Change failure rate for low-deploy teams

Under 5 deploys per week, CFR is statistically noisy. Here is how to report it without overclaiming.

Oliver Wakefield-Smith

Founder, Digital Signet. Built siemcost / monitoringcost / itbudgetcalculator.

Updated 21 Jun 2026

Direct answer

Below roughly 5 deploys per week, weekly change failure rate swings on single events. Use a 30 or 90 day window, report the count alongside the percentage, and consider deploy-frequency improvement before CFR optimisation IT Revolution DORA2024.

The statistical floor

Standard rule of thumb: aim for at least 30 events in the window before publishing a percentage. A 2-deploy-per-week team needs a 15-week window to clear that floor. The fix is not to invent deploys; it is to widen the window or report the raw count.

Sample-size warning thresholds

Dashboard convention: green if > 50 events, amber 20-50, red < 20. This site renders an amber banner under the gauge when total_deploys in window falls below 30DORA 2024 Report2024.

Deploy frequency comes first

For a team shipping less than 5 times per week, the highest-leverage DORA improvement is usually deployment frequency itself. Faster, smaller deploys reduce blast radius and make CFR easier to measure DORA Four Keys2024.