Several data practitioners have chronicled the challenges faced by new managers charged with building a data function: tying data work to business value, standardizing tools and metrics, establishing data literacy throughout the org, etc. However, fewer have put forth actionable guidance on how to transform a dysfunctional data team and culture. In this post, Erik Bernhardsson tells the fictional tale of a new Head of Data at an early-stage startup to reveal important lessons about how companies can become data-native. He advocates for a data team structure wherein reporting and people management are centralized, but the backlog and work management are decentralized, and data scientists are embedded within business units. In addition, his narrative is interspersed with analysis of how and why certain behaviors emerge and clever tips on how to become data-driven (e.g., when implementing self-service tools, designate data scientists to educate users on their use).