Standard : Change Failure Rate (CFR)

Description

Change Failure Rate (CFR) measures the percentage of changes that result in a failure in production—such as incidents, degraded service, or urgent fixes. It is one of the four key DORA metrics and reflects the quality and reliability of your delivery process.

A low CFR indicates strong testing, effective review, and stable deployment practices. CFR helps teams balance speed with safety and build confidence in frequent releases.

How to Use

What to Measure

  • Track how many production changes (deployments) caused a failure.
  • Failures include P1/P2 incidents, customer-visible errors, rollbacks, or hotfixes.

Formula

CFR = (Failed Changes / Total Changes) x 100

Instrumentation Tips

  • Use incident tracking systems to flag failed deploys.
  • Define a consistent taxonomy for “failure” (e.g. severity levels, SLA breaches).
  • Correlate deployment events with incident timelines or rollback activity.

DORA Benchmarks

Performance Level Change Failure Rate
Elite 0–15%
High 0–15%
Medium 16–30%
Low 46–60%

Why It Matters

  • Quality signal: Reflects deployment health and reliability.
  • Customer trust: High CFR leads to more disruptions and lower satisfaction.
  • Team confidence: Frequent failures erode confidence in the release process.
  • Learning: CFR insights help guide investment in testing, observability, and resilience.

Best Practices

  • Invest in robust automated testing before merge and deploy.
  • Use progressive delivery (e.g. canary, blue-green, feature flags).
  • Instrument for observability and alert on post-deploy anomalies.
  • Standardise rollback processes for failed changes.
  • Review failures in blameless postmortems and track recurring patterns.

Common Pitfalls

  • Using inconsistent or vague definitions of “failure.”
  • Tracking only rolled-back changes (some failures persist unnoticed).
  • Ignoring near-misses or degraded performance incidents.
  • Overfocusing on CFR at the expense of deployment frequency.

Signals of Success

  • Fewer incidents caused by changes, even with frequent deployments.
  • Failure causes are understood and addressed at the root.
  • Post-deploy issues trend downward over time.
  • Testing and monitoring practices evolve based on CFR insights.
  • [[CoE/Measures/Delivery Performance/Deployment Frequency]]
  • [[Lead Time for Change]]
  • [[Mean Time to Recovery (MTTR)]]
  • [[Postmortem Completion Rate]]

Aligned Industry Research

  • State of DevOps / DORA (2021–2023): Elite performers maintain CFR below 15% while deploying multiple times per day.
  • Accelerate (Forsgren et al.): CFR is a key measure of software delivery performance that correlates with organisational outcomes like profitability, productivity, and customer satisfaction.