Software factory throughput

Choose the completion boundary

Throughput can mean merged pull requests, completed work items, production deployments, or changes that reached a user. These are related events, but they are not interchangeable. A team that merges changes behind a feature flag may have high merge throughput while customer-visible delivery remains unchanged.

State the work population and period with the measure. Count completed items over time, and keep the type and size of the work available for interpretation. A team doing a large migration may complete fewer items while making important progress.

Pair quantity with flow and stability

Throughput is more useful with cycle time, work in progress, and queue time. DORA places deployment frequency and change lead time alongside stability measures because a delivery system should move changes quickly and safely. A higher count accompanied by more failed deployments or rework is a different result from a higher count with stable quality.

Avoid converting throughput into an individual quota. The system determines much of the available flow, and increasing work in progress can make the queue larger without increasing completed work.

How Weave can help

Weave can show code output and delivery patterns together, with pull request and review context available for investigation. Teams can ask whether throughput changed because the factory removed a bottleneck, because work became smaller, or because more recovery work entered the system. Product outcomes and deployment records are still needed to establish whether the added throughput created value.