Inventory and distribution strategies for retail/e-tail organizations

https://doi.org/10.1016/j.cie.2009.09.005Get rights and content

Abstract

In the retail sector many traditional bricks-and-mortar companies have added online sales channels to their supply chains. Unfortunately, even though the combined retailer/e-tailer is becoming a common business model, there is very limited research addressing retail/e-tail operations. To address this gap, this research considers where and how much inventory should be allocated and held at each site for a company that satisfies both in-store and online demand. Specifically, we determine how many and which of a firm’s capacitated locations should handle online sales to minimize total cost (holding, backorder, fixed operating, transportation, and handling costs). Our primary findings include the following: (i) when all costs are considered the percentage of sales occurring online plays a critical role in determining the number of sites providing e-fulfillment; (ii) when holding and backorder costs are the only consideration (i.e., the customer pays for shipping), the standard deviation of in-store demand controls where online inventory should be located, regardless of the percentage of demand occurring online; and (iii) an increase in unit shipping costs does not necessarily imply that adding online fulfillment locations will decrease total cost. Results from a computational study illustrate that the model provides good solutions even when demand is correlated or not normally distributed.

Introduction

The advent of e-commerce has made retailing more complicated and more competitive. Today consumers can use the Internet to sidestep their corner store and patronize shops across the country or around the world. It is within this business context that many traditional bricks-and-mortar companies have attempted to increase sales and improve profitability by adding online retail channels for consumers. Unfortunately, even though the combined retailer/e-tailer appears to be emerging as the dominant business model for the retail sector, there is little research specifically addressing retail/e-tail operations and how the addition of an online sales channel should affect a firm’s supply chain network design (Hill et al., 2002). This paper contributes to the literature by filling this gap and examining strategies that retailer/e-tailers can employ to leverage synergies between their online channels and “bricks” locations.

In this paper we present an online fulfillment assignment problem for a two-echelon distribution system to determine where and how much inventory is held for retail/e-tail operations. Our model accounts for potential channel synergies in inventory control, location, and distribution decisions that can influence how a retailer/e-tailer handles its online fulfillment. For example, online grocers like Albertsons and Safeway have considered tradeoffs amongst inventory pooling, overhead cost, and proximity of store locations to online customers before deciding to pick and pack Internet orders at some of their existing stores (Boyer and Hult, 2006, Grocer, 2003, Montgomery, 2003, Retail World, 2003). Numerous other retailers face similar considerations, such as online orders at Lowe’s being assigned to the customer’s “local Lowe’s store” (www.lowes.com). To account for such synergies our model seeks to identify those “bricks” sites that can fulfill online customer demands at the minimum expected cost of holding inventories, incurring backorders, transporting products, handling online sales, and maintaining e-fulfillment capabilities.

To solve the resulting optimization problem, an algorithm is developed to determine the online fulfillment locations for our online assignment problem. Although many researchers have solved general assignment and location problems (Erlenkotter, 1978, Gavett and Plyter, 1966) there has been very little work addressing dual sales channel environments with stochastic demands, as discussed in the next section of the paper.

Our analysis focuses on a two-echelon distribution system operating under periodic review. The system consists of a break-bulk central warehouse and J + 1 retail/e-tail locations. These lower echelon members could be retail stores or regional warehouse facilities and may be dedicated to online fulfillment or capable of handling both in-store (retail) and online (e-tail) demands. Within this context our solution method is used to identify the expected minimal cost for assigning “bricks” facilities to handle Internet sales. Although many other alternate retail/e-tail strategies exist, to keep the scope of the paper reasonable, we will focus on the setting described above. Our study provides various insights for retail/e-tail organizations. First, when holding and backorder costs are the only consideration (i.e., the customer pays for shipping), the standard deviation of in-store demand controls where online inventory should be located, regardless of the percentage of demand occurring online. Obviously, inventory cost is just one component of total cost and may not lead to a selection of online fulfillment locations that minimizes total cost for a firm. When all costs (holding, backorder, fixed operating, transportation, and handling costs) are considered the percentage of sales occurring online plays a critical role in determining the number of sites providing e-fulfillment. Results from a computational study also illustrate how an increase in unit shipping costs does not necessarily imply that adding online fulfillment locations will decrease total cost.

The remainder of this paper is organized as follows. The next section presents a brief overview of relevant literature. Section 3 formulates the dual channel assignment problem to minimize per period combined holding, backorder, transportation, handling, and fixed operating cost. Section 4 develops a set of results which specify how inventories should be placed to minimize holding plus backorder cost and uses these results to construct valid lower and upper bounds on total cost. The results are then applied within an efficient branch-and-bound algorithm to solve the nonlinear integer optimization problem. Section 5 reports on a numerical study to evaluate the effectiveness of the solution method and address the following questions: (1) What impact does percentage of online demand have on supply chain network design? (2) How does demand variability impact the supply chain network design? and (3) What impact do pooling, operational, transportation and handling costs have on optimal online inventory allocation policies? Section 6 evaluates the impact of demand correlation and non-normal demand distributions. Section 7 presents an extension of the base model. In the final section we present some concluding remarks and directions for future research.

Section snippets

Literature review

In contrast to the substantial quantity of multi-channel research in the marketing and economics areas, research on multi-sales channel inventory systems is relatively sparse. Agatz and Fleischmann (2008) review the current state of research in e-fulfillment and multi-channel distribution and conclude that there is room for significant contributions in all areas of e-fulfillment, particularly those examining the interaction between e-fulfillment and other distribution channels. Related reviews

The online fulfillment model

To obtain a robust model we begin by formulating the general, two-echelon facilities network shown in Fig. 1. In this model, the central warehouse is permitted to hold inventory at an offsite depot for online sales but not for in-store sales. (Note that the model can easily handle settings with more than one online depot.) Each period we assume that the following sequence of events takes place. First, the central warehouse observes the inventory positions at the sites and allocates orders to

Solving the online fulfillment model

In this section we propose a branch-and-bound algorithm to solve the nonlinear integer problem (P). First, we construct valid lower and upper bounds on the system cost. Using these bounds a branch-and-bound algorithm is developed to solve problem (P).

Computational study

To illustrate how the model and solution methodology can be used for designing retail/e-tail systems and to gain insights into cost tradeoffs that must be considered in such a design, we develop a set of test problems based on a 6-location kitchenware retailer/e-tailer and a 14-store sporting goods company looking to add an online sales channel. In both examples we assume that each period the central warehouse allocates inventory to its lower echelon locations (in a process that takes one week).

Correlated demand and non-normal demand distributions

A computational study was undertaken to determine if problem (P) provides good solutions even when demand is correlated and non-normal. The demand distributions examined include positively correlated normal distribution, negatively correlated normal distribution, uncorrelated triangle distribution, and uncorrelated uniform distribution.

To assess the impact of correlation and alternate demand distributions, the general 2-echelon system in Fig. 1 was modeled in a computer simulation. In each

Satisfying online demand from one or more locations

In this section we describe how the solution methodology presented in Section 4 can be extended to permit fractional eij decision variables. That is, we relax the requirement that all of a region’s online demand must be satisfied from only one location. The resulting formulation, (P1), is identical to (P) except that: (i) constraint (7) is replaced by constraint (7′) below0eij1i=0,1,J;j=0,1,Jand (ii) constraint (4) is removed from the formulation (to permit other online fulfillment

Conclusions

The Internet has already changed the way that companies around the globe conduct business. In fact, for many businesses online presence has become an order qualifier over the last decade. Often this has resulted in the haphazard addition of a new channel to an existing supply chain.

This paper provides a method of assessing where and how much online inventory is held for retail/e-tail operations such that overall system costs are minimized. Our theoretical results provide the basis for computing

References (26)

  • D. Erlenkotter

    A dual-based procedure for uncapacitated facility location

    Operations Research

    (1978)
  • N. Erkip et al.

    Optimal centralized ordering policies in multi-echelon inventory systems with correlated demands

    Management Science

    (1990)
  • A. Federgruen et al.

    Allocation policies and cost approximations for multilocation inventory systems

    Naval Research Logistics Quarterly

    (1984)
  • Cited by (0)

    This manuscript was processed by Area Editor William G. Ferrell Jr.

    1

    Tel.: +1 812 855 3487; fax: +1 812 856 5222.

    2

    Tel.: +1 812 855 3491; fax: +1 812 856 5222.

    View full text