What's DAT? Three Case Studies of Measuring Software Development Productivity at Meta With Diff Authoring Time
Moritz Beller, Amanda Park, Karim Nakad, Akshay Patel, Sarita Mohanty, F. Garberson, Henri Verroken, Andrew Kennedy, Ian G. Malone, Vaishali Garg, Pavel Avgustinov · 2026
This paper introduces Diff Authoring Time (DAT), a precise, yet conceptually simple approach to measuring software development productivity that enables rigorous experimentation. DAT is a time-based metric, which assesses how much active work time engineers take to develop self-contained changes. It uses a bespoke telemetry system integrated with version control, the Integrated Development Environments, and the Operating System. We validate DAT through observational studies, surveys, visualizations, and descriptive statistics. At Meta, DAT elevates the internal tool development workflow to the scientifically grounded, experiment-driven development flow already present for external-facing products. As such, DAT enables rigorous experimentation on long-standing software engineering questions like “do types make development more efficient?” DAT has powered experiments and case studies on more than 20 projects at Meta. Here, we highlight (1) a controlled experiment on introducing mock types, which showed that typed mocks in tests yield a 14% DAT improvement, (2) a case study on the development of automatic memoization in the React compiler, a 33% improvement, and (3) a framework to estimate the thousands of DAT hours saved annually through code sharing (> 50% improvement). Overall, this paper contributes a novel, yet straightforward way to measure development velocity, DAT. It describes how the use of this metric has enabled rigorous experimentation on development productivity at industry scale, and paves the way toward a more scientific mindset in internal product development.