Definition
The elapsed time between the first period in which a model's accuracy or population divergence crossed its declared threshold and the retraining action that addressed it.
The Measurement Gap
There is no metric that tells you how much time elapses between a measurable drop in model accuracy and the retraining that corrects it, which means a manual retraining process carries no accountable latency.
Formula
RRL = median(t_retrain - t_breach) across breach episodes
Proposed formula, pending validation pilots.
At a Glance
Decision Supported
Whether manual retraining is keeping pace with drift, and whether the operation needs an automated trigger rather than a monitoring habit.