Cycle time is how long a unit of work takes from start to finish. For code changes, it is often measured from the moment a pull request leaves draft status until it is merged, though teams define the start and end points differently. Shorter cycle times generally mean work flows through review and testing without long waits.

Teams track it to find bottlenecks. If most of the time is spent waiting for review or for slow builds, that points to where to improve, and trends over time show whether changes to process or tooling help.

The number needs careful reading. It depends entirely on how it is defined, so only figures measured the same way can be compared. Averages can be distorted by a few very long items, so the median and higher percentiles usually give a fuller picture. A shorter cycle time is also not automatically better, since rushing review can raise defects, and a measure that is turned into a target tends to be gamed.

Define the start and end points in writing, report the median and a high percentile, and look at it alongside quality indicators such as failed deployments or reverted changes.