What is change failure rate?

Change Fail Rate (CFR) is the percentage of production deployments that result in degraded service and require immediate remediation such as a rollback, hotfix, fix-forward, or patch. It is one of DORA's five current metrics and measures deployment instability rather than deployment frequency.

CFR = (Failed deployments ÷ Total deployments) × 100

A "failure" is any deployment that causes a degraded service experience requiring corrective action. This includes rollbacks, emergency patches, and hotfixes — but not planned feature flags or A/B tests.

How it fits the current DORA model

Change Fail Rate is one of five current DORA software delivery performance metrics. It belongs to the instability factor together with Deployment Rework Rate. The other three metrics—Change Lead Time, Deployment Frequency, and Failed Deployment Recovery Time—describe throughput.

The current DORA Quick Check reports normalized 0–10 scores rather than permanent Elite, High, Medium, and Low thresholds. This page therefore shows the measured percentage and a transparent directional score: score = 10 − (CFR ÷ 10). Use DORA's official Quick Check for research-based industry benchmarking.

How to reduce change failure rate

The most effective levers are: comprehensive automated testing (unit, integration, and end-to-end), feature flags to decouple deployment from release, smaller and more frequent deployments (smaller changes have lower blast radius), and automated rollback on failed health checks.

Counterintuitively, increasing deployment frequency often reduces CFR over time — smaller deployments are easier to test and easier to diagnose when they fail.

Measurement rule: Count only production changes that require immediate intervention, and keep the same application and measurement window when comparing results over time.

CFR vs error budget

CFR measures deployment quality. Error budget measures service reliability. They are related but distinct: a low CFR means your deployments rarely cause incidents, but your service could still have reliability issues from infrastructure failures, traffic spikes, or third-party dependencies. Use both together for a complete reliability picture. See the Error Budget Calculator for the other side of this equation.