfections of the data upfront, and then use your qualitative judgment as a product manager to make the best decisions given the predetermined constraints.?

Gather feedback and create internal advocates to champion your data products.?

Companies often try to measure everything and attach a metric to it. Eric suggested that while data is extremely important for making business decisions, there are often data quality issues that can prevent data from being as insightful as we want it to be. Eric recommended focusing on creating internal advocates for the data product instead. It is crucial to get the users to realize the challenges, risks, and constraints of not having the data product in the first place. If you can demonstrate the value to internal stakeholders that the data products provide, then you can gain the internal champions to help your product succeed. Getting feedback from stakeholders early and often, in all forms, including the good, the bad, and the ugly, can make the difference between the success and failure of your data product.??

Create documentation with the intended audience and future use in mind.

Documentation has become a common practice at many top companies. Eric mentioned that while many companies do focus on documentation, there are many companies that have nothing written down at all, or if they do have something written down, it may be one single Google Doc owned by someone no longer at the company. Eric shared that when you are planning on documenting something, ask yourself ‘what do you think is going to allow people or enable people to consume that information the most easily?’ By always keeping the future use of the data and documentation in mind, you can better plan for the most efficient way to capture that information so that a particular audience can best consume it.

Make your data product easy to use and focus on the benefits.

One pitfall that product managers want to avoid is creating a data product that has so many features, or is too overly complicated, that stakeholders aren’t very clear on how to use it. To avoid this confusion, it can be beneficial to think about both the direct users of the product and the edge users. Thinking through not only how these users could benefit, but also, and more importantly, how the users could understand the benefit they are experiencing. Solutions to combat this could be creating an FAQ document or having a tutorial for first time users. Thinking about other less common use cases upfront can help product managers preemptively create the necessary extra features that prevent technical debt going forward.?

Break down the role separations and focus on end-user benefits to bring out the best of the product manager and data scientist duo.?

One way to promote collaboration between product managers and data scientists is not to think about the separate responsibilities of each role, but to think about the end user and what would make the end user's experience better. If data scientists are more open to think about what it means to build and own a product, and product managers are more open to working to understand the nuances of data science, while both keeping the end users in mind, they can both benefit immensely.?

?? Special kudos to Andrew Altschuler for drafting this article.

Next up,

• July 29: Building Impactful Educational Products with Ope Bukola
• August 5: [Special Edition] Breaking into Product Management with Diego Granados
• August 12: B2B Growth and Strategy with Elena Verna
• August 19: Building High-Performing Teams & Transitioning to People Leadership with Liz Li
• September 9: Applying the Lean Product Process to Build Great Products with Dan Olsen
• September 23: [Special Edition] Breaking into AI Product Management with Marily Nika
• October 7: Use Behavioral Economics to Design Better Products with Kristen Berman
• October 14: Building Subscription Products & Becoming a Thought Leader with Robbie Kellman Baxter
• October 21: Content Creation & Product Community Building with Felix Watson

More sessions to be added later… subscribe to stay tuned!?

Learn more about the Product Management Learning Series and view past recordings here.