Stop measuring developer productivity

The author argues that engineering leaders should shift their focus from measuring developer output through metrics like code commits and pull requests to assessing developer effectiveness and eliminating operational friction. Common sources of friction such as slow CI/CD pipelines, flaky tests, and poor diagnostics waste significant engineering hours that are typically hidden in productivity dashboards. Furthermore, the author warns that while AI coding tools can speed up initial code writing, they risk shifting bottlenecks downstream if the underlying engineering systems and pipelines remain slow. Open article →