RE-ENGINEERING A REVERSE SUPPLY CHAIN FOR A DIRECT RETAILERS PRODUCT RETURNS SERVICES

Carol C. Bienstock, Ph.D. * Assistant Professor Department of Management and Marketing Radford University, USA M. Mehdi Amini, Ph.D. Professor Department of Marketing and Supply Chain Management Associate Director of FedEx Center for Supply Chain Management The University of Memphis, USA Donna Retzlaff-Roberts, Ph.D. Professor Department of Management The University of South Alabama, USA ABSTRACT An important service management activity, particularly in a retail environment, is return services. This article discusses the strategic issues surrounding the effective management of product return services and the importance of the role of effective reverse logistics operations to the design and execution of successful and profitable reverse supply chains to support product return activities. We present a case study to illustrate how a reverse supply chain and the logistics activities that support it were reengineered to enhance the effectiveness and profitability of the product returns process for a major direct retailer in the US. Keywords: simulation; retail; product returns; supply chain management; reverse logistics
*Corresponding Author Carol C. Bienstock, Ph.D. Radford, VA 24142, USA Tel: 540.831.5301 Fax: 540.831.6261 cbienstoc@radford.edu

Acknowledgements: This research was supported by a research grant from the FedEx Center for Supply Chain Management, at the FedEx Institute of Technology, The University of Memphis 1

RE-ENGINEERING A REVERSE SUPPLY CHAIN FOR A DIRECT RETAILERS PRODUCT RETURNS SERVICES
INTRODUCTION How do companies differentiate themselves when operating in industries where most, if not all firms offer high quality products and customer service at the time of sale? As James Stock put it, After a while, those features just become your admission to the game (Lambert and Stock, 1993, p. 28). A potential solution to this dilemma is offered by Dennis and Kambil (2003), using what they term service management, which provides both competitive differentiation and an opportunity to increase profits. Service management is the sum of all customer interactions that follow a products sale . . . (Dennis and Kambil, 2003, p. 42). The benefits of service management can also be related to the service profit chain framework, which integrates investments in service operations with customer loyalty and firm profitability (Heskett, Jones, Loveman and Sasser, 1994; Wagner, Mittal and Mazzon, 2002). One of the most important of these service management activities, particularly in a retail environment, is return services. In an effort to enhance service management activities and, thus, engage in what Flack and Evans (2001, p. 19) term marketing on customer terms to increasingly demanding customers, a growing number of retailers are liberalizing return policies and becoming more reliant on consignment inventory, activities which can result in a greater number of returned products. Because catalogue and online retailers typically face higher rates of return than

traditional retailers, effective service management of their product returns is even more important (Daugherty, Autry and Ellinger, 2001). Not surprisingly, the existence, effectiveness, and efficiency of service management activities, such as return services, depend heavily on effective reverse logistics operations. Because reverse logistics operations and the supply chains they support are significantly more complex than traditional manufacturing supply chains (Dennis and Kambil, 2003) an organization that succeeds in meeting the challenges presents a formidable advantage that is not easily duplicated by its competitors. Effective reverse logistics operations benefit both the organization and its customers. Successfully accomplished service management activities, such as product return operations, positively impact customers satisfaction and, consequently, customer loyalty and return sales (Cohen and Whang, 1997; Fitzsimmons and Fitzsimmons, 1998; Retzlaff-Roberts, 1998; Daugherty, Autry and Ellinger, 2001). In the next section, we discuss the issues surrounding the value of product returns as service management activities. We also discuss the importance of the role of effective reverse logistics operations to the design and execution of successful and profitable reverse supply chains to support product return activities. In the last section we present a case study to illustrate how a reverse supply chain and the logistics activities that support it can be reengineered so that the effectiveness and profitability of a direct retailers product returns process are enhanced. PRODUCT RETURNS SERVICES Product returns have been and remain an essential part of the retail landscape. Customers return products for a variety of reasons, e.g., they change their minds, the
3

product shipped to them is defective; the product is damaged in transit; the wrong quantity or the wrong product is shipped. Customers also return products that are

under warranty or products that are the subject of manufacturers recalls. Particularly in the case of direct retailers, e.g., catalogue and online retailers, where customers generally perceive more risk associated with product purchases (Schoenbachler and Gordon, 2002), a solid record of product return services can significantly enhance customer loyalty and increase the probability of repeat purchases (Daugherty, Autry and Ellinger, 2001). While the return of a particular item is generally not expected at the time of sale, many organizations have some means of forecasting what percent of their sales volume is typically returned. The magnitude of this percentage depends upon the nature of the business and the organizations return policy, and can vary from as low as 2% to as much as 50% (Lambert and Stock, 1993). Generous return policies have made the structuring of the required reverse supply chains and the management of the reverse logistics that support these unplanned returns particularly difficult because organizations do not know what products will be arriving when (Meyer, 1999). REVERSE LOGISTICS Reverse logistics accounts for 5-6% of total logistics costs in both the manufacturing and retail sectors. One of the more interesting and significant trends in supply chain management is the recognition of the strategic importance of reverse logistics operations (Retzlaff-Roberts and Frolick, 1997; Handfield and Nichols, 1999; Daugherty, Autrey and Ellinger, 2001). These reverse logistics operations support a
4

variety of activities ranging from what is termed green logistics, i.e., efforts to reduce the environmental impact of the supply chain (Rogers and Tibben-Lembke, 2001, p. 130), to activities that encompass product returns, repairs, and refurbishment. Estimates of the costs of reverse logistics operations range from $37 - $921 billion annually. Despite this, four in ten logistics managers consider reverse logistics operations to be a very low priority for their companies. Obviously, the type and extent of reverse logistics activities vary according to industry, but the extent of these activities are already significant in many industries and they continue to grow (Rogers and Tibben-Lembke, 2001). Although recognition of the strategic importance of reverse logistics operations is not by any means universal, but there is some evidence that this is changing. According to Meyer (1999), the . . . new frontier of management is reverse logistics . . . after companies have downsized, reengineered, TQMed, racheted up customer service, and wrung out every conceivable cost efficiency, it may well be one of the last business frontiers business can conquer (p. 27) Most logistics systems are not well-equipped to manage product movement in a reverse direction. In addition, the costs associated with reverse logistics may be nine times higher than moving the same product in a forward channel. Another complicating factor is that returned products that are handled by reverse logistics operations often cannot be transported, stored and/or handled in the same manner as when they are distributed in a traditional supply chain (Lambert and Stock, 1993). Since the activities involved tend to be so varied, reverse logistics operations are quite

complex to manage, In addition, demand can be difficult to predict, making product and information flows challenging to manage. Complicating the problem of managing reverse logistics operations is the fact that very few, if any, standardized software solutions designed for reverse logistics operations exist (Meyer, 1999; Rogers and Tibben-Lembke, 2001). Although reverse logistics operations in general can be quite difficult to manage, there are some particular challenges to managing reverse logistics operations for product returns. Not only does a retailer have to effectively manage the actual product return, which is a challenge in itself, as discussed above, but, once returned, the product must be disposed of in some way. The most common disposal method for returned product is return to the manufacturer, but some returned products are repackaged and resold, resold as is, destroyed, or sold at other retail outlets (e.g., off price retailers or manufacturers outlets) (Daugherty, Autry and Ellinger, 2001). Despite the fact that effectively managing the complex reverse logistics operations required to support what Dennis and Kambil (2001, p. 42) term service to profit supply chains, requires considerable skill and integration, Dennis and Kambil stress the potential advantageous competitive positioning and market opportunities for firms that handle these important activities effectively. Dennis and Kambil also point out the value of using reverse logistics activities to develop service-centric supply chains to adequately support customers in such activities as product returns. Such supply chains are vital tools as companies seek to differentiate themselves from their competitors, increase customer loyalty, and boost profit margins. For this reason, firms that can effectively implement and manage the necessary reverse
6

logistics operations to meet these needs will significantly enhance their competitive position. With this background discussion of the strategic role of managing product returns service management activities and the role of reverse logistics operations in supporting these service-to-profit supply chain operations, we present a case study of a project that reengineered reverse logistics operations for a major direct retailer in the US. The project was designed to assist the company as it considered an

innovative approach to create a more convenient product return process and reduce the cycle time for customers to receive refunds and exchanges when items are returned. The primary objective of the project was to enhance customer service quality by reengineering the retailers reverse logistics processes in order to reduce the cycle time of providing refunds and exchanges to customers. A secondary objective was to enhance the internal efficiency of processing returned products by exploring opportunities for lowering product returns and related operational and capital costs. In order to accomplish the reengineering effort, computer simulation models were developed and examined to compare the current process with a proposed new reverse logistics process under different operational scenarios. CASE STUDY OF A DIRECT RETAILERS PRODUCT RETURNS PROCESS The case study presented here involves a direct retailer located in the US. This retailer markets apparel and household goods with sales in excess of one billion US dollars annually. The forward supply chain and the attendant logistics processes are efficient and effective with most customer orders being shipped within 24 hours.
7

During the holiday season well over 100,000 packages may be shipped daily. Customer satisfaction is a high priority. In the spirit of continuous improvement, the retailers management was eager to explore new opportunities to enhance customer satisfaction with the product returns process. In general, increasing customer satisfaction with the product returns process means (1) reducing the cycle time of customer receipt of the refund or exchange, and (2) increasing the convenience of sending a return. For customers of direct retailers, one of the disadvantages of transactions is the inconvenience and time involved in returning an item. Many direct retailers try to mitigate this inconvenience by providing a return form and a preaddressed shipping label with each order. Nevertheless, the typical return process for customers of direct retailers is typically something like this: 1. Fill out the return form or write a letter to the retailer to indicate the reason for the return and the requested action, e.g., exchange for another item, issue a refund check, or credit a card credit account. 2. Repackage the item and enclose the paperwork, 3. Go to the post office and stand in line to have the package weighed and postage assessed. 4. Wait for the package to be received and processed, and the requested action to be completed by the retailer. All of this is time consuming and inconvenient for the customer. The

inconvenience of the return process is often cited by customers of direct retailers as a major deterrent to initiating retail transactions (Cho, Im and Hiltz, 2003). Finding
8

ways to reduce the inconvenience and cycle time of a customer return can increase customer satisfaction, thus enhancing customer retention and sales. The Direct Retailers Current Product Returns Process The direct retailers current product returns process begins when a customer decides to return one or more products. The majority of the time these are new products which the customer has recently ordered and received. Reasons for product return are numerous, e.g., the customer may have changed his/her mind, the customer may decide that they want a different color or size, In addition to return of new products customers also return used products that they feel did not live up to their expectations. Regardless of whether the product is old or new the customer will request either an exchange or a refund. Figure 1 depicts a simplified version of the current reverse logistics process map. This process includes a number of main processes and a large number of subprocesses to effectively manage arrival of a large volume of packages containing items from a variety of product lines. These packages need to be sorted out and routed to the correct locations. This can be a difficult task, since the only indication of what the package contains is the size and shape of the box. Each package is processed by opening it, reading the contained documentation, and assessing the package contents. This is the point at which the customer transaction is separated from the merchandise and the two processes proceed independently and in parallel. The returns documentation is transferred to the financial transaction process, where depending on the initial means of transaction, e.g., credit card, personal check, gift certificate, customers are reimbursed for the returned merchandise.
9

If an

exchange has been requested, the appropriate information proceeds to the distribution center, from which the exchange item is shipped. This completes the customer transaction process. Meanwhile the returned products have been removed from the package, are sorted into the various product groups, and are conveyed to the merchandise preparation area. Here the quality each item is assessed and the item is prepared as needed for its destination. First-quality items are repackaged for return to the

distribution center. Lower quality items go to a variety of destinations depending on their condition. For example, some items are donated to charity, while the lowest level of quality is discarded. Items are consolidated and shipped to the appropriate destination. Notice that the customers financial transaction waits to commence until the package has been received, opened, and its contents assessed. Only at this point can the information needed for the customer transaction be separated from the merchandise. This is the usual procedure in virtually all return processes; the The

merchandise must be in hand before any further transaction takes place.

majority of the cycle time for the product returns process is due to shipping time through the reverse supply chain. The Direct Retailers Proposed Product Returns Process The proposed reengineering of the product returns supply chain for this direct retailer hinges on the fact that the shipping time is removed from the customer transaction by having customers call first and use a scanable postage-paid label. As shown in Figure 2, the reengineered process proposes that customers telephone the
10

direct retailer to indicate what they are returning and specify the details of the desired exchange or refund. A postage paid return label is provided with the initial order, thus providing a significant convenience to the customer. The customer is charged a nominal fee for this postage paid return label. When the carrier (e.g., FedEx, UPS, USPS) receives the package containing the returned product, the label is scanned and the information transmitted to the direct retailer. This allows the documentation containing information about the product return to be separated from the merchandise at a much earlier point in time, so that the customers return transaction can be completed without the delay of waiting for the product return package to arrive. When the package arrives at the returns center all that remains is to reconcile the transaction and complete the merchandise preparation. Since the scanable return label included in the initial order allows information on the returned product to be transmitted prior to the retailer actually receiving the package containing the returned product, the returns center can be restructured to combine package processing and merchandise preparation operations based on product lines. In the current product returns process the contents are unknown until the package is opened, making it impossible to process returned products by product line. This is the reason for having package processing and merchandise preparation as separate operations in the current process. Being able to route packages to the right location based on product lines offers the potential for increasing the efficiency of operating the returns center.

11

However, this streamlined process can be followed only if the customer calls and uses the scanable label. If the customer does not call or does not use the label, then the customer transaction cannot be completed prior to package arrival and the package must follow traditional product returns processing, with its separate package processing and merchandise preparation operations. Comparison of Current and Proposed Product Returns Processes In order to evaluate the proposed product returns process described above it must be compared to the current process. The cycle times and other characteristics of the current process are known, but the proposed process is very much a what-if? scenario. process: 1) How long would it take for a customer returning product to receive their desired exchange product or credit for the returned merchandise (what is the customer cycle time)? 2) How long would it take for returned products to be prepared for resale or disposal (what is the product cycle time)? 3) How many FTE (full time equivalent) employees would be needed for the proposed process? The customer cycle time (CCT) is defined as the time from when a customer ships a package until receipt of the refund or exchange. The product cycle time (PCT) measures the time from when a customer returns an item until it is shipped out from the returns center for resale or disposal. Under the proposed product returns process, the CCT time would clearly be decreased, which was the major motivation for the
12

Answers are needed to the following questions regarding the proposed

proposed change.

The PCT was expected to remain approximately the same

(reducing this time was not an objective) and was measured simply as a characteristic of the process. The required number of FTEs for staffing the proposed process was unknown because many of the tasks are restructured by the reengineering the returns center in the proposed process (i.e., combining the package processing and merchandise preparation processes). Fewer FTEs should be needed to staff the

returns center because the use of scanable labels and customer calls should allow packages to be sorted and processed very efficiently. However, the proposed product returns process created a new job that did not previously exist personnel to answer the phone calls for returning merchandise. Computer Simulation Modeling Due to the complexity of the reverse logistics activities for product returns, the answers to the questions above were not easily determined. There appear to be two possible methods of assessing which process the retailer should adopt. One method was to adopt the proposed process, collect data, and evaluate in hindsight whether it was great idea or a mistake. The second method was to create a computer simulation model of the current and proposed processes to allow the organization to perform analyses that would enable it to compare the current and proposed processes, as well as to fine tune the proposed process and make an informed decision on adoption. Computer simulation modeling is known as an effective approach for process reengineering, particularly when the level of complexity is high. It allows for accurate and effective study of alternative operational scenarios without costly and timeconsuming interruption of the real physical process. Also, simulation models are
13

capable of capturing the probabilistic nature of the processes under study, where simpler analytical methods fail. When using simulation to compare a proposed process with an existing one, it is advisable to first model the current process to allow validation against reality. After working out any bugs the current model can be modified for any number of what-if scenarios to evaluate proposed changes (Law and Kelton, 1999; Rabinovich and Evers, 2003). Therefore, the reengineering of this direct retailers product return operations involved the simulation of both the current product returns operation, as well as the proposed product returns operation. The complexity of the reverse logistics activities for this direct retailers product return process is driven by the probabilistic nature of the activities, events, and man-machine interactions within the different sub-processes. Figure 3 shows major steps involved in a computer simulation modeling and analysis (Law and Kelton, 1999). Using this framework, the discussion below details the computer simulation modeling and analysis process involved in the re-engineering efforts for the direct retailer in this case study. Step 1 begins with a clear objective and identification of what questions are to be answered. The objective of the current study was to reduce cycle times (CCT and PCT) and operational costs (in the form of FTE employees in the returns processing center). In order to accomplish this objective, it was necessary to evaluate the proposed process by comparing it to the current one, since measuring and benchmarking the related cycle times (CCT and PCT) and required FTE requirements under different operational scenarios was required to enable the direct retailer to
14

determine which of the two product return processes accomplished the objectives of the reengineering. In Step 2, process maps were prepared for the current and proposed processes. These maps were designed to provide a clear view of the processes, facilitate communication between the research team and the practitioners, and clarify the types of data that would be required to construct the simulation model. This step was one of the most time-consuming phases of the project, taking 70% to 80% of the project duration. However, these process maps were vital, since they formed the basis for the computer simulation model. Operations within the product Returns Processing Center (RPC) of this direct retailer include 15 major processes. Each of these 15 major processes themselves consists of a network of sub-processes. During the reengineering effort, development of a detailed process map of the RPC consumed approximately 60% of the total time dedicated to the project. In addition to the fact that the authors signed a legally binding confidentiality agreement with the direct retailer, the sheer size of the current and alternative process maps and prohibit complete representations of readable versions on standard sized paper. For example, the smallest readable versions of the process maps require 25 by 17 sized paper. The actual process maps, which guided the simulation models for the current and alternative processes, included tracking of returned packages and items from each process to the next. In addition, the related financial papers were tracked until a returned item was either disposed of, or prepared for resale and reshelved.

15

During the time the required data was being collected, for Step 3, Arena 3.01 (1997) software was utilized to develop the simulation model for the current product return process (Step 4). Step 3 involved the collection and analysis of the data for the simulation model. Corresponding to the large size of the processes to be modeled, the data requirements were quite large. Table 1 shows the number of data elements involved. Since returning product and its associated paperwork are separated shortly after entering the system, and passed though different processes, the values for product and paperwork are shown separately in Table 1. A total of 85 separate processing time distributions were needed. Although the delay time distributions in Table 1 do not refer to entities waiting in a processor queue, distributions for delay times were needed because returned items were batched at many points in the system after being processed, causing a delay before being sent on to the next processing step. Thus data was needed on these delay time distributions, of which there were 120. The term splits in Table 1 refers to decision points in the return process where some items went one way and some items another way. Proportions were needed for each of these, with the total being 113. Given the large size of the model, an explanation of how each distribution was fitted would be impractical. The process used to collect the required data and fit distributions for the model was two-fold. First, a data set was utilized which consisted of 445 returned items that had been time stamped at various points in the process was obtained from the direct retailer. Using these data, distributions were fitted using Arenas distribution fitting capability. The fitted distributions obtained from this
16

pricess included the beta distribution, normal distribution, and exponential distribution. A second process was used to fit distributions for the remaining

processes in Table 1, which would have required data that had time stamps at both the beginning and the end of each process. Since this level of data collection was not possible, in these cases, personnel who worked in these process areas (both managers and operators) were interviewed to obtain their answers to typical, minimum, and maximum times for these processes and the triangular distribution was used. As the data for these various returns processes became available, we used Arenas Input Analyzer capability and Fit All option to identify the best distribution fitted to the collected input data. The best was defined as the

distribution with minimum square error, as determined by the p-value of the Chisquare statistic for goodness of fit using a Kolmogorov-Smirnov analysis. Steps 4 and 5 involved the development, verification, and validation of the Arena simulation model. We created and verified (with management of the direct retailer) a detailed process map of the current product returns process (Figure 1). Using this map, we developed an Arena simulation model to depict the current return process. For the purpose of ease of communication with management, as well as verification and validation of the simulation model, the format of the Arena model mimicked the layout of the process map. As each major process and its related sub-processes were populated with the identified best distributions, the logic involved in simulating each process was validated. In addition, validation runs were conducted for each major process in conjunction with the other major processes with which each process was interlinked.
17

During this exercise, we applied Arenas animation capabilities to the extent allowed by hardware memory limitations. The validation process was completed when the network of all fifteen major processes along with their related sub-processes were simulated simultaneously. For verification and validation purposes, simulation results for all major processes and their related sub-processes, were communicated and discussed with the project team, including representatives from direct retailers management. . In addition, results from the model of the current process were successfully validated against existing operational data. As discussed earlier with respect to the process maps, confidentiality concerns, as well as the sheer scope of the product returns process prohibit a detailed representation of the entire Arena simulation model for either the present or the proposed product returns processes. For example, because of the scope of the simulation model, a readable representation of the Arena simulation of the direct retailers present product returns process requires 66 x 60 paper. However, in order to provide the reader with an idea of the simulation model developed for the present product returns process, we present, in Figure 4, a small section of the Arena simulation, depicting the sub-processes within the current apparel and footwear returns processes. As Figure 4 indicates, when a return package in a previous return process has been identified to include apparel or footwear return items, the package enters the apparel and footwear returns process. If any of the returned items require repair, these are added to a batch. When this repair batch attains a certain size, it is conveyed to the repair process. If the apparel and footwear items do not require

18

repair, they are processed, batched, and conveyed to the next sub-process within the major apparel and footwear returns process. After data collection was completed (step 3) and the simulation model of the present product returns process was developed, verified and validated (steps 4 and 5), full model experimentation and scenario analyses were conducted on the model of the current product returns process (steps 6 and 7). Cycle times were measured under alternative operational scenarios that characterized the direct retailers product return operations under different product return levels. The return process center operates five days, two shifts per day. Hence, we decided a steady-state simulation approach should work very well. To determine the simulation run, we ran a five-replication simulation of the entire returns process, with a normal volume of packages, for a period of one, two, three, four, and five months. Analyses of the collected cycle time statistics and graphs generated from these statistics, which depicted changes in the cycle times for the five replications, indicated no significance differences between two, three, four or five month simulation periods. Hence, we decided to use a two-month simulation period. As a result of the steady-state analyses described in the previous paragraph, we realized that system steady-state is achieved within the first five days of simulation. Thus, for the comparative study of the current and the proposed product returns processes, all simulation runs used a two-month simulation period with a fiveday warm-up time. In addition, each simulation run assumed that system and related resources are idle.

19

To develop a basic understanding about the performance of the current product returns process, management of the direct retailer was interested in an exercise that included three simulation scenarios. The only difference between the three scenarios was the daily volume of packages arriving at the return facility. These three package volumes were, according to management, the three typical volumes historically processed at the center. These volumes represented a spectrum of a low volume, the typically expected volume, and a high volume of package returns. Assuming that A depicts the base, or low volume scenario, the volumes for scenarios B and C relative to A increase by 43% and 114%, respectively. At

managements request, the focus of the simulation of the three package return volumes was on the differences among the average cycle times associated with customer reimbursement for four different customer purchase methods (1, 2, 3, 4) and the cycle times associated with reshelving of six general categories of products (1, 2, 3, 4, 5, 6) returned to the returns facility.1 A summary the simulation results for the exercise described above is shown in Table 2. The table shows Relative Average Percentage (RAP) changes in the customer reimbursement cycle times when product return volumes for scenarios B and C are compared to the base scenario A. When the RAP of customer reimbursement cycle time associated with the four customer purchase methods under scenario B was compared to scenario A, the RAP increased by a fraction of a percentage for all four purchase methods. A comparison of the RAP for scenario C versus scenario A

The confidentiality agreement with the direct retailer prohibits us from providing specific details of the four customer purchase methods or the six product categories.

20

indicated an increase of between 4.81% to 10.45%, with customer purchase method 2 experiencing the largest increase, and customer purchase method 4 experiencing the smallest increase in the RAP. In addition, Table 2 shows the cycle times related to reshelving for each of the six categories of products returned to the returns processing facility. The top three product categories represent approximately 90% of the returned products. In

comparing scenario B versus scenario A, the reshelving cycle times show an increase from 1.64% to 5.34%, where the minimum and maximum increases occur for product categories 1 and 4, respectively. The reshelving cycle times for scenario C versus scenario A, reveal a 93.49% increase between these two scenarios for product category 5; a 72.42% increase for product category 1; and only a fraction of a percent increase for product category 6. As Table 2 demonstrates, the associated variances among product categories between scenarios C and A is much larger than for scenario B versus A. These results discussed above helped management to objectively understand how the typical returned package volumes within the current return process interacted with (a) the customer purchase method to influence the customer reimbursement cycle times and (b) the category of returned products to influence the returned product reshelving cycle times. Also, this simulation exercise allowed

management to develop a deeper understanding of the nonuniform nature of the impact of these factors on the cycle times. As a result of these simulations,

management realized that (a) they could use the information gleaned from the simulations to effectively manage customer expectations with regard to when to
21

expect reimbursements from product returns; and (b) they should take into consideration the information on product reshelving cycle times provided by the simulations to help manage demand for products that are in short supply. Using the thorough understanding of the current return process provided by the model experimentation and scenario analyses discussed above, a simulation model of the proposed new returns process was created. As discussed earlier, the key

difference between the current and proposed new returns processes is the percentage of customers expected to call the returns processing center prior to returning their products and provide detailed information about the products they anticipate returning. Analysis of the model of the proposed product returns process required a number of iterative scenarios in which bottle necks were identified and resolved. The throughput capacity of the product returns sub-processes were determined by the probability distribution that describes the time needed for task completion and the number of these tasks that could be performed in parallel. Many of the product returns tasks were performed by people since returns processing is labor intensive. Determining the number of FTEs needed for these various labor intensive tasks was essential for eliminating bottlenecks and identifying FTE staffing needs. In simulating the proposed new product returns process, the management of the direct retailer desired to base the model experimentation and scenario analyses on two different estimates of the percentage of customers who would call the returns center prior to their product returns. The first estimate was a conservative one and was believed to represent the percentage of customers who would call in advance of
22

returning products when the new return policy was initially being introduced. The second estimate, 10% higher than the first one, was believed to be the long-term percentage of customers who, once they became advised and further educated about the new return process, would call in advance of returning products. Using the two estimates of the percentage of customers who would call prior to returning products, simulation scenarios D and E were designed. In simulating both of these scenarios, we assumed the volumes of product return would be the same as in scenario B (described above in the analysis of the current product returns process), a two-month simulation period, and a five-day steady state period. The same level of resources and number of processors/process centers, were considered for both scenarios D and E. In addition, for these two scenarios, management wished to focus only the top three product categories, since, as discussed above, constitute approximately 90% of returned products. Comparison of Current and Proposed Product Return Operations Comparison of the current and proposed product returns processes involved an analysis of scenario D for the proposed product returns process with scenario B of the current product returns process. Table 3 depicts the relative average percentage (RAP) cycle times associated with customer reimbursement for four different customer purchase methods and the relative average percentage (RAP) cycle times for reshelving of the top three product categories. In comparing scenario D versus

scenario B, we can see that improvements in RAP customer reimbursement cycle times for the four customer purchase methods range from 19.91% to 35.39%. The customer reimbursement RAP cycle times for customer purchasing methods 1
23

through 4, when scenario E is compared to scenario B, vary from 24.75% to 44.52%. This is more impressive than the relative improvement realized by scenario D versus scenario B. Similar comparisons between scenarios D and B and scenarios E and B regarding the RAP in reshelving cycle times for the three top product categories demonstrated mixed results. There is an increase in RAP for reshelving cycle times of from 3.31% to 10.05% for product categories 1 and 2 (in other words, the reshelving cycle times worsened), but an improvement of 4.87% to 5.31% for product category 3 (i.e., the reshelving cycle times decreased). The simulation exercise with the new product returns process enabled management to understand two important issues. First, the new product returns process, regardless of the percentage of customers who called prior to returning products, significantly improved the cycle times associated with customer reimbursement. Secondly, the impact of the new product returns process on product reshelf cycle time for two product categories is negative (product categories 1 and 2) and for the third product category (product category 3) is positive. However, the difference in product reshelf cycle time between scenario D versus B and between scenario E versus B is only a fraction of a percentage, indicating that variation in how many customers call in advance of returning products has a minimal impact on the product reshelf cycle time. A follow up simulation exercise showed that adding one additional processor to a bottleneck sub-process would improve the RAP product reshelf cycle time 10% when compared to scenario B. Based on this follow up simulation, management was
24

convinced that the increase in cost for the additional processor resource was well justified when the significant benefit of reduced product reshelf cycle time was considered. Results showed that reengineering of the returns center according to the proposed product returns process would indeed improve efficiency and productivity and would require approximately 65% of the current FTE staffing level for returns processing of packages and merchandise in the returns center. However, since the proposed product returns process creates the additional task of answering phone calls for returns, the net staff FTE levels would be approximately 85% to 90% of current levels. Staff that handles the financial transactions associated with product returns remains essentially unchanged under the proposed process. Note that the degree of reduction in staff FTEs in the returns processing center under the proposed product returns process depends on the volume of customers who fully utilize the new process by calling and using the scanable label. The reduction to a net of 85% to 90% of current FTE staffing levels in the returns processing center is based on the assumption that 35% of customers returning products will use the new process. If the percentage of customers using the new process increases, the net FTE staffing requirements in the returns processing center would decrease; conversely, if fewer than 35% of customers returning products use the new process, the net FTE staffing requirements in the returns processing center would increase. Under the proposed product returns process, customer cycle times (CCT) were substantially reduced, since the initiation of customers return processing begins prior to shipping the product back to the returns processing center. Customers who use a
25

credit card can receive credit in only a few days. For other customers, who request a refund via check or product exchange, the CCT would involve an additional three to four days to complete these transactions. These reduced customer cycle times along with the convenience provided by the postage paid return label represent a significant increase in customer service levels. SUMMARY AND CONCLUSIONS One of the most important service management activities in a retail environment is product return services. These services are important from a strategic point of view because of their ability to positively impact customers satisfaction, engender customer loyalty, and consequently, increase products sales. Successful product return services depend on the design of competent reverse supply chains and support of those supply chains by effective reverse logistics operations. Organizations that are able to achieve competence in these service management activities have the potential to enjoy significant advantages over their competitors, since the design and operation of these activities are not easily duplicated. This study presented an analysis of a set of reverse logistics activities to support a proposed new product returns process for a major direct retailer in the US. The objective of the project was to improve customer service quality and reduce operational costs. To capture the complexity and dynamism of the reverse logistics activities that support the products returns processes, a computer simulation modeling technique was used. The simulation model allowed the comparison of multiple scenarios for both the organizations current product return process as well as a proposed new product returns process. Analysis of the simulation model
26

facilitated the organizations decision making with respect to the design of its reverse supply chain for product returns, enabling it to reduce the time for customers to receive credit or products in exchange for product returns. In addition, the organizations operational resources in the form of returns processing center staffing requirements were able to be reduced.

27

REFERENCES
Arena 3.01, 1997, System modeling corporation, Sewicley, PA Cho, Y., I. Im, and R. Hiltz, 2003, The impact of e-services failures and customer complaints on electronic commerce customer relationship management, Journal of consumer satisfaction, dissatisfaction and complaining behavior, 16, 106-118. Cohen, M.A., and S. Whang, 1997, Competing in product and service: a product lifecycle model, Management science, 43 (4), 535-37. Daugherty, P. J., C.W. Autry, and A. E. Ellinger , 2001, Reverse logistics: the relationship between resource commitment and program performance, Journal of business logistics, 22(1), 107-123. Dennis, M.J. and A. Kambil, 2003, Service management: building profits after the sale, Supply chain management review, (January/February), 42-48. Fitzsimmons, J.A. and M.J. Fitzsimmons, 1998, Service management: operations, strategy, and information technology, Boston, MA: McGraw-Hill. Handfield, R.B. and E.L. Nichols, 1999, Introduction to supply chain management, Upper Saddle River, NJ: Prentice-Hall. Heskett, J.L., T.O. Jones, G.W. Loveman, and E.W. Sasser, Jr., 1994, Putting the service-profit chain to work, Harvard business review, Mar/Apr, 72 (2), 164-74 Lambert, D.M., and J.R. Stock, 1993, Strategic logistics management, 3rd Edition, Irwin, Boston Law, A. M. and W. D. Kelton, 1999, Simulation modeling and analysis, New York: McGraw-Hill Book Company. Meyer, H., 1999, Many happy returns, Journal of business strategy, 20 (4), 27-31. Rabinovich E. and P. Evers, 2003, Product fulfillment in supply chains supporting internet-retailing operations, Journal of business logistics, 24 (2), 205-236.

28

Retzlaff-Roberts, D.L., 1998, Return customers and profits to your bottom line, A FedEx White Paper. Retzlaff-Roberts, D.L. and M.N. Frolick, 1997, Reducing cycle time in reverse logistics, Cycle time research, 3(1), 69-78. Schoenbachler, D. D. and G. L. Gordon, 2002, Multi-channel shopping: understanding what drives channel choice, The journal of consumer marketing, 19 (1), 42-54. Wagner, K.A., M. Mittal, F. de Rosa, and J.A. Mazzon, 2002, Assessing the serviceprofit chain, Marketing science, 21 (3), 294-317.

29

FIGURE 1 A SIMPLIFIED MAP OF REVERSE LOGISTICS ACTIVITIES FOR THE CURRENT PRODUCT RETURNS PROCESS
Returns Package Receives Refund or Exchange

CUSTOMER

Sort Packages

Process Packages & Sort Merchandise RETURNS PROCESSING CENTER

Financial Transaction

Assess Quality, Prepare Merchandise & Consolidate Products Other Destinations No First Quality?

Exchanges Only

Yes Ship Exchange Items Re-shelve Products DISTRIBUTION CENTER

Indicates movement of products Indicates movement of documentation and financial transactions

30

FIGURE 2 A SIMPLIFIED MAP OF REVERSE LOGISTICS ACTIVITIES FOR THE PROPOSED PRODUCT RETURNS PROCESS

CUSTOMER

Customer Calls Retailer

Customer Ships Package

Customer Receives Refund/Exchange

CARRIER

Carrier Receives Package & Scans Label

Carrier Ships Package to Retailer RETURNS PROCESSING CENTER Enter Transaction

Match Transaction

Financial Transaction

Sort by Scanable Label

Traditional Package Processing

Open Package, Reconcile Transaction, & Prep Merchandise To Other Destinations No First Quality?

Traditional merchandise prep

Yes

Distribution Center
Re-shelve Items Ship Exchange Item(s)

Indicatesmovement movementof ofproducts products Indicates Indicates movement of documentation and financial transactions

31

FIGURE 3 COMPUTER SIMULATION MODELING PROCESS

Step 3 Model Data Collection & Analysis Step 2 Preparation of Maps for Current & Proposed Product Returns Processes

Step 4 Simulation Model Development Step 5 Model Verification & Validation Step 6 Model Experimentation

Step 1 Objective: Reduce Cycle Time & Operational Costs

Step 8 Presentation of Results and Recommendations

Step 7 Scenario Analyses

32

FIGURE 4 A SECTION OF THE ARNEA SIMULATION DEPICTING THE PRESENT PRODUCT RETURNS SUB-PROCESS FOR APPAREL AND FOOTWEAR
C. PROCESS APPAREL & FOOTWEAR(AFW)
FW A PA C K A G E S PR O C E SSE D & B A T C H E D IT E M S C O N V E Y E D

ORDERSET PROCESS TIME Assign

A FW TY P MA IL A P FP rocesstm

Duplicate
N ITE MS

Enter
A FW P R O C E S S

MAIL TYPE? Chance

AdvServer
W ith0.779 E lse A FW TY P MA IL A P FP rocesstm A FW P R O C E S S IN G A FW TY P MA IL .E Q . 1

Choose

ORDERSET

REPAIRS? Chance
W ith1.00 W ith.000 W ith0.0

Assign WHITEMAIL PROCESS TIME Assign NOT BATCHEDP K G TY P Assign

P K GR E P B A TC H E D

If E lse

WHITEMAIL CONVEY TIME Delay

0.

Duplicate Leave
N ITE MS

Delay
0.

LeavePACKAGES CONVEYED TO REPAIRS

33

TABLE 1 SIMULATION MODEL DATA REQUIREMENTS

DATA REQUIREMENTS Processing Time Delay Time Distributions Distributions 69 101 16 13 85 120

Product Paperwork Total

Splits 105 8 113

34

TABLE 2 CURRENT RETURN PROCESS RELATIVE AVERAGE PERCENTAGE (RAP) INCREASE IN CUSTOMER REIMBURSEMENT ANDPRODUCT RESELF CYCLE TIMES
Customer Purchase Method 1 2 3 4 Product Category 1 2 3 4 5 6 Scenario B Versus A 0.47% 0.17% 0.13% 0.12% Scenario B Versus A 1.64% 3.81% 2.51% 5.37% 3.04% 2.44% Scenario C Versus A 5.75% 10.45% 6.19% 4.81% Scenario C Versus A 72.42% 8.18% 2.66% 5.43% 93.49% 0.17%

Customer Reimbursement

Product Reshelf

35

TABLE 3 NEW PRODUCT RETURNS PROCESS RELATIVE AVERAGE PERCENTAGE (RAP) CHANGES IN CUSTOMER REIMBURSEMENT AND PRODUCT RESELF CYCLE TIMES
Customer Purchase Method Customer 1 Reimbursement 2 3 4 Product Category Product Reshelf 1 2 3 a (-) Suggests reduction in RAP cycle time. Scenario D Versus B -19.91% a -35.39% a -20.02% a -21.77% a Scenario D Versus B 3.31% 10.05% -5.31% a Scenario E Versus B -24.75% a -44.52% a -24.93% a -25.22% a Scenario E Versus B 3.49% 10.45% -4.87% a

36