Most teams collect more build and compile figures than a release council can absorb. The failure mode is not missing data—it is bringing the wrong sheet into the room.
Three questions
For each candidate measure, ask:
- Decision — If this number is bad, what concrete action do we take before the cut?
- Freshness — Is the figure young enough to describe this release train?
- Owner — Who will speak to it without reading from a dashboard legend?
If any answer is empty, the measure belongs in a lab notebook, not on the checklist.
A workable maximum
Four measures is plenty for a go/no-go call. Beyond that, the council starts debating definitions instead of risk. Park the rest for a monthly build guild review.
Related reading
See the assessment ledger for how we column these decisions during an engagement.