Every team measures forecast accuracy differently, and almost nobody states which one they mean. MAPE punishes low-volume SKUs unfairly; WMAPE fixes that but hides SKU-level misses. Worth writing up properly at some point.