Your question is Diagnosing Production Model Degradation. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
If a model's performance degrades in production, what is your diagnostic process?
Describe and implement a reproducible workflow that distinguishes data-quality failures, training-serving skew, population drift, label-delay effects, calibration problems, and genuine concept drift. Use reference validation data and recent production prediction logs with delayed ground-truth labels where available. Define the monitoring metrics, statistical tests, slice analysis, rollback criteria, and retraining decision. The implementation should avoid leakage and should identify whether the issue affects the whole population or specific segments.