The manufacturing industry faces major challenges in managing often uncertain demand fluctuations. This applies to PT. XYZ, a company operated in the production of automotive spare parts. Uncertainty in demand results in the risk of overproduction which increases storage costs or underproduction which has the potential to reduce customer confidence. The aim of this research is to predict production demand implementing the decomposition method, which separates historical data into trend, seasonal, cyclical and random components. This method is able to provide more accurate predictions by utilizing historical data for the last three years. The research results show that there are seasonal trends that influence inventory levels, with peak inventories usually occurring in K1 or K3, while K4 tends to be the period with the lowest inventories. With this method, PT. XYZ can increase production efficiency, minimize operational costs, and optimize production capacity according to market needs. Based on inventory estimates for 2024 and 2025, a consistent fluctuation pattern can be seen every quarter. In 2024, the first quarter (K1) is estimated to have inventory of 23,107 units, which then decreases slightly in K2 to 22,058 units. Inventory increased again in K3 to 23,210 units, before experiencing a significant decline in K4 of 21,602 units. This pattern repeats in 2025, with K1 showing an increase to 24,381 units, followed by a decrease in K2 of 23,258 units. Inventory rose again in K3 to 24,455 units and fell again in K4 by 22,746 units.