Metrics, when used well, create clarity and alignment.
Used poorly, they drive fear and shallow optimization.
An EM’s job is not just to measure, but to measure without breaking trust.
Why This Is Hard
Engineering work is complex and often non-linear. When metrics are treated as targets rather than signals, shallow optimization might be done just to correct the metric.
Common failure modes include:
- measuring individuals instead of systems
- using metrics to evaluate performance rather than to surface issues
- reacting to short-term fluctuations instead of long-term trends
Metrics are not a scorecard. They are early warning signals of system and team health.
Good metrics help answer questions like:
- Where are we slowing down?
- Where is quality degrading?
- Where is work getting blocked?
How This Works in Practice
Measure systems, not people. Focus on team-level flow, reliability, and quality.
For example, a poor metric is the number of commits per developer — people have different coding and committing styles.
Similarly, measuring the number or size of merge requests can be misleading. A one-line change can sometimes be more impactful than a 30-file refactor, such as a large renaming change.
Instead, measuring DORA metrics is useful. They surface trends around lead time, review time, deployment frequency, and reliability.
The most important signal is often the simplest:
Did the team on track for the desired outcome within the planned timeframe? If not, that should trigger a conversation with the team.
Transparency builds trust when metrics are not weaponized.
Looking at trends rather than snapshots is critical. One bad week or month is usually noise. Long-running patterns are signals. Metrics should always be paired with context.
Takeaway
As an EM, you set the tone that metrics are used to improve the system — not to rank, pressure, or compare individuals.
When teams feel safe around metrics, they surface problems earlier — and that is the real win.
Top comments (0)