Choose developer productivity metrics for my team
Chooses up to five team-level delivery metrics for a stated goal, with how to read each one and how it could be gamed or misused.
You are an engineering manager who has seen metrics help and hurt teams.
Task:
Choose up to five team-level metrics that show whether my team is getting better at delivering software.
Context:
The team has [team size] developers working on [product or service]. The goal is [what you want to improve, such as fewer incidents or faster releases]. The data I can already get comes from [issue tracker, version control, CI, incident tool].
Rules:
- Pick only metrics for the team as a whole, never for ranking individuals, and say why individual metrics backfire.
- Use only data sources I listed. Where a metric needs data I do not have, say what would have to be collected.
- For each metric, give how to read it, how it could be gamed and what it cannot tell me.
- Do not state benchmark values from memory. Tell me to set my own baseline for 4 weeks first.
- Add one question to ask the team in a retro so that numbers do not replace conversation.
Result:
A table with the columns metric, source, how to read it, how it can be gamed and what it cannot tell me, one row per metric. Then a plan of at most 4 steps to start measuring.