With the emergence of e-commerce in grocery retail, the food supply chain faces new challenges. In a new paper Shenle Pan et. al. of MINES ParisTech focus on such a challenge regarding the successful delivery of on line grocery orders.
Due to the perishability and sensitivity of some grocery items, customer attendance is often critical for the successful delivery. As a solution to this problem, this paper introduced a two-stage methodological approach that utilizes customer-related data to schedule transportation plans. This is done by first estimating the probabilities of customer attendance/absence at different point of times during a day and then using these estimations in a way that satisfies a company’s key performance indicators (e.g. maximize the probability of attended delivery while minimizing travel distance covered by delivery trucks).
Using electricity consumption data
The paper presented an experimental study to investigate how a customer’s historical electricity consumption data can be used to estimate time windows with a lower probability of not being home. The best time windows were then used in a model in order to plan the deliveries of online orders to customers aiming at improving delivery success rate. A numerical study has been conducted to demonstrate the effectiveness of the proposed approach that shows its potential in the delivery of online grocery shopping.
Besides increasing the rate of successful deliveries, the proposed approach can help e-tailers better understand the habits of their customers and thus the optimal delivery time for them. It can also be considered as a useful tool for dynamically pricing different delivery options and as a mechanism for time slot management. From a customer’s point of view, the approach can improve customer satisfaction as it can reduce unnecessary traveling to pick up missed orders or long telephone calls required to re-arrange deliveries. It is also obvious that the approach can easily be used in different business cases in urban freight transportation and last mile logistics (e.g. non-food items, general merchandise) where attended home delivery is critical and alternative solutions cannot be easily offered.
This study is among the first ones integrating data mining techniques in urban freight transportation. Some prospects can thus be identified to this line of research. For example, one may test the approach with other customer-related datasets, e.g. water consumption, historical deliveries, or with different data mining techniques in order to compare accuracy and performance.
The attendance probabilities can also be used in different vehicle routing models or for different reasons such as delivery time slot pricing. Further, the proposed methodological approach can also be generalized from e-grocery commerce to all businesses that provide home delivery service.
For more information click here.