DORA Metrics

No Change
adopt
First Added:June 22, 2026 Updated: July 1, 2026

DORA metrics are the DevOps Research and Assessment team’s validated measures of software delivery and operational (SDO) performance. The framework tracks throughput and stability with four core metrics, plus reliability as a fifth measure. We adopt them as the default engineering health baseline under DevOps. Treat them as directional delivery evidence, not executive value metrics on their own.

Blurb

DORA’s research program represents nine years of research and data from over 36,000 professionals worldwide. Our team conducts research to understand the practices, processes, and capabilities that enable teams to achieve higher levels of software delivery and organizational performance.

Summary

The four keys (software delivery performance):

MetricDimensionWhat it measures
Deployment frequencyThroughputHow often you successfully release to production
Lead time for changesThroughputCommit (or work start) to running in production
Change failure rateStabilityShare of changes that degrade service or need remediation
Failed deployment recovery timeStabilityHow long it takes to recover from a failed deployment or incident

DORA reports and tooling may still label the fourth metric time to restore service (MTTR). Same intent: restore user-impacting service quickly.

Fifth metric (operational performance):

MetricWhat it measures
ReliabilityHow well services meet user expectations (availability, latency, performance, scalability)

Read all metrics together. High deploy frequency with a high change failure rate is not high performance.

Performance tiers: DORA research clusters teams from survey data each year. Tiers are emergent benchmarks, not fixed universal thresholds. Use DORA Quick Check for current industry comparison. Use tiers for trend tracking, not vanity rankings.

Reporting altitude and audience:

AudienceWhat to showWhat to avoid
Delivery teamRaw DORA metrics, drill-down dashboardsBusiness outcome claims the data cannot support
Dependent teamsTeam-level trends tied to shared servicesOrg-wide rollups that hide local context
Executives and budget holdersValue story with 1-2 supporting metricsLeading with DORA alone

DORA answers “how healthy is delivery?” It does not answer “why should we fund the platform?” Map each metric to a stakeholder concern before rolling it up.

Example link chain (stability → capacity → outcome):

DORA signalIntermediate capabilityStakeholder outcome
Lower change failure rateFewer fire drillsMore predictable release dates
Faster recovery timeLess unplanned reworkMore capacity for features
Higher deploy frequencySmaller batchesFaster feedback on product bets

Without that chain, measurement cost looks like overhead.

Where DORA fits in the garden:

PracticeRelationship
DevOpsDORA measures delivery culture outcomes
Continuous DeliveryLead time and deploy frequency reflect CD maturity
Continuous IntegrationShorter lead time starts with fast, reliable CI
SRERecovery time and reliability align with error budgets and ops learning
Incident ManagementChange failure rate and recovery time need sane incident flow
DashboardingWhere teams visualize DORA trends for operators

When to use: Team and org dashboards, improvement experiments, platform direction checks, and postmortem prioritization.

When to skip as the whole story: Executive business cases, platform ROI narratives, and product outcome reviews. Pair DORA with stakeholder-defined value metrics and narrative.

Details

TopicNotes
ScopeSystem-level outcomes, not individual developer output
Data sourcesCI/CD events, deploy logs, incident records, VCS timestamps
ToolingDORA Quick Check; native DORA views in some platforms; custom Dashboarding on OpenTelemetry or pipeline metadata
Anti-patternsGaming metrics, local optimization, comparing unlike systems, executive decks that start with four keys
ComplementsSPACE and flow metrics for developer experience; business KPIs for altitude

Definitions (practical):

  • Lead time for changes: elapsed time from code committed to successfully running in production.
  • Change failure rate: percentage of production changes that result in degraded service, rollback, or hotfix.
  • Failed deployment recovery time / MTTR: elapsed time from production failure to restored service for users.

Collecting honestly:

  • Define “deployment” and “failure” the same way across teams before comparing.
  • Split metrics by service or product line when architectures differ.
  • Review quarterly against last year, not against a static “elite” label from an old report.

References