# DORA Metrics

> A research-backed set of software delivery measures that keeps throughput and instability in view together.

- HTML version: https://robbiepalmer.me/ideas/dora-metrics
- Source: https://dora.dev/guides/dora-metrics/

DORA's current software delivery performance model uses five measures. Three describe throughput:

* change lead time;
* deployment frequency; and
* failed deployment recovery time.

Two describe instability:

* change fail rate; and
* deployment rework rate.

The combination is more valuable than any single number. Delivery frequency without failure data
can reward churn. Failure rates without lead time can reward avoiding change. The tension makes it
harder to improve a dashboard while making delivery worse.

These are system measures, best applied to one application or service over time. They are not a fair
league table for teams working in different contexts, and they do not explain which local practice
caused a change. I use them as outcomes and pair them with narrower evidence when testing a specific
intervention.

## Related ideas

- [Goodhart's Law](https://robbiepalmer.me/ideas/goodharts-law.md): When a measure becomes a target, pressure to improve the number can break its connection to the outcome it once represented.

## Where it appears

- Project: [Agentic Code Review](https://robbiepalmer.me/projects/agentic-code-review.md)

---

Markdown index of this site: https://robbiepalmer.me/llms.txt