Start with measurement, not tooling
Most DevOps programs fail because they start by buying tools. We start with the four delivery metrics popularised by the DORA research: deployment frequency, lead time for changes, change failure rate and time to restore service. These show where delivery is hurting, and they give you a baseline against which any improvement can be shown honestly. Surveys of developers help here too, because the people doing the work often know exactly which step hurts, and they rarely need a tool to tell them.
We then trace a change from commit to production, noting every wait, handoff, manual approval and environment queue. Typical bottlenecks are long-lived branches, slow builds, shared staging environments, manual deployment steps and weak monitoring that makes every release feel dangerous, so people batch changes and make releases bigger and scarier. We present these findings as a visual value stream map, which makes the discussion about waiting time and rework concrete for managers and engineers alike.
