참고 답변
I track metrics across three categories, delivery performance, quality, and team health, because optimizing only one dimension creates problems. For delivery performance: velocity (story points completed per sprint, tracked as a trend over time, not used for team comparison), cycle time (time from starting work to deployment, broken down by work type), throughput (number of items completed per time period), and sprint goal achievement rate (percentage of sprints where committed goals were met). For quality: defect escape rate (bugs found in production vs. caught in testing), code coverage percentage for automated tests, mean time to recovery (MTTR) when issues occur, and technical debt ratio (tracked through SonarQube or similar tools). For team health: team satisfaction scores (from regular anonymous surveys), employee retention and turnover, team predictability (how often estimates are accurate), and participation in ceremonies (attendance and engagement). What's critical is that I use these metrics to drive conversations, not to judge people. Low velocity doesn't mean the team is lazy, it might indicate complexity in the backlog, technical debt slowing development, or unrealistic estimates. I present metrics in retrospectives and ask ‘What does this tell us? What should we try differently?' I also avoid vanity metrics, measures that look good but don't actually indicate success. For example, high code coverage means nothing if the tests are poor quality. The goal is actionable insights that drive continuous improvement.