Online retailers have an important decision to make: when a customer places an order for several days ahead, do I commit my current inventory to the customer, or do I wait with committing, since another customer may arrive that wants the product immediately.
Together with my co-authors Willem van Jaarsveld of Eindhoven University of Technology, and Goncalo Figueira and Pedro Amorim of the University of Porto, we have developed a novel allocation mechanism that rewards customers that place orders early, while still properly managing inventory. This is a very challenging problem, since a retailer faces both customers that order early and expect reliability, and customers that expect immediate delivery. The strategic insight resulting from the application of our mathematical model is that it generally pays off to commit inventory early.
Online retailers struggle with fullfilment costs and an increasing diversity in consumer requirements
Online retail is one of the fastest growing industries despite the thin, often negative, margins. At the end of 2019, the e-commerce share of total retail sales in the US reached 11.2%. In terms of margins, however, even large online retailers are struggling. Amazon has averaged just 1.8% in operating margins over the past three years, while online grocery retailers have netted as low as 0.5%. These thin margins result from high fulfillment costs, which are triggered by the more complex and fragmented operations in the online retail environment. Fulfillment in online retail includes picking, packing and shipping to individual customers, the latter being particularly costly (e.g., for Amazon shipping represents 13% of net sales).
In an attempt to counter these costs, online retailers strategically offer alternative delivery options — such as different leadtimes — to allow for consolidation in picking and transport. In grocery retail, different time slots with different lengths and different horizons are offered for attended home delivery. The latter does not only increase options for consolidation, but also helps consumers plan in more detail when they plan to be home such that they are able to receive the groceries they ordered. Apart from many grocery retailers having deployed such strategies over many years, increasingly other online retailers are offering such detailed advance delivery options.
Backordering also becomes a more common strategy in online retail
However, service levels in online grocery retail are low and stockouts are common. It is increasingly common for retailers to backorder missing items. For instance, Target, one of the largest online retailers in the US, backorders items that are missing in a basket, explaining to their customers: ‘Backordered’ means there’s a delay on your order because an item is temporarily out of stock. The item will ship as soon as it’s in stock. Electronics retailer Best Buy offers similar service. Obviously, such backorders are costly, since they involve additional, potentially separate shipments to consumers. Offering backorders as an option also complicates the service level experience by the customer: apart from the fill-rate at the moment of ordering, also the delay experienced in getting the backorder delivered becomes an important metric. Hence, online retailers now have a larger portfolio of managerial decisions to make in their service strategy: inducing customers to place orders early provides additional information and allows for order consolidation, while at the same time the option to have products backordered adds an additional dimension to the service level experience.
Information from customers ordering early can be leveraged
Despite the additional complexity, these settings provide an opportunity for retailers to leverage additional available information and improve their operations. A key piece of information in this setting results from the fact that customers place orders for a future due date. While in some cases it is the vendor that determines the corresponding leadtime, in more and more cases it is the customer who specifies the date and time he or she wants to get the goods delivered. Actually, having the ability to select a convenient delivery window has been of great importance to online consumers for a while now. In online grocery retail, attended home delivery – where consumers need to indicate a preferred time slot in advance – it is actually the norm in the industry.
Irrespective of who determines the delivery time, there is a time window that is larger than the shipping time, and which results in `advance demand information’ (ADI) that the retailer can use to better manage its inventory, and hence increase operating margins.
Managing inventory, allocation scarce inventory, and nudging the consumer to order early are difficult to balance
In current inventory models with ADI, the commitment to the consumer is made upon delivery (CUD). This is actually at the moment when most information is available and the best decision can be made in terms of minimizing delay. However, the growing reality in online retail is that customers pick their preferred leadtime and expect a higher service level when they place their order in advance. This implies that in addition to using ADI for inventory control, this information should also be used for inventory allocation. This allocation decision may need to be made when the customer places the order rather than when the order is due for delivery. Obviously, this fundamentally changes the inventory logic and the associated strategic choices. Current ADI inventory models lack this option to allocate early and hence do not take advantage of this new reality in online retail.
The service level of an online retailer can be measured in terms of the fill-rate of the orders placed and the delay of the orders that are backlogged. Any policy needs to be evaluated against these performance measures. Strategically, however, another important property of any policy is important, and this is whether a policy is advancement-inducing. An advancement-inducing policy is a policy that rewards consumers that place orders early by committing early to such consumers and by providing them with a better service than customer ordering late. Such commitment needs to be firm, and thus prevent the item to be reallocated to a later order. Note that the inducing of advancement may go at the expense of the overall delay.
Committing to an order upon delivery (CUD) hence may not be the best choice for the allocation decision in online commerce, since it is not advancement inducing. In fact, consumers that would place orders early would not be able to get a firm commitment at the moment that they place the order. For such consumers, it would be rational to actually wait with placing their order until they are close to the moment of delivery. Such delay will lead to considerably less opportunities for the retailer to realize efficiency gains in the operation.
Instead, we propose it is better to commit upon arrival of the order, which is advancement-inducing. In our study, we define such an allocation policy and evaluate it both analytically and numerically. Our numerical evaluation is based on actual data of a European grocery retailer.
Committing upon arrival needs to be done with great care
Our analysis demonstrates that when committing upon arrival of an order (CUA), it is not a good idea to allocate units of the supply pipeline following First-In-First-Out. This is too naive and does not make effective use of the time between the arrival of an order and its due date. Its service level is worse than CUD. Therefore, we propose a second rule, which allocates the ‘least-critical unit’ (LCU), i.e., the unit in the supply pipeline that will arrive latest, but still in time to fulfill the order. The CUA-LCU policy dominates CUA-FIFO in every indicator. When compared to CUD, it results in longer delays, but higher fill-rate. This is a relevant trade-off in the online setting. To study this trade-off analytically, we define and study sensible policies, i.e. policies that satisfy an intuitive constraint on back-order clearance that is reasonable in environments where customers are all equally important. The corresponding broad class of policies contains CUD, CUA-LCU, and CUA-FIFO as well as a wide range of other policies. A particularly interesting analytical finding of our work is then that the sensible policy that maximizes the fill rate is precisely the CUA-LCU policy, while the classical CUD policy is the sensible policy that minimizes the average waiting time.
Managing online inventory requires a strategic choice
With the changing landscape in online retail, customers are increasingly placing baskets of orders that they would like to receive at a planned and confirmed moment in time. Especially in grocery this has been growing fast. This fundamentally changes the strategic management of inventory, not only because the customers may have different expectations, but also because the willingness of customers to order early provides an opportunity for cost reduction and service level improvement for thin-margin online retailers. We demonstrate in this paper that online retailers have the tools at hand to do so: they should commit early to customer orders to enhance the customer service experience, and — eventually — to also create opportunities for reducing the cost of operations. Our application in a grocery retail setting demonstrates that the effect size of our novel strategy is substantial and hence would need to be considered by many retailers.
While we have conducted our research with the utmost care, the work is currently undergoing peer review. Until then, we refer to our working paper that is fully available online at SSRN. Any feedback is appreciated. We are interested in exploring further applications of our model, so if you are intrigued and interested, and are willing to share data, please get in touch.
Jan C. Fransoo is a professor of Operations Management and Logistics and Dean of Research at Kuehne Logistics University in Hamburg, Germany. Has published over 130 papers in academic journals and books on topics ranging from production planning and supply chain management to inventory management and transportation optimization.
Photo: Amazon