DORA Metrics

DORA metrics tell you whether your delivery engine is healthy. They say nothing about whether you're building the right thing.

The DORA research program (Forsgren, Humble, and Kim, starting in 2014) identified four metrics that reliably predict software delivery performance: deployment frequency, lead time for changes, change failure rate, and mean time to recovery. The four metrics hold each other in tension; nobody games deployment frequency by shipping empty releases because the other three would expose it.

In the Impact-Outcome Model, DORA metrics live at the Activity and Output levels. Deployment frequency is a practice metric (how often do we ship?). Lead time, change failure rate, and MTTR are flow metrics (how well does our delivery pipeline perform?). They measure the capability of your delivery system, not the value of what it delivers.

That distinction matters. A team can achieve elite DORA scores — deploying multiple times per day with sub-hour lead times and near-zero change failure rate — while shipping features that don’t move any customer or business outcome. The metrics tell you the engine is running smoothly; they say nothing about whether you’re driving somewhere worth going. Without pairing DORA with outcome and impact metrics, you risk optimizing a delivery machine pointed at the wrong target.

DORA is most valuable when a team is maturing its software delivery and operations capability. If deployments are painful, infrequent, and failure-prone, DORA gives you a clear signal of what to improve and whether your improvements are working. But once delivery is healthy, the question shifts from “can we ship?” to “should we ship this?” That’s where outcome metrics take over.

The Discover-Option-Action Cycle treats DORA as one option in a broader toolkit: validate that delivery performance is the constraint you need to relieve, then decide whether DORA’s metrics are the right signal for your situation. A team running embedded firmware on a two-week release train doesn’t need daily deployment frequency as a north star. Without that validation step, DORA becomes another mandated dashboard and Goodhart’s Law takes over. DORA is powerful when it’s in service of outcomes; it’s just a vanity dashboard when it isn’t.

Resources

• Nicole Forsgren, Jez Humble, and Gene Kim, “Accelerate: The Science of Lean Software and DevOps: Building and Scaling High Performing Technology Organizations” (IT Revolution Press, 2018)
• Impact-Outcome Model — where DORA metrics sit in the Activity-to-Impact causal chain
• Discover-Option-Action Cycle — the model for deciding which metrics fit your context
• Goodhart’s Law — the risk when any metric set becomes mandated