This study examines the problem of minimizing job lateness in the paper manufacturing industry, focusing on cut-size machine scheduling under fluctuating demand. Historical demand data (2018-2019) were forecast using Double Exponential Smoothing (DES) and Holt-Winters' Triple Exponential Smoothing (TES), with accuracy assessed via Mean Absolute Percentage Error (MAPE). The forecasts informed scheduling models for single- and parallel-machine environments using dispatching rules, including Earliest Due Date (EDD), Shortest Processing Time (SPT), Critical Ratio (CR), Longest Processing Time (LPT), and Least Slack Time (LST). Results show Holt-Winters' TES achieves the most accurate forecasts, while EDD consistently minimizes lateness, reducing delays by more than 70% compared with alternatives. These findings highlight the value of integrating forecasting and scheduling to enhance machine utilization and delivery performance. The framework offers practical guidance for demand planning and resource allocation in export-oriented manufacturing sectors facing high demand variability.