Rework rate
Also known as Engineering rework, Rework ratio
Definition
Rework rate is the proportion of work that must be repeated or revised because the earlier result was incomplete, incorrect, or no longer suitable. Its calculation depends on the work unit and the evidence used to identify rework.
Define the repeat
Rework can mean a change reverted after release, a pull request reopened after review, or an implementation revised because a requirement changed. Those are different events and should not be combined without a reason.
An illustrative team might define rework as an unplanned production deployment made in response to an incident. That is close to DORA's deployment rework rate. Another team might track changes to the same file within thirty days. The second measure can reveal churn but cannot establish that the change was incident-driven.
Connect the signal to a cause
High rework can reflect unclear requirements, missing tests, a difficult codebase, or healthy iteration on an uncertain problem. Review comments and incident timelines help distinguish those possibilities.
Do not use rework as a score for an individual. Shared systems, product decisions, and review practices influence the result.
Use the rate to choose an improvement
Look for concentrated patterns. If rework clusters around one integration, improve its contract or test environment. If it follows a particular release process, inspect the handoff and rollback path. If it appears after large changes, reduce batch size and improve feedback earlier.
Track the rate with quality outcomes and delivery timing. A lower rate is useful when it reflects fewer avoidable repeats, not when work is simply moved outside the reporting boundary.
How this relates to Weave
Weave's quality and output views help teams investigate repeated edits, reverts, and changes associated with a release. Use those signals to find likely sources of rework, then confirm the cause through code and team context. Repeated edits do not automatically mean the original work was poor.
Explore Engineering intelligence