What do you measure?
| Metric | Definition | Denominator |
|---|---|---|
| Decision latency | Median days from receipt to recorded call | All gated opportunities in the quarter |
| Override rate | Share of calls reversed by authority above the gate | All recorded calls in the quarter |
| Late-reversal rate | Share of pursuits killed or withdrawn after week two | All submitted pursuits in the quarter |
| Evidence completeness | Share of records with all evidence fields filled at the call | All records in the quarter; missing rows reported, not imputed |
Volume and capacity reporting (workload, stage age, delivery overlap) belongs to pipeline metrics; the data-entry register is the pipeline register. This page measures decision quality: whether calls come fast, stand, and rest on evidence. Rising win rate alongside falling late reversals suggests the gate works: suggests, not proves, because market mix confounds everything and this page claims no causation.
How do you review?
Quarterly, from the register: received, gated, and decided cohorts kept distinct, denominators stated, missing data declared. Overrides get reasons reviewed one by one: a pattern of delivery vetoes means the gate misreads capacity; a pattern of late reversals means week-one reads are decorative. Sunk-cost persistence has its own page (the zombie bid) and its rate inside these numbers is the one to watch first.
What does this mean for the bid/no-bid call?
Nothing this quarter, and everything next quarter. KPIs tune the process that makes the calls; they do not make any single call. A gate measured honestly improves faster than a gate defended loudly.