Delivery time reliability in on-demand food delivery: Heterogeneity from attribution effects

The popularity of on-demand food delivery (ODFD) has been observed globally, in which platforms connect restaurants with consumers with digital infrastructure and crowdsourced riders. With fierce competition, it is imperative for practitioners to optimize decision-making models which heavily rely on insights into consumer behaviors. Time perception is considered a key determinant of consumers’ ODFD usage. Scarce attention is paid to consumers’ attitudes towards delivery time reliability. Therefore, with a stated choice experiment, this study aims to investigate and monetize consumers’ value of reliability (VOR) by adopting the scheduling-based approach. The study explicitly models random heterogeneity using a mixed logit specification, and systematic heterogeneity with an integrated choice and latent variable model. Specifically, consumers’ attributional attitudes of service outcomes pertaining to different service parties, namely delivery riders, restaurants, and platforms, are incorporated in heterogeneity modeling. The results show that cost and expected delivery time negatively affect ODFD usage utility, while the negative effects of long expected times are significantly weaker for orders placed during peak hours or from distant restaurants. On average, consumers hold positive attitudes toward early delivery, while the attitudes are highly heterogeneous. Moreover, late delivery negatively affects utility of ODFD usage, and the effects are also weaker for orders from distant restaurants. Regarding the attributional effects, the attribution of late delivery to restaurants would intensify the dissatisfaction, while the converse is true for attribution to platforms. In contrast, attribution to delivery riders has no significant effects on consumers’ dissatisfaction. In conclusion, this study contributes to the theoretical development of both reliability perceptions and attribution theory. Besides, empirical observations can effectively support the development of decision support systems in ODFD, allowing practitioners to improve ODFD services with more approaches. The findings would support the integrative optimization of service promise and dispatch algorithms, and context-aware optimization. Furthermore, ODFD practitioners can implement effective psychosocial intervention to manage consumer satisfaction with the findings on attributional effects as well.