What, Exactly, Does a Data Product Manager Do?
In 2014 I excitedly looked at job descriptions for this new role called “data scientist”. After reading about 50 job descriptions, I messaged a friend at Facebook at the time and asked “what exactly does a data scientist do?” Then I asked the same question to 4 others I knew across tech, finance and real estate. I got 5 different answers.
Data product management (and technical product management) are in much the same place that data science was in 2014. People generally knew it was important. People generally understood that it had some different qualities than an analyst role. People described what it was at their company, creating a huge variety of “what is a data scientist” articles across the industry.
So, what does a data product manager do? Rather than get into the fine-grained details which likely vary across companies and industries (just like data science), I think it is useful to focus on 5 themes of a data product manager’s work. I’m creating this from my understanding of the role, discussions with other product leaders across industry and my belief about where the role is headed in the near future.
If you read the sentence above carefully, you’ll notice I am presenting my perspective on what a data product manager does. This isn’t intended to be the gold standard. Instead, I’m aiming for something useful to spur discussions and help us think about defining the role.
Theme 1: They are a product manager. Earth shattering insights here, available all day. The reason I say this is that data product managers must be skilled at the fundamentals of product management. Success in this role doesn’t start with data, it starts with the customer and understanding their needs. Success requires mapping those needs to potential product solutions. Success requires creating a roadmap of how to get from here to there and understanding how that affects the customer experience.
Theme 2: They understand the value of data as an asset. There’s a lot of conversation about the value of data. But data doesn’t inherently contain insights or value (except in some very special cases). Data product managers need to think about the investment required to get the value out of data. The investment required to turn data into an asset. This is not trivial. It often requires being able to understand algorithms, data management, data quality and modeling, on top of the product expertise.
Theme 3: They find the balance between manual and automated. The number of emails I get about the next great “automated insight” is likely larger than the number of words in this article. Automation sounds pretty great, but getting automation right and making sure it is the right thing for the customer is really hard. There’s a strong tendency to want to automate things around data because the manual version is “hard” or “slow”. Knowing when the manual version allows the customer to extract the most value, and then knowing when it is time to move to automation is a critical skill.
Theme 4: They understand what others are doing in the industry and learn as much as they can. The data product industry is moving incredibly fast. If data product managers put their head down and only focus on what is happening internally, they are going to miss leveraging the learning, mistakes and changes happening at other companies. The best data product managers read a lot. They talk to people outside of their company. They go to conferences. They learn continuously.
Theme 5: They navigate and think about the build vs. buy conversation. As the data ecosystem becomes more complex, there are also multiple products that exist to reduce that complexity. Some things that were hard 5 years ago are now solved problems because companies have created solutions around them. Knowing when to leverage an existing product vs. build it internally requires understanding internal capabilities, external product capabilities and thinking about the costs to the business.
Over the coming years, you’ll read many job description for data product managers across areas of practice (experimentation, analytics, data management), across companies and across industries. I strongly believe that all of these roles will entail the 5 themes above.
If you’re looking to get into this space, looking at your comfort level in the 5 themes above is a good place to start.
Thanks for reading!