<![CDATA[This study examines the dynamic impact of TikTok advertising expenditure on skincare product demand using a time-series framework. The objective is to evaluate whether incorporating digital advertising spending as an exogenous variable improves sales forecasting accuracy on social commerce platforms. Monthly secondary data from Company X, covering shampoo and cream products sold on TikTok from October 2024 to October 2025, were analyzed. The study compares ARIMA, ARIMA with Trend, and ARIMAX models. Stationarity was tested using the Augmented Dickey–Fuller test, while model selection was based on AIC, MSE, RMSE, and MAPE. The results reveal heterogeneous demand characteristics across products. Shampoo demand shows strong persistence and relatively stable patterns, with advertising expenditure having a positive but limited incremental effect on forecasting accuracy. In contrast, cream demand is highly sensitive to advertising intensity. The ARIMAX model significantly outperforms alternative models for cream products, producing substantially lower forecast errors. These findings indicate that promotional elasticity differs across product categories. Managerially, the results suggest that promotion-driven products require tighter integration between marketing expenditure planning and operational forecasting, while habitual products may rely more on historical demand patterns. This study contributes to digital marketing and forecasting literature by empirically demonstrating the product-specific effectiveness of social media advertising within a dynamic time-series context.]]>
