Avg. from merge to deploy
The average amount of time it takes for code changes to go from being merged into the main codebase to being deployed to production (where users can actually use the new features).
Where to find it
- Health > Benchmark > Add new metric
- Health > Team Insights > Add new metric
Interpretation
Larger companies
-
High: Less than 10 hours - indicate near real-time CD: auto-promote on green, small batches, canary/blue-green, infra-as-code, and quick rollbacks. Watchouts: insufficient soak time or reviews, off-hours deploys without coverage. Mitigate with progressive delivery, strong observability, runbooks, and staffed deploy windows.
-
Medium: Between 10 and 24 hours - a healthy enterprise norm that balances safety and speed: accommodates change windows, compliance checks, regional rollouts, and coordination with SRE/ops. Maintain with policy-as-code (auto-approve low-risk changes), prod-parity staging, draft release notes, and pre-scheduled cutovers.
-
Low: More than 24 hours - suggests bottlenecks: manual CAB gates, release trains, long/flaky end-to-end suites, artifact promotion delays, environment contention, cross-team dependencies, or freeze periods. Improve by slimming gates via risk-based approvals, parallelizing/caching tests, pre-building artifacts, adding ephemeral envs, region-by-region rollout, feature flags to decouple deploy from release, and setting SLOs for deployment latency.
Smaller companies
- High: Less than 1 hour - indicate a very fast deployment process, often driven by strong CI/CD automation, well-tested code, and solid DevOps practices. These processes should be maintained and continuously refined.
- Medium: Between 1 and 4 hours - reflect a reasonable deployment speed, typical for teams with some automation and a balanced focus on speed and safety. There may be manual steps, so further automation opportunities should be explored.
- Low: More than 4 hours - suggest a slow deployment process, potentially caused by manual steps, infrastructure issues, or testing bottlenecks. These inefficiencies should be investigated and the deployment pipeline optimized.
Custom Dashboards
Don't forget you can anytime include any metric in your Custom Dashboards.
Updated 10 days ago