This paper examines how commonly used demand forecasting accuracy metrics, particularly Mean Absolute Percentage Error (MAPE), behave across hierarchical product levels in manufacturing environments and which structural operational risks may remain concealed when analysis is limited to aggregated indicators. Using a 24-month dataset of actual and forecasted demand, product mix structure, and inventory indices, the study distinguishes between volume forecast accuracy and mix forecast accuracy. The results show that while aggregate-level MAPE may appear acceptable, substantial forecast errors persist at lower hierarchy levels and are not neutralized by aggregation. A key contribution of the study is the treatment of product mix deviation as an independent source of operational risk rather than a secondary effect of volume forecast error. Empirical evidence indicates that mix deviations can persist even under high aggregate forecast accuracy and are associated with non-linear and asymmetric inventory responses. In particular, demand overestimation leads to sharper inventory increases than the reductions observed under demand underestimation. Adopting a decision-oriented perspective, the study conceptualizes forecasting dashboards as decision support systems that support managerial interpretation of forecast errors and help identify structurally hidden demand risks in operational contexts.

Keywords

Demand forecasting; MAPE; Product mix deviation; Inventory behavior; Hierarchical forecasting.