Calculate release velocity trends across sprints to track delivery acceleration.
Release velocity normalizes release counts from two consecutive periods to a common releases-per-week rate and compares them to detect a trend: change% = (current_velocity − previous_velocity) ÷ previous_velocity × 100. A change beyond ±5% is classified as accelerating or decelerating; smaller fluctuations are treated as stable, since release counts over short periods naturally have some noise.
velocity = releases / time_period_weeks
Comparing raw release counts across periods of different lengths (or with different numbers of working days/holidays) can be misleading — normalizing to a rate makes the comparison meaningful regardless of exact period length.
A gently accelerating or stable trend generally reflects a maturing delivery pipeline; a sharply decelerating trend can signal growing technical debt, testing bottlenecks, or an increasingly risk-averse release process worth investigating.
Deployment frequency measures a single period's rate against DORA's performance tiers; release velocity specifically tracks the trend between two periods to answer 'are we speeding up or slowing down,' which deployment frequency alone doesn't capture.
Be consistent — if your deployment frequency metric counts all production deployments including hotfixes, use the same definition here, since mixing definitions between periods would create a false trend signal.