Optimization of Sales Forecasting for Mipcinta 50 WP Product Based on Tracking Signal

Abstract

PT. XYZ is one of the leading companies in Indonesia engaged in the production of bio-products, pesticides, and various agricultural chemicals. The company is supported by adequate production facilities, an extensive distribution network, and competent resources, enabling it to compete and continue to grow in the industry. Nevertheless, PT. XYZ still faces various operational challenges, such as excessive production capacity (overload), unmet production targets, and suboptimal utilization of labor. Information regarding the planned production quantity becomes an important factor to ensure that production targets can be met on time. In addition, fluctuations in consumer demand and sales uncertainty become the main challenges in the production planning process. To address these issues, this research applies two forecasting methods, namely Weighted Moving Average and Exponential Smoothing. Validation of the forecasting methods is conducted using a tracking signal to ensure the reliability of the prediction results. This study aims to estimate the sales of Mipcinta 50 WP products at PT. XYZ and determine the most accurate forecasting method. The analysis results show that Exponential Smoothing yields a Mean Absolute Percentage Error (MAPE) of 12.165% with a sales estimate of 22890.01 kg for August 2024. The Weighted Moving Average method obtains a MAPE of 12.774% with a sales estimate of 22050 kg for the same period. Based on these results, the Exponential Smoothing method provides a higher level of accuracy and demonstrates better stability in following demand patterns. This is also supported by the tracking signal values, which mostly fall within the range of +6 to -6, making the method more consistent and more accurate for forecasting the sales of Mipcinta 50 WP products at PT. XYZ.