Skip to main content

Leadership through Data Science at Roblox

June

05, 2020

by kragle_rock


Product & Tech

By all measures, Roblox has grown extremely fast in the past few years. Whether through the lens of consumer engagement, developer content, or employees, each kind of growth has posed unique challenges to scaling. The Data Science & Analytics organization’s mission is to increase our speed, frequency, and acumen in making decisions to build the right product. We provide insightful recommendations that shape the product roadmap, and by building and productionalizing models that automate decision-making. And in order to do this effectively, we embed ourselves as partner members of product teams.

Product managers at Roblox are in the driver’s seat for making decisions regarding the product roadmap… not exactly a groundbreaking concept. Roblox has a strong vision-based culture to building product, providing a beacon of light far in the distance rather than short-term goals. To be successful, each leader must be able to communicate their vision, and even to go so far as pausing the building process until that vision is clear. Data science can and should play a significant role in building that vision and how it will be achieved. Our data scientists serve as navigators to our product managers in the driver’s seat.

As navigators of the product roadmap, data scientists answer the big question, “how do we get from here to there?” With such a large question to answer, a service-oriented approach would be inefficient on a PM’s cognitive load. We don’t want product managers to be asking specific questions like “should I turn left here?” every 30 seconds. Therefore, we can only succeed if data scientists are an autonomous partner in this journey to build product.

Autonomy in any organization can be difficult due to the many cross-functional relationships required to achieve success. But it becomes easier when everyone is guided by the same principles. Roblox is driven by five core values:

  • Take the Long View
  • Own It
  • Self Organize
  • Get Stuff Done
  • Respect the Community

Data scientists apply those values in their own functional ways. It starts by “taking a long view” for our product goals, translating the qualitative goals the product is looking to achieve into a quantitative metric. Context is extremely important as proxy metrics may be simplified abstractions and need to mature with our data capabilities. A data scientist must also own these metrics and become the stakeholder for what makes sense to track and, oftentimes more important, what not to track in order to keep the team focused. “Self-organization” is expressed through autonomy as data scientists strive to discover the underlying structure of a metric and look for opportunistic levers. They anticipate the next set of questions to answer and don’t wait for others to initiate a request, in essence building out their own analytical roadmap of understanding.

Analytical excellence is achieved through “getting stuff done.” This can mean making recommendations to the product roadmap with profound analytical insight, designing hypothetical experiments where needed, and scaling the effort it takes to answer questions through automation and modeling. “Respecting the community” can be viewed through the lens of mentorship and teamwork across the organization and product team. Data scientists share their learnings, techniques, and context to those closest around them and expect the same in return. We don’t play a zero-sum game.

To apply these values productively, a data scientist must bring certain characteristics to tackle the problems appropriately. Having an intrinsic drive, a strong curiosity and the need for betterment are all important characteristics to hold. Accompanying those with a strong level of communication and the ability to provide feedback and ask for feedback in return are the pillars to ensure that we better ourselves just as much as we look to better the product.

When you realize data science and analytics is more than a compilation of techniques and methodologies, and focus on improving the quality and accelerating the frequency of decision-making, then you have truly achieved the status of data scientist at Roblox. We look to bring the world together through play and have a fun time doing so.


Neither Roblox Corporation nor this blog endorses or supports any company or service. Also, no guarantees or promises are made regarding the accuracy, reliability or completeness of the information contained in this blog.

This blog post was originally published on the Roblox Tech Blog.