DORA's four
- Deployment frequency — how often production deploys happen. High performers: multiple per day.
- Lead time for changes — commit to production. High performers: under a day.
- Mean Time To Recover (MTTR) — how long to fix a production incident. High performers: under an hour.
- Change-failure rate — % of changes that cause a production issue. High performers: 0–15%.
Why DORA's four are powerful
They balance speed (deploy freq, lead time) with stability (MTTR, change-failure rate). A team gaming any one metric regresses the others. Optimizing all four together is what high-performing engineering orgs actually do.
↳ code-level metrics with caution
Defect density (defects per KLOC) and test coverage (% of code under test) are useful internally but easily gamed. A team mandated to hit 90% coverage will write trivial tests; a team mandated to drive defect density to zero will stop logging defects. Use as inputs, not as targets.
What PMs should ask
- What's our deploy frequency? (Want: not weekly batches.)
- What's our MTTR for the last 5 incidents? (Want: trending down or stable.)
- What % of our last 50 deploys caused an incident? (Want: under 15%.)