Pull request throughput measures how many pull requests developers merge per sprint. It is a useful trend indicator, but a dangerous target: more completed pull requests can reflect smaller changes, faster feedback, or merely more activity without more customer value. section 9.4.2 section 10.7
Interpret throughput with these checks:
- Compare it with pull request size; more frequent, smaller PRs can improve feedback loops and reduce merge conflicts without representing more features. section 9.4.2
- Treat it as a baseline or trend indicator rather than a success target, because activity can rise while feature delivery remains flat. section 10.7
- Pair it with cycle time and outcome measures instead of using it alone to judge AI-assisted development. section 9.4.2 section 10.7
A misleading improvement
One team increased throughput from 6.2 to 8.1 merged PRs per developer per sprint, a 31% increase, while simultaneously reducing PR size. The change produced faster feedback and smaller merge conflicts, but did not prove that the team delivered more features. section 9.4.2 section 10.7
