In progress Designing sustainable last-mile delivery services in online retailing



The continuous growth of online sales together with the inefficiency of the last-mile of the e-commerce supply chain puts a lot of pressure on urban areas in terms of congestion, emissions and pollution. It is critical to increase the efficiency of the last-mile deliveries to enhance the financial and environmental performance of internet retailers. The primary goal of this project is to develop and evaluate decision support models and tools to facilitate the optimal design of different delivery service models in online retailing and to identify service models and corresponding operating strategies that provide the most benefits in terms of various sustainability criteria. To create more sustainable last-mile operations, the proposed research is organized around two PhD projects, one that particularly focusses at optimizing the delivery operations and one that focusses on the optimal design of the retail network. The research is part of larger collaboration between the Rotterdam School of Management, the Erasmus School of Economics and a consortium of companies in the area of online retailing.


E-commerce, distribution, transportation, vehicle routing

Time frame

2015 - 2019


Online retail sales continue to grow at a fast pace, even in times of global economic downturn. Online sales represent the only growth sector in the declining retail market. However, despite the potential of the internet as a sales channel, online retailers face many logistical challenges in the fulfillment of online demand. Internet order fulfillment, also called e-fulfillment, is generally considered the most challenging and critical part of the operations of companies selling physical goods online. Handling small individual orders and shipping them to the customers’ homes in a timely and cost-efficient manner has proven difficult. This is particularly the case for the “last-mile” delivery, i.e., the last leg of the e-fulfillment supply chain in which the deliveries are made to the customer. Due to the large number of small individual orders, the last-mile often covers many stops in urban areas. This results in disproportional high costs of the last-mile deliveries.

The efficiency of the last-mile not only impacts the profitability of internet retailing but also affects environmental and social performance criteria such as emissions and traffic congestion. For instance, vehicle-miles are directly related to emissions and the required number of vehicles for delivery has an impact on congestion. The continuous growth of online sales together with the inefficiency of the last-mile delivery puts additional pressure on the already congested urban areas. To increase the efficiency of the last-mile delivery, the proposed research develops and evaluates innovative service models and retail network designs to provide more sustainable business models for the future.

Next, we describe the most important and unique aspects of last-mile delivery operations in e-fulfillment. We highlight important challenges for which we believe that the most gains in sustainability and profitability can be made. Moreover, we suggest an area of research that will yield a promising new service model for e-fulfillment. Furthermore, we describe the unique aspects of retail networks which include online retail. We show that in this case there is much room for consolidating or differentiating different flows of goods, depending on the design of the network. We argue that research is required to determine when consolidating flows of goods increases sustainability and when consolidating flows is harmful. The design of these retail networks needs to be studied to facilitate innovative service models and to cope with the tremendous growth in demand that characterizes online retail.

Delivery operations:

Delivery operations - Home delivery

The logistical challenges of e-fulfillment are especially apparent in attended home delivery, where most often the customer has to be home to receive the ordered goods [1,2,4]. This is typically the case for products that cannot be delivered in the mailbox, such as groceries, electronics and apparel. To minimize the risk of not finding a customer at home for delivery, and thus creating unnecessary vehicle movements, the retailer and the customer may agree on a delivery time window. Several online retailers recently started to offer customers a menu of delivery time window options to choose from. The design of such a time window menu involves decisions on the timing, the length and the number of the offered time windows and the associated delivery fees.

The design of a time window menu to offer to customers involves trade-offs between customer service and operational efficiency. Offering wide time windows to customers for example, increases transport flexibility but will be disliked by customers, as they might have to wait a long time for their delivery. On the other hand, offering very small time windows at many different times during the day will be preferred by customers. However, this way there is the risk of many customers booking the same small time window, forcing deliveries to be made simultaneously by different vehicles. In practice this might even lead to disappointing the customer by making the delivery outside the time window when not enough delivery vehicles are available.

E-commerce allows online retailers the unique opportunity to differentiate the service offered to different customers in real-time [5]. E-grocer Ocado, for example, is appealing to the customers' environmental concerns by indicating which delivery time window would minimize the fuel consumption for their order. The company uses a ‘green’ van to indicate that a delivery van is already in the customer's neighborhood at a certain time. A better coordination of demand and supply through the smart use of time windows can help to further improve the performance of the last-mile delivery system.  

Delivery operations - Pickup points

In addition to different delivery service options to the customer’s home, more and more online retailers (such as Wehkamp, Amazon, Ahold and Google) also offer customers the option to directly have the goods shipped to a pickup location instead of their home address. This alternative way to bridge the last-mile allows customers to choose a convenient time to collect the goods without having to be home for the delivery. The pickup point may be staff-assisted or unmanned self-service lockers. Pickup points often also serve as drop-off points to facilitate the easy return of goods ordered online.  If pickup locations are suitably located, many customers do not even have to make a detour to visit them. For instance, a customer might have to visit a supermarket or gas station anyway, so a pickup point at such a location ensures delivery at no additional effort for the customer, while at the same time transport efficiency will be increased. An illustrative example of this is the recent rollout of a network of pickup points at metro stations by British supermarket chain Asda.

Delivery operations – Innovative service models

In recent years, we have seen the rise of several innovative last-mile delivery service models such as deliveries in the trunk of the customer’s car (, local shopping services (eBay Now), crowd-sourcing of last-mile deliveries ( and immediate return services for apparel products ( A major driver in these developments is the desire to reduce customer effort for receiving goods.

In essence, for the transfer of goods to take place, the supplier and customer need to agree on a time and location. For home delivery, the customer provides the location and the supplier and customer need to agree on a time. This usually requires efforts from both parties, as these agreements are usually made in the form of time windows. Hence, the customer needs to be at a specific location during the time window and the supplier needs to arrive at the location within the time window. For pickup points, the supplier and customer do not need to be at the same location at the same time for the transfer of goods. Still, effort is required from the customer to go to the pickup point, but the customer receives more flexibility in timing.

Currently, it has become very easy for customers to share information with the supplier. This offers new opportunities to coordinate the last-mile deliveries with the customer’s time schedules and travel patterns. The customer may for instance provide the flexibility to deliver at different locations based on the timing, e.g. before 17h at work or after 18h at home. This way, the customer does not need to change its behavior; it will receive the goods at no additional effort. At the same time, this offers the supplier relatively high flexibility, which can be utilized to select the time and place for transferring goods that maximizes profitability and sustainability. In this project, we plan to further develop this concept and determine its feasibility and its benefits.

Retail networks:

Retail networks – Flow consolidation

Multi-channel retailers typically use their existing conventional stores as potential pickup locations [3]. Not only do these pickup points require low investment costs, they also provide the opportunity to consolidate the online retail transport flow with the regular flow of goods to resupply the stores.

Furthermore, home deliveries are usually distributed using a hub system; goods are transported by truck to a local hub, typically a parking lot or (small) cross docking facility, where goods are transferred to smaller vans for home delivery. Letting the hub locations coincide with pickup points offers the option to consolidate flows of goods destined for pickup points and home delivery. For multi-channel retailers, conventional stores might even serve as both pickup points and hubs for home delivery.

In the design of the retail network, increased consolidation options might ensure that fewer trucks are necessary to ship all goods. However, their effects on transport efficiency and sustainability are unclear. Locating a home delivery hub at a pickup point or conventional store might yield longer distances to be travelled by the home-delivery vans. Even more detrimental for sustainability is the fact that most convenient locations for pickup point and conventional stores are in densely populated, and more congested, areas. Therefore, it is unclear whether the benefits of consolidating different transport flows outweigh the disadvantages.

Retail networks – Flow differentiation

Contrary to consolidation, online retailers might also improve efficiency and sustainability by differentiating transport flows. For example, business-to-business (B2B) flows and business-to-consumer (B2C) flows have very dissimilar characteristics. Business customers typically order large volumes, on a regular basis and prefer a specific fixed time of delivery. Whereas consumers typically order smaller volumes, at a highly irregular frequency and prefer varying times of delivery. Many online retailers do not distinguish between these customers and consolidate B2B and B2C flows without question.

However, differentiating might allow the supplier to exploit the regular ordering pattern of one customer group and mitigate the effects of irregularity of the other customer group. For customers with a regular ordering pattern, supply operations can be fine-tuned to increase transport efficiency and increase sustainability performance criteria. Furthermore, increased service by for instance offering a small time window in which the delivery will be made comes easier when customer demand is highly stable. Moreover, when differentiating, customers with an irregular ordering pattern do not disrupt delivery operations for the regularly ordering customers. The question remains whether the benefits of this strategy outweigh the benefits of consolidating flows.

Retail networks - Growth

The online retail market is currently characterized by fast growth, even in times of economic downturn. This poses challenges for designing an efficient and sustainable retail chain. When determining locations for pickup points and hubs and deciding on transport capacity, these long term decisions should take growth into account.

Furthermore, many multi-channel retailers deal with a decrease in turnover at their conventional stores. It seems that such retailers face undercapacity for their online channel, i.e., a limited number of hubs pickup points and vehicles, whereas they face overcapacity for their conventional channel, i.e., too much storage space and floor space at (expensive) store locations. In managing this shift from conventional retail to online retail, it should be investigated whether the overcapacity at the conventional channel can be used to manage growth of the online channel.


To implement the above described delivery operations and to design a retail network including an online channel in a sustainable way, economies of scale are required for which customer demand has only become sufficiently voluminous in recent years. This area is still very much in development and standard practices have not yet emerged. The challenge now faced by online companies is to deal with these large and complex planning and coordination problems.

Research goal

The primary goal of this project is to develop and evaluate decision support tools to facilitate innovative operating strategies and to analyze the design of retail networks including an online channel that provide the most benefits in terms of various sustainability criteria.

The proposed research is organized around two PhD projects, one that particularly focusses on optimizing the delivery operations and one that focusses on the tactical design of the last-mile retail network. These decision making problems are related in the sense that the performance of tactical decisions is dependent on the operational efficiency and vice versa. Decision making in both cases asks for sophisticated decision support tools. We aim to identify and formulate new optimization problems and to develop innovative optimization approaches to solve these problems. Our specific research questions include:

1.      How to optimize delivery operations?

1.1  What menu of time windows to offer to improve the sustainability of last-mile deliveries? The design of the delivery time windows, delivery lead-times and corresponding delivery fees involve complex trade-offs between customer convenience and operational efficiency. Online retailers are in a unique position to differentiate the service offering to different customers and to change this offering relatively easily, even in real-time. This allows the service provider to steer demand in a way that facilitates cost-efficient and environmentally sustainable last-mile deliveries. Exploiting these opportunities requires sophisticated decision support tools that incorporate the impact of the service offering decisions on the underlying routing and scheduling decisions.

1.2  Does offering both home delivery and pickup point delivery provide a sustainable service model? It is becoming more common that a customer is allowed to choose between home delivery and delivery to a pickup point. Depending on the number of customers choosing for the different options, delivery operations are affected, and hence the performance in terms of efficiency and sustainability. In potential, such a service model outperforms a model offering only home delivery or only delivery to pickup points. However, to reach the full potential, delivery operations need to be optimized, for which new decision support tools need to be developed. Moreover, the choices of customers determine whether in the end this service model is sustainable at all. This needs to be investigated, as well as the possibilities to influence customer choices using incentive schemes, such as using different delivery fees for the different options.

1.3  How to employ time based customer location information to improve sustainability? Modern information technology opens up the possibility for a customer to share his or her whereabouts with the supplier in a quick and easy way. Specifically, a customer can inform the supplier where he or she will be at different times of the day. A supplier can now choose a time and place of delivery that fits the customer’s schedule, such that the customer does not have to alter his schedule for receiving a shipment. Moreover, compared to home delivery within a specific time window or delivery to a pickup point before a specific time, the supplier gains a lot of flexibility to choose a time and place for transferring the goods. This flexibility can be used to improve transport efficiency and increase sustainability, while at the same time reducing customer effort for receiving a shipment. To exploit this additional flexibility, complex decisions have to be made to coordinate delivery operations. The development of new decision support model and tools is crucial to this end.

 2.      How to design the retail network?

2.1  How to design a retail network including home delivery hubs and pickup points in a sustainable way? The location of home delivery hubs and pickup points drive transport efficiency and sustainability. Suitably chosen locations limit the total distance travelled by delivery trucks and vans. Moreover, the relative distance of hubs and pickup points, and even other delivery locations such as conventional stores of a multi-channel retailer, determine whether consolidation of different transport flows is beneficial. Because of the large growth of online retail in recent years, and the expected continued growth, many online retailers now face the complex problem of expanding their retail network. New decision support tools are required to help decide where to locate new home delivery hubs and pickup points and which flows to consolidate to improve transport efficiency and sustainability.

2.2  Does differentiating transport flows for different customer groups improve sustainability? The main idea is that delivery operations are most efficient when dealing only with customers with a regular demand pattern. Therefore, it might be prudent to separate customers with regular and irregular demand patterns. This way, sustainability of the delivery operations for the regular customers can be improved by fine-tuning their delivery schedules, while the negative effects of irregular demand patterns on sustainable operation remain limited to the operations pertaining to these irregular customers only. Also in this case, decision support tools are necessary to attain the full benefits of separating, in order to properly answer this research question.

2.3  How to manage the fast growth of online retail, especially in the case of multi-channel retailers? The expected continued growth is an important factor in the process of deciding how to set up the retail network, e.g., where to place home delivery hubs and pickup points. Furthermore, multi-channel retailers typically face excess capacity in their conventional channels and under-capacity in their online sales channel. An important question therefore is whether the excess capacity in one channel can be used to make up for the undercapacity in the other channel. For instance, setting up a home delivery hub requires space for cross docking, or setting up a pickup point requires storage space. Excess space at conventional store locations might be used for this, limiting investment costs. The question that remains to be investigated is whether balancing out capacity of different channels, even though this is cost efficient, leads to a sustainable retail network.

The mathematical models and decision support tools that will be developed to answer these research questions are aimed at improving transport efficiency and reducing traffic congestion and emissions. In our analysis, it is vital to take into account the following features:

  1. Environment. How do the characteristics of the environment impact the efficiency and sustainability of the different last-mile delivery models? Which last-mile models are more appropriate for urban areas with a high population density and which ones work better in a low-density rural area?
  2. Customer behavior and service flexibility. How far is a customer willing to travel to visit a pickup point? Do customers want to make a detour from their commute? How flexible is the customer with respect to the offered time windows? How price sensitive is the customer?
  3. Demand volumes and growth. How does the efficiency and sustainability of a last-mile delivery system depend on the demand volume? What are the critical demand volumes to make different service models viable?
  4. Governmental sustainability policies. What impact do governmental policies such as vehicle restrictions, load-factors and time-access restrictions have on the last-mile deliveries?

Supervisory Team

Albert Wagelmans
Professor of Econometrics (Management Science)
  • Promotor