Vercel leverages dozens of metrics created from one heterogenous table to drive business, technical, product, and operations decisions across the company. Our approach has empowered technical and non-technical stakeholders to jump into their analytical discovery from the metrics table with more frequent iterations and less involvement from the data team. In addition, our practice has increased our stakeholders' ability to self-serve by connecting the data with a documented lineage and ownership of how and where the metrics are compiled. Before our current structure, stakeholders and the data team spent significant amounts of time deconstructing complicated queries and tracking down the data sources' owners to understand why shifts were happening in the data. Today, when an unexpected change occurs in the data, our stakeholders can quickly identify what actions are driving the abnormal activity, reach out to the data source owners, and inform the data team what information needs to be updated and why. Centralizing data and metadata used in creating our many metrics has increased the number of stakeholders that can participate in analytics, decreased the time needed to troubleshoot outlier events, and removed the data team as a dependency for all data-related tasks.
Thomas leads analytics and data science initiatives at Vercel, scaling insights across engineering, product, and design. He focuses on making data modeling, analytics, and decision-making more accessible for all users. Before landing in tech, Thomas was an elementary teacher and a carpenter. This experience has enabled him to think about the non-technical users of data and how data can be intimidating without the proper support. Outside of work, Thomas likes to spend his free time working in a community garden, dreaming up construction projects, and hanging out in the backyard with his family.