Abstract

Purpose: Online food delivery platforms face challenges to operational efficiency due to increasing demand, a shortage of drivers, and the constraint of a one-order-at-a-time delivery policy. It is imperative to find solutions to address the inefficiencies in the food delivery industry. Bundling multiple orders can help resolve these issues, but it requires complex computations due to the exponential increase in possible order combinations. Research design, data and methodology: This study proposes three bundle delivery systems—static, dynamic, and hybrid—utilizing a machine learning-based classification model to reduce the number of order combinations for efficient bundle computation. The proposed systems are analyzed through simulations using market data from South Korea's online food delivery platforms. Results: Our findings indicate that implementing bundle systems extends service coverage to more customers, increases average driver earnings, and maintains lead times comparable to standalone deliveries. Additionally, the platform experiences higher service completion rates and increased profitability. Conclusions: This suggests that bundle systems are cost-effective and beneficial for all stakeholders in online food delivery platforms, effectively addressing the inefficiencies in the industry.