Evaluating system reliability in online food delivery with stochastic food processing time

In the online food delivery (OFD) market, service reliability is often compromised by operational uncertainties. This study evaluates system-level reliability under the dual uncertainties of courier availability and stochastic food processing times (FPT). We introduce the Delay-Aware Multistate Online Food Delivery Network (DMOD), formally integrating spatiotemporal constraints and FPT uncertainty into reliability assessment. An efficient three-stage exact computational framework renders this evaluation tractable. Numerical experiments across three spatial topologies reveal that targeted ETA rationalization for mismatched orders approximately doubles system reliability with marginal impact (approximately 3%) on service promises, especially under stable preparation. Furthermore, optimal courier allocation is topology-dependent: targeted deployment excels in clustered networks, whereas uniform expansion suits dispersed ones. Managerially, dynamic ETA calibration and topology-aware resource allocation serve as effective mechanisms to enhance reliability.