The Value of Patience in Online Grocery Shopping

TL;DR

The paper addresses the urban externalities of rapid online grocery delivery and investigates customer patience as a scalable demand-side lever. It introduces a theoretical framework where patient delivery windows increase shareability probability $P(\theta)$ and enables efficient bundling, paired with a network-based dispatching approach that minimizes mileage under pickup-delivery constraints. Using two large Dubai datasets ($>8$ million orders), the authors show a convex trade-off: modest patience yields substantial mileage and CO2 reductions (approximately $30\%$ and $20\%$, respectively, for a $5$-minute delay) with diminishing returns beyond $10$ minutes, and higher order density enhances savings. They further quantify life-cycle emissions via GREET with Dubai inputs and find potential cost savings for operators (around $13\%$ with a $10$-minute patience) and argue that the strategy generalizes to other cities and last-mile contexts, offering a practical path to balance convenience with urban sustainability.

Abstract

Since the COVID-19 pandemic, online grocery shopping has rapidly reshaped consumer behavior worldwide, fueled by ever-faster delivery promises aimed at maximizing convenience. Yet, this growth has also substantially increased urban traffic congestion, emissions, and pollution. Despite extensive research on urban delivery optimization, little is known about the trade-off between individual convenience and these societal costs. In this study, we investigate the value of marginal extensions in delivery times, termed customer patience, in mitigating the traffic burden caused by grocery deliveries. We first conceptualize the problem and present a mathematical model that highlights a convex relationship between patience and traffic congestion. The theoretical predictions are confirmed by an extensive, network-science based analysis leveraging two large-scale datasets encompassing over 8 million grocery orders in Dubai. Our findings reveal that allowing just five additional minutes in delivery time reduces daily delivery mileage by approximately 30 percent and life-cycle CO2 emissions by 20 percent. Beyond ten minutes of added patience, however, marginal benefits diminish significantly. These results highlight that modest increases in consumer patience can deliver substantial gains in traffic reduction and sustainability, offering a scalable strategy to balance individual convenience with societal welfare in urban delivery systems.

Figures (4)

• Figure 1: Overview of the developed order bundling and fleet dispatching framework. (a) Overall workflow. Orders are bundled based on spatial and temporal constraints (in light green except for vehicle characteristics), then dispatched to available vehicles. Key outputs, including delivery delay, mileage, fleet size, and emissions, were estimated. (b) Bundling strategy to form an order shareability network. (b-I) Orders that meet the predefined spatial and temporal proximity constraints in a batch are considered as bundling candidates; (b-II) These candidates form a shareability network, where nodes represent orders and cliques indicate feasible bundling opportunities. (c) The fleet dispatching algorithm. (c-I) At the end of a batch, orders issued during the batch are collected (and bundled) and vehicles idle positions are estimated; (c-II) Order ready times and the weighted order-vehicle matching are computed; (c-III) depending on the assignment and PUD, vehicles able to arrive at the pickup point on time or earlier are assigned, and their next idle positions are updated; (c-IV) For orders unreachable in time by any vehicle, a new vehicle is generated at the pickup point (vehicle V4), thus increasing the fleet size.
• Figure 2: (a) Fraction of shareable orders for different values of vendor popularity $\lambda$ as a function of additional delay (patience $\theta$). (b) Estimated fraction shareable orders as a function of the customer patience based on theory and data.
• Figure 3: The tradeoff between average delivery delay (patience $\theta$) with (a) total mileage savings, (b) life-cycle CO2 emissions, and (c) fleet size.
• Figure 4: The influence of order density on the tradeoff between average delivery delay (customer patience $\theta$), and total mileage savings.