Factory metric tree

Start at the decision

A metric tree should begin with a decision the factory needs to make. The top question might be whether delivery is improving safely, whether a platform change reduced friction, or where additional capacity would help. The next branches describe the dimensions that can explain that question.

For a delivery goal, branches might include throughput, flow, stability, quality, work mix, developer experience, and customer outcome. Each branch should lead to defined measures and source events. Change lead time might use commit and production deployment events. Review waiting might use pull request and review timestamps. Customer value may require a product analytics system.

Keep proxy and outcome separate

A metric tree is useful when it shows which measures are direct outcomes and which are proxies. More pull requests may indicate activity, but it does not establish customer value. Lower queue time may indicate better flow, but it does not establish that the released change worked. Label these relationships so a convenient leaf does not become the goal.

How Weave can help

Weave can populate the development branches of a metric tree and retain the examples behind a trend. Teams can use that evidence to choose a focused experiment, then join the result to deployment, incident, product, and developer feedback sources.