changefailurerate.com

Change failure rate FAQ

The questions that come up most. Answers, not vendor sales talk.

OW
Oliver Wakefield-Smith
Founder, Digital Signet. Built siemcost / monitoringcost / itbudgetcalculator.
Updated 21 Jun 2026
What change failure rate is good?

Below 2% puts a team in the top ~16.7% of the 2024 DORA distribution (the legacy ‘Elite’ band). Below 10% is roughly above median. 10-22% is the broad middle. Above 22% is the long tail.

Does CFR include rollbacks and hotfixes?

Yes. A deploy that requires a rollback or a hotfix is counted as a failed deploy in the numerator. The rollback or hotfix itself is typically excluded from the denominator as a fix-only deploy if the exclusion is enabled.

How is CFR different from defect-escape rate?

CFR is failures per deploy. Defect-escape rate is bugs reaching production per release, regardless of whether they caused a failure. Two teams can have identical defect-escape rates and very different CFR if their deploy cadences differ.

Should AI-generated code be tagged separately for CFR?

Yes, if you want to attribute CFR changes to AI adoption. Tag via commit-message footer or Git note. The 2024 DORA AI capabilities research finds throughput-up, stability-modestly-down is the dominant pattern.

Why did DORA drop Elite / High / Medium / Low?

Because by 2024 the cluster structure of the survey data no longer matched the original tier boundaries. Continuing to publish the named tiers would have mis-described the distribution. DORA shifted to percentile bands.

Is this calculator's data sent anywhere?

No. All math is client-side. Nothing is logged or transmitted. CSV export runs entirely in the browser.

Why no GA4 / no demo / no signup?

Independent reference, not a vendor funnel. The site exists to give a VP Eng a defensible number in 90 seconds. Signup forms are the opposite of that.

More questions?

See the full Methodology and Glossary pages, or run the calculator and tune the inputs.

Continue reading