Your question is Diagnose Underperforming Model. 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).
You shipped a model that looked strong in validation, but it is underperforming after launch. The team wants you to figure out why the results are worse and whether the issue is in the model, the data, or the decision threshold.
What would you do if your model is underperforming?