28_2013_IJRFT_Reducing Out of Stock, Shrinkage and Overstock Through (1) (1)

International Journal of RF Technologies 4 (2012/2013) 107125DOI 10.3233/RFT-120040IOS Press

107

Reducing out of stock, shrinkage and overstock throughRFID in the fresh food supply chain: Evidence from anItalian retail pilot

Massimo Bertolini, Gino Ferretti, Giuseppe Vignali and Andrea VolpiDepartment of Industrial Engineering, University of Parma, Parma, Italy

Abstract. The paper shows how to leverage RFID technology in fresh fast moving consumer goods(FMCG) industry, and, in particular, to optimally manage stock levels on shelves, shelves replenishment,and shrinkage prevention for fresh perishable products.

We deployed a pilot project on a FMCG retail supply chain, encompassing a distribution centre andtwo stores of a major Italian retailer. About 60 products have been RFID tagged at case level. We real timetracked cases of products for 4 months, through the distribution chain all the way to the stores, where theRFID deployment made it possible to punctually monitor shelves stock levels, backroom stock levels andproduct shrinkage.

We demonstrate that the out of stock (OOS) problem is just one piece of a broader picture, that is shelvesstock optimization. The higher the stock level on the shelves, the lower the need for shelf replenishment,and thus the likelihood of an OOS. However, the capital holding costs and the risk for product shrinkageincrease. The latter issue is particularly relevant for fresh perishable products. This pilot demonstrates thata retailer could reduce OOS, shrinkage and capital holding costs all together, by means of efficient RFIDdata management. Potential savings for fresh products account approx 1.7% of sale turnover.

This study is the first pilot project which assesses the impact of RFID technology on the supply chainof fresh perishable products; a full roll-out of the project is being planned for the next future.Keywords: RFID, out of stock, fresh perishable fast moving consumer goods, pilot study

1. Introduction

This paper presents the results of a pilot project carried out with the purpose ofdemonstrating the potential of Radio Frequency Identification (RFID) to improveOn-Shelf Availability (OSA), which represents a key focus within the Fast MovingConsumer Goods (FMCG) industry. More precisely, it provides a sizeable opportunityfor both retailers and manufacturers to better meet the needs of their consumers, atthe same time maximising sales and profit (Fernie and Sparks, 2004). Based on these

Corresponding author: Massimo Bertolini, Eng., Ph.D., Assistant Professor-Mechanical IndustrialPlants, Department of Industrial Engineering, University of Parma, Via G.P. Usberti 181/A, 43124 Parma,Italy. Tel.: +39 0521 905861; Fax: +39 0521 905705; E-mail: [email protected].

1754-5730/12/13/$27.50 2012/2013 IOS Press and the authors. All rights reserved

108 M. Bertolini et al. / Reducing out of stock, shrinkage and overstock

premises, the purposes of the present study are twofold. From a technological pointof view, the study aims to test the feasibility of backroom and shop-floor inventorymanagement systems based on RFID technology, and to integrate RFID and residentprocesses, as well as EPCISs and legacy systems, in a real working environment. Fromthe managerial perspective, the research shows how RFID technology can be used toenable real time goods monitoring and sharing of OSA data, reducing shrinkage andoptimizing store shelf inventory levels of fresh perishable products. Results presentedin this paper are real data, grounded on a pilot study involving a major FMCG playerof Northern Italy.

The remainder of the paper is organized as follows. In the next section, we reportthe state of the art regarding OSA related issues, then we describe the context wherethe research was carried out, i.e. the pilot study. The data collected from the in-fieldmeasurement and subsequent elaborations are described in Section 4. In Section 5,we present the main findings from the pilot study. Section 6 concludes by discussingthe managerial implications of the study and indicating future research steps.

2. Literature review

On-Shelf Availability is the measure of a product being available for sale to ashopper, in the place he expects it and at the time he wants to buy it. It typicallycharacterised by three key dimensions (Mitchell, 2012):

i) Shelf Availability: it scores zero, in case the item is not on the shelf. There maystill be stock in the store, but it is hidden, in a different location or it is still inthe backroom.

ii) Store Availability: it scores zero when the product is not available anywhere inthe store. It may however be stocked in the distribution centre or en route to thestore.

iii) Warehouse Availability: it scores to zero when the product is not available toorder, as there is no stock in the Distribution Centre (DC).

The FMCG supply chain has been battling the OSA problem for long time. Forretailers and suppliers, the key for success is getting the right product in the rightquantity in the right place and at the right time, with the minimum cost. Despite this,OSA, overstock and Out Of Stock (OOS) still remain persistently high in all retailproduct categories, but they are especially critical when they affect products with shortshelf life (i.e. fresh foods, or all those products that have a due date). Accordingly,recent research has shown that the OOS rate is around 8%, ranging to 4% of sale lossfor a typical retailer and up to 17% for promotional items (Bharadwaj et al., 2002;Corsten and Gruen 2003; ECR Europe 2003; Bottani et al., 2009; Bertolini et al.,2012).

Many authors argue that the adoption of Auto-ID technologies, such as RadioFrequency Identification (RFID), and the use of Electronic Product Code (EPC) are

M. Bertolini et al. / Reducing out of stock, shrinkage and overstock 109

powerful allies in improving OSA and thus reducing OOS problems (e.g., Rekik et al.,2008).

In particular, Hardgrave et al. (2005, 2006) assessed the impact of OOS causedby poor store process execution, such as shelf replenishment, in Wal-Mart Stores.The 29-week study analysed out-of-stock merchandise at 12 pilot stores equippedwith RFID technology, and at 12 control stores not exploiting this technology. Theresearchers found a 16% reduction in OOS in stores equipped with RFID technology;additionally, the study showed that OOS items managed with EPCs were replenishedthree times faster than similar items managed through standard bar code technology.Equally important, Wal-Mart experienced a remarkable reduction in manual orders,resulting in a reduction of excess inventory.

Bottani and Rizzi (2008) argue that an EPC-enabled RFID solution providesenhanced visibility of items and supplies the required information to determine theirlocation once at a facility. This new visibility enables retailers to better manage theirdemand data and replenish their stock more effectively; also, it helps suppliers toverify when the items were received and moved out to the sales floor, with a directimpact on the reduction of OOS at the store (Bottani et al., 2009).

Bertolini et al. (2012) underline that the adoption of RFID technologies to managethe FMCG supply chain could be sustainable from an economical point of view forboth the retailer and the manufacturer. Knowing the exact value of OOS on business,and especially knowing how much of OOS occurrence can be solved through theadoption of Auto-ID technologies, allows a quantitative economic evaluation so tiedto a technology implementation of a RFID project in the FMCG sector. The globalsavings for the manufacturer are estimated to approximately range from 0.7 to 4.5%of the sales turnover; moreover, the results show that the highest savings for boththe retailer and the manufacturer can be achieved for fresh products (i.e. dairy, meatsand frozen foods departments). These results would be largely sufficient to justify thereturn on investment for both players.

Typically, it is observed that a simple re-distribution of the existing stock couldlead to a 10 15% improvement in product availability. This does not imply thathigher stock is the solution; rather, it suggests that the stock needs to be in the rightplace, which can be achieved with correct allocation and management. From a shelfavailability perspective, understanding what needs to be done in store by means ofshelf replenishment, how long this will take, and when it should be done to maximisecustomer satisfaction is at the basis of successful retail operations (Papakiriakopoulos,2005).

From the analysis of the literature, it can be argued that the impact of RFID adoptionat case-level on OOS reduction is threefold:

first, inventory levels at the backroom and at the store can be managed inde-pendently; therefore, the replenishment process can be fully automated. Forinstance, when a product is near out of stock on the shop floor and someinventory is available in the backroom, the shop operator will be assigned a


replenishment task. A RFID handheld reader can be used to locate the stockkeeping unit (SKU) in the backroom and to track the replenished cases;

second, RFID technology reduces inventory inaccuracies (Aloysius, et al., 2008,2011; Hardgrave, 2009), since it improves:

the quality of order preparation and shipping processes at the DC; and receiving, order reconciliation and replenishment process at the retail store,

at the same time enabling a better management of shrinkage; finally, if inventory visibility is extended upstream the supply chain, potential

out of stocks can be foreseen and managed directly by the manufacturer, thusavoiding or reducing them.

RFID enables real-time visibility and better coordination between marketing andmanufacturing activities (Taylor et al., 2003; Handfield and Zahay, 2004), as well asincreased inter-firm collaboration/vertical integration, that represent a major driverfor nurturing innovation capabilities (Bigliardi et al., 2011). On a more operationallevel, RFID has potential to significantly improve the supply chain efficiency, interms of increased processes automation, labour efficiency, and improved accuracyof logistics activities (Agarwal, 2001; Frazier et al., 2005). It is important to note thatthe improvements resulting from automated product identification may only be thetip of the overall RFID benefits, which also includes new business opportunities andstrategies (Krotov, 2008; Bertolini et al., 2010).

3. Context: The RFID Logistics Pilot2 projectThe RFID Logistics Pilot2 (RLP2) project was carried out as the second step of a

research path started at RFID Lab - University of Parma in 2007, when RFID Labresearchers began working on a supply chain project aimed at assessing the feasibilityof automatic data capturing in the supply chain by means of RFID technology. Thatpilot, carried out during the spring and summer of 2008, involved the tracking of12,000 cases and 800 pallets of sliced ham, sandwiches and other fresh food as thegoods moved from production to the retail sales floor. The major results of that projectcan be summarized in a 68% drop in the amount of time spent to check shipmentsfrom a manufacturers warehouse to a retailers DC, where goods were checked duringreceiving process saving 80% of time (Bertolini et al., 2009). As a consequence of thatproject, RFID Lab carried out intense research activities aimed at assessing the causesof OOS and thus the potentials of Auto-ID technologies to reduce OOS occurrence(Bertolini et al., 2010, 2012). These latter results, however, needed to be verified bymeans of an in-field pilot.

FMCG companies that shared the results of the previous research activities thusagreed to setup a pilot project aimed at testing how the RFID technology could beutilized to improve OSA. Hence, the first logistics pilot project has been followed bythe RLP2 in 2010. The project was shared and developed by a consortium of eightItalian companies and supervised by the University of Parmas RFID Lab. The board


consists of retailers Auchan, Coop Italia and Conad, as well as goods manufactur-ers Danone, Lavazza, Nestle` (Buitoni and Purina brands), Parmacotto and Parmalat.Although all the players, including competitive companies, participated actively tothe design phase of the project, brought their contribution, observed the processes andshared the results, only one player (i.e., Auchan) had an active role in the deploymentphase, by testing the technology at its DC and stores.

The development of the research project took place in three steps: in the first one, AS-IS analysis, researchers and companies members mapped

the logistics processes that could be impacted by the implementation of RFIDtechnology;

the second phase, TO-BE reengineering, was aimed at developing a new sce-nario, which encompassed case-level RFID UHF tagging;

in the third phase, researchers developed ad hoc models and metrics for mea-suring the OOS reduction and deriving meaningful results.

During the experimental campaign, which took place from April to July 2011,cartons filled with products were tracked into and out of an Auchans DC, and atthe dock doors of two of its stores. At the store, during the replenishment process,goods were removed from the cardboard cases and placed onto the shelves. Emptytagged cardboard cartons were read eventually in the stores trash compactor, therebyindicating that the products previously packed within must be located on shelves onthe shop floor.

The RFID deployment of the project encompassed fixed RFID UHF readersinstalled at the dock doors of Auchans DC in Calcinate (Bergamo, Italy), as wellas in the receiving area at two Auchan stores, in Curno (Bergamo, Italy) and Ron-cadelle (Brescia, Italy). At the Auchans DC, during the course of the pilot projectthe operators tagged approximately 30,000 cases of goods by means of EPC Gen 2RFID labels. Such cases included about 60 different products such as fresh pasta andsauces, cheeses, hams and other perishable goods.

4. Tracking points and data collection

During the RLP2 project, the following processes were impacted by the implemen-tation of RFID technology, as shown in Fig. 1:

1. DC processes:a. slap & ship;b. shipping;

2. retail store processes:a. product receiving;b. replenishment from the backroom by means of trash process;c. inventory counts;d. check-out.


Fig. 1. The RLP2 supply chain and related RFID tracking points.

The processes mentioned above were mapped before the RFID implementation inthe AS-IS scenario analysis, and reengineered applying RFID technology in TO-BEscenario, in order to assess the improvements achieved thanks to this technology. TheTO-BE scenario was engineered and deployed encompassing different RFID readpoints, called tracking points to reflect their ability to punctually trace the who,where, when, why, and how of every item moving downstream in the supplychain. The whole deployment is based on EPC global standards, as well as on RFIDcase-level tagging, with tag inserted in a product label. Cases were tagged usingEPC SGTINs, and information records were captured in EPC Information Services(EPCIS) according to EPC global standards.

The Auchans DC in Calcinate receives dry and fresh goods from the manufacturers;fresh goods are cross docked directly from the DC to the retailershypermarkets withinthe region, while dry products are stored on shelves in the DC until they are pickedto fulfil orders from stores.

The first tracking point is located at the DC, where individual cases are taggedduring the slap & ship process. Thanks to a software integration with the Auchanlegacy systems, when the warehouse operators process the tested products whichhave been ordered by the two stores participating to the pilot project, a RFID labeltag is printed using a mobile RFID printer and applied to the cases. The tag is a paperlabel embedding a EPC Gen2 passive UHF RFID inlay. This process thus creates aunique association between the products SKU number, the EPC serial and the productcharacteristics available on the Auchan ERP (i.e. product type, production lot, andexpiry date). All the caseslying on the same pallet are then associated together tothe same virtual SSCC code, in a father-son relationship. Hence, during this process,two EPCIS events took place concurrently: the first one is objectadd event, and thesecond one is aggregation event, which links each EPC case to the respective virtualSSCC code pallet.

The second RFID tracking point is connected to the shipping process. Shipmentsusually leave the DC during the evening, with retailers receiving those products earlythe next morning. The warehouse operator uses a couple of RFID portals equippingtwo dock doors for fresh products. Expected cases are checked while being loaded


in the truck, enabling a proof of delivery for pallets shipped by the DC. The shippingprocess generates a transaction event in the DCs EPCIS, and data is then forwardedto the Auchans EPCIS-based software that made such information available to thepilot participants, thereby indicating that the product had left the DC.

After shipping, products arrive at the retail store, where the third RFID read tookplace during receiving of goods. In order to provide the store with the detailed list ofitems to be received in a delivery, a Discovery Services (DS) has been developed toresolve the IP address of the EPCIS where shipment data can be downloaded. TheDS is hosted on a server sited in the RFID Lab of the University of Parma. Thanks tothis application, the RFID door portal is able to retrieve the complete list of expecteddelivery upon the read of the first tag of the pallet. In the store EPCIS, the receivingprocess generates on the one hand an object-observe event and on the other one aquantity event on the backroom of the store.

Then, some of the products are moved to the sales floor for replenishment beginningat around 5 am. The remaining goods are stored in the backroom until they are used torestock the sales floor. During the replenishment, once the products are unpacked fromthe cardboard cases and placed on the shelves, the empty boxes are destroyed in a trashcompactor. Here, a further interrogator captures their IDs and updates once again theinventory of shop floor records to show that the products are now on the shelves andconsequently decrease the inventory level in the backroom. During the replenishment(trash) process, the stores EPCIS automatically generates an object-observe event.

The last tracking point gathering information on product availability are check-outand shrinkage processes, although these are not strictly RFID-enabled tracking points.More precisely, data coming from the Auchans own point-of-sale and shrinkage-management software is integrated with the RFID system as well. Every 15 minutes,data is shared and the EPCIS software updates the products status, determining whichitems have been sold and which have been discarded because of shrinkage (either dueto product spoilage or expiry). The inventory of the shop floor shelves is updatedaccordingly.

The Auchans WMS/ERP software is consequently updated to indicate which prod-ucts are no longer on the shelf, because of in out of stock condition (OOS), and whichare at risk of becoming out of stock, i.e. in near out of stock condition (NOOS). Duringthe pilot, such information is clearly visible on a big display located in the backroom;accordingly, the Auchans staff can manage the shelf stock, replenish products, aswell as reorder OOS products.

The last tracking point is related to inventory counts. Regular inventory countsof the items tracked were performed in the store backroom, using a handheld reader.Backroom inventory count is performed using RFID technology, since the products arestill in cases, while barcode technology is used in the store to scan individually itemsEAN codes. After every RFID inventory count, the store EPCIS captures an objectevent, updating the last known location of an item, i.e. either store floor or backroom.Moreover, an aggregated quantity event is generated to update the inventory level.


Table 1Summary of the reengineered logistics processes

Process Position ObjectiveSlap & Ship DC SGTIN case-level tagging and SSCC

aggregationShipping DC Tracking shipmentsReceiving Store Receiving dock doors in the

backroomBackroom stock update

Trash Store Fixed readers in thebackroom trash compactors

Backroom and shop floor inventory update

Check-out &demarque

Store shop floor Not RFID process Shop floor inventory update

Table 2Example of case tracking on EPCIS

Event time ID actor Event Action Biz step24-JUNE-2011 16:05 Auchan DC Object Add Slapship24-JUNE-2011 16:05 Auchan DC Aggregation Add Slapship24-JUNE-2011 21:32 Auchan DC Object Observe Shipping25-JUNE-2011 05:15 Auchan shop Object Observe Receiving25-JUNE-2011 05:15 Auchan shop Quantity Receiving25-JUNE-2011 09:25 Auchan shop Object Delete Trash

The Table 1 summarizes the RFID reengineered supply chain processes and Table 2reports an example of the EPCIS records for a case which has been read through thetracking points described above, except check-out (not RFID enabled) and inventory.

The tracking points described are used to gather data related to the RFID reengi-neered processes. In addition, four Business Intelligence Modules (BIMs), named Outof Stock, Track & Trace, Inventory, Check-out & Demarque, have been developedto query the EPCIS data warehouse and derive value-added information for researchpurposes and process management, as well as relevant statistics and graphs.

The network infrastructure, shown in Fig. 2, is built in order to create an Internetof things aimed at sharing RFID data with each supply chain player. Auchan hosts amain server in Milanofiori (Milan) server farm and satellite servers in Roncadelle andCurno stores and Calcinate DC; all of them are connected to the same local intranet.

As reported in Fig. 3, the main server hosts the EPCIS repository, that storesthe RFID data, and interfaces for query and capture data, while the satellite servershost a middleware named RSA (RFID System Administrator) that implements thebusiness processes and manages devices, processes and data capturing. The capturingapplication of RSA gathers RFID data from physical RFID devices, by means of webservices (i.e. from RFID printers) or by means of dedicated adapters and middlewarelayer (i.e. from handheld devices and fixed RFID readers). Only the main server


Fig. 2. The RLP2EPCglobal network infrastructure.

accesses the Internet, showing data to authorized clients, while satellite servers canshare data only with the main server. In the RFID Lab, a dedicated server hoststhe Object Name Service (ONS) and the Discovery Services (DS), deployed to getdata from Auchans EPCIS, and the BIMs. The local database is fed with data fromAuchans EPCIS by means of standard EPCIS queries; data is processed and analysedin order to produce quantitative KPIs reported in the business intelligence dashboard.

5. Project resultsThe impact of RFID technology on the store can be manifold; in particular, on the

basis of the findings from the RLP2 project, four main pillars shown in Fig. 4 can beidentified which support RFID adoption in the fresh perishable products category.

As a global result, it is possible to notice that the overall RFID values is +1.75% insales turnover of product category, plus an increased product freshness up to +18%.In the next section we punctually quantify each of the above mentioned contributions.

5.1. RFID impact on OOS

The impact of RFID technology on the OOS has been quantitatively determinedby analysing aggregate data reported in the BIMs. For each product the value of eachOOS occurrence has been computed by multiplying the average sales per hour by thetime the product has been in OOS.

The stock level of a product in the shop floor is real time available thanks to theintegration of the RFID system with the Auchans ERP/WMS, which, as mentioned,


Fig. 3. The RLP2 software architecture.

Fig. 4. The main pillars of RFID adoption pointed out by RLP2 project.

are updated every 15 minutes with check-out and shrinkage data. An example of theshop floor stock level and OOS detection for a given product is charted in Fig. 5.

In order to evaluate the impact of RFID technologies on OOS it is necessary toidentify the cause of each OOS occurrence, as well as to estimate the saving RFID


Fig. 5. Daily stock level for a specific product.

Table 3Different kind of OOS causes

Conditions OOS causesIF Shop floor inventory = 0 AND Backroom

inventory = 0 AND Many days of OOSAND product out of reorder list

THEN Strategic OOS supply chain issues, productdelisting, stores personnel deliberatelydecide not to put that products on shelves(i.e. promote sales of an alternativesize/format)

IF Shop floor inventory = 0 AND Backroominventory = 0

THEN Reorder errors/Inventory inaccuracy

IF Shop floor inventory = 0 AND Backroominventory >0

THEN Missed replenishment

can generate. To identify the cause, according to Bertolini et al. (2012), we haveinvestigated the boundary conditions of inventories when shop floor inventory scoreszero. These boundary conditions define the OOS causes as reported in Table 3.

Table 4 reports an example of daily value of the OOS occurrences for differentproducts, coupled with the corresponding OOS causes. Looking at the first row ofthe table, one can see that product 1 was detected in OOS on June 25, 2011, for twohour. The OOS duration is an average value, which has been computed thanks to thesynchronization of RFID data and check out data, that takes place every hour. TheOOS cause is determined according to the boundary conditions proposed in Table 3.

Overall, as regards the causes of OOS, it can be observed in Table 4 that some ofthem could be efficiently reduced by the adoption of RFID technology, while other


Table 4Daily OOS value for different products

Product OOS from OOS to Duration OOS value OOS cause[hours]

Product 1 25/06/2011 25/06/2011 2 D 2.34 Missed replenishmentProduct 2 17/05/2011 17/05/2011 3 D 8.47 Reorder errors/Inventory inaccuracyProduct 2 20/05/2011 20/05/2011 3 Missed replenishmentProduct 2 25/06/2011 25/06/2011 6 Missed replenishmentProduct 2 29/06/2011 29/06/2011 8 Missed replenishmentProduct 3 09/06/2011 09/06/2011 6 D 8.43 Reorder errors/Inventory inaccuracyProduct 3 12/07/2011 12/07/2011 10 Reorder errors/Inventory inaccuracyProduct 4 30/06/2011 30/06/2011 6 D 1.41 Missed replenishmentProduct 5 04/06/2011 05/06/2011 6 D 6.54 Missed replenishmentProduct 6 04/07/2011 05/07/2011 17 D 9.42 Missed replenishmentProduct 7 10/05/2011 25/05/2011 203 D 46.57 Strategic OOSProduct 7 30/05/2011 30/05/2011 6 D 24.11 Missed replenishmentProduct 7 23/06/2011 23/06/2011 7 Missed replenishmentProduct 8 14/06/2011 15/06/2011 23 D 3.33 Reorder errors/Inventory inaccuracyProduct 9 03/06/2011 03/06/2011 4 D 0.77 Reorder errors/Inventory inaccuracyProduct 10 18/07/2011 18/07/2011 4 D 5.36 Missed replenishmentProduct 11 14/06/2011 15/06/2011 18 D 22.43 Reorder errors/Inventory inaccuracyProduct 11 11/07/2011 12/07/2011 16 Reorder errors/Inventory inaccuracyProduct 11 13/07/2011 13/07/2011 5 Reorder errors/Inventory inaccuracyProduct 12 11/06/2011 20/06/2011 109 D 14.91 Strategic OOSProduct 13 07/07/2011 09/07/2011 39 D 9.22 Reorder errors/Inventory inaccuracy

Table 5Normalized average daily OOS value

OOS cause Can RFID impact on OOS? Normalized value of OOSStrategic OOS No 0.17Reorder errors/Inventory inaccuracy Yes 0.14Missed replenishment Yes 0.14

ones have no potentials to be affected by such technology. This is, for instance, thecase for a product which is OOS and it is intentionally not reordered, to promote asimilar product (strategic stock-out).

The cumulative value of the OOS for the whole project has been computed byadding up the values of the daily OOS of each product and relating such value to thesales turnover with respect of each OOS cause. The obtained values are reported inTable 5.

The reader can appreciate unusual low values of OOS. Typical figures of OOS, asreported in literature, account approx. for 8% of store turnover. The reason for thisoutcome is at least threefold. First, the two stores chosen for the test are among theAuchans best performer hypermarkets, with very low OOS values, compared to otherstores of the retail chain. Second, products chosen for the test have medium or low


Fig. 6. Daily stock levels for two different products.

sales velocity. For such products, as Hardgrave et al. (2006) show, shelf replenishmentissues are minor. Third, we could exploit dedicated store personnel during the project.This was an advantage, because the store managers managed to keep the systemproperly tuned and fully working. On the contrary, results obtained under such settingmight be somehow biased. As a matter of fact, store personnel precisely knew whatwe were looking at (i.e., OOS measurement) and what products we were monitoring;therefore, for these products, unusually high attention might have been paid to shelfreplenishment. This is demonstrated by the replenishment operations performed outof usual time windows (i.e., at early morning, 2:00 PM, or 8:00 PM).

As shown in Table 5, we found that RFID technology can improve the productavailability when the OOS is caused by reorder errors, inventory inaccuracy or missedreplenishment; as a matter of fact, the real time visibility of products is an efficientleverage for accurate stock and order management. The RFID impact on OOS ismeasured as the increase in products availability due to the mentioned causes, andaccounts for approx. +0.28% increase in sales turnover. Conversely, strategic OOS iscaused by products which intentionally are no longer in the reorder list of the retailstore; under that circumstance, the RFID technology cannot contribute to reduce OOSoccurrence.

5.2. RFID impact on shelf inventory managementThe analysis of the shelf cycles and safety stock management for different products

is detailed in this sub-section. The pilot project pointed out that the stock level of someproducts is managed very efficiently (as shown in Fig. 6 product A), while it is lessefficient for other products (as reported in Fig. 6 product B).

The reorder process efficiency can be assessed for both the cycle stocks and thesafety stocks.

In particular, the cycle stock in Fig. 6 product A seems to be appropriate becausethe order quantities are proportional to sales and lead time, and the average inventoryrotation index is 2.09 days of product sales. Safety stocks are optimised too, because


Fig. 7. Determination of potential stock reduction.

they are frequently used, and turn out to be proportional to the sales standard deviationand lead time; the safety stocks rotation index is 1.51 days. Conversely, product Bis not managed with the same efficiency. First, it can be appreciated from Fig. 6 thatthe cycle stock is inappropriate: in fact, order quantities are not proportional to salesand lead time, and the average inventory rotation index is excessively high (approx.21.04 days of sales) compared to the product shelf life of 40 days. The safety stocksare oversized, since they are constant and not proportional to the standard deviationof demand and lead time; the safety stock rotation index for the same product is 17.41days.

The RFID impact on shelf inventory optimization can be quantitatively assessedconsidering a conservative scenario, where only the safety stock level is optimized,while cycle stocks are kept unchanged. In the proposed conservative scenario, thesafety stocks of each product at the shop floor have been quantitatively reduced by aconstant factor, computed as the average of the minimum peaks of inventory levels(for detail refer to Fig. 7). For every product, the average stock level correspondingto the lowest peaks is computed and the corresponding cost is calculated multiplyingthe value of the item by the average minimum stock level. This is a capital invested


Table 6Different kind of shrinkage causes

Conditions Shrinkage causesIF Physical inventory = WMS inventory

AND Shop floor inventory >0 ANDDemarque on shop floor

THEN Reorder errors (strict expiration date,damage)

IF Physical inventory = WMS inventoryAND Backroom inventory >0 ANDDemarque on backroom

THEN Replenishment policy errors (strictexpiration date, damage)

IF Physical inventory /= WMS inventoryAND Demarque

THEN Other causes (shoplifting, employee theft,paperwork errors, supplier fraud, damage)

Table 7Normalized average daily shrinkage value

Shrinkage cause Can RFID impact on shrinkage? Normalized value of shrinkageReorder errors Yes 0.60Replenishment policy errors Yes 0.54Other causes No 0.61

in unnecessary stocks; multiplying the value by the weighted average capital cost(WACC) of the company, and referring the result to the product sales turnover, weobtain the saving that can be achieved by RFID adoption. The savings for all productsare consequently added up, resulting in +0.33% of the category sales turnover overall.

5.3. RFID impact on shrinkage

The shrinkage of products is a decrease in inventory due to shoplifting, employeetheft, paperwork errors, supplier fraud, damage and strict expiration date (The GlobalRetail Theft Barometer, 2010). While the physical inventory is affected by the above-mentioned causes, the inventory level on the Auchans WMS is manually adjustedaccording to the effective stock, which is usually lower (this process is internallynamed demarque). Shrinkage is inevitable; hence, the causes of shrinkage were inves-tigated, grouping them into three categories, as reported in Table 6. In this regard, thein-field research pointed out that some of those categories can be removed by means ofRFID implementation; hence, in order to evaluate the corresponding shrinkage reduc-tion, it is necessary to identify the cause of the observed shrinkage and to estimate itsvalue. To identify the shrinkage cause, we have investigated the boundary conditionsof the products and the inventory (i.e. expiry date, physical and WMS inventories ofbackroom and shop floor, receiving and last replenishment dates) during the executionof the demarque process. These boundary conditions define the shrinkage causes asreported in Table 6.


Table 8Inventories: A short blanket

Shelf display Shelf display Out of stock Out of stock Demarque Demarque Invested capitals Invested capitals Product freshness Product freshness

Fig. 8. Fresh product shelf-life schema.

The cumulative value of the shrinkage for the whole project has been computed byadding up the daily demarque value of each product and normalizing such value tothe sales turnover. For the considered different causes of shrinkage, its average valuesare reported in Table 7.

It can be seen from Table 7 that RFID technology can improve the availability ofproducts when shrinkage is caused by reorder or management errors; in fact, the realtime visibility of goods and related data can be exploited to set up more efficientreplenishment and order management processes. Accordingly, the overall increase inproducts availability achievable thanks to RFID adoption accounts for approx +1.14%.

Other causes of shrinkage we observed are shoplifting, theft, and damage; however,RFID technology cannot contribute to reduce them.

5.4. RFID impact on fresh product freshnessA qualitative result achieved by the project is related to the fact that product

freshness could be increased through RFID deployment. In fact, as a result of theoptimization of the safety stock levels on the shop floor, a decrease in the rotationindex of the products on the shelves was observed, which enhances the freshness ofthe products available to the final customer.

In the Italian fresh perishable fast moving products industry, manufacturers keeps1/3 of product shelf life while guarantees to the retailer the remaining 2/3. For pilotproducts the average shelf life guaranteed at the retailers DC inbound dock doors


results in 44.6 days. Further 2.9 days are spent for DC receiving, cross docking, andshipping, plus store receiving and replenishment. The AS IS shelf stock has a rotationindex of 7.6 days of sales. As a result, the average shelf life available to customer is34.1 days as stated in Fig. 8.

The optimisation of the shelves stock level, on the basis of the criteria mentionedin Section 4.2, allows reducing the average rotation index of products on the shopfloor to approx. 1.6 days, thanks to the stock reduction for every product by a constantfactor (i.e., average of the local minima). The resulting average shelf life available tocustomer is thus increased to 40.1 days, gaining +18%.

6. Conclusions

In this paper, we have presented the numerous benefits that can be achieved throughRFID case-level tagging deployment in the supply chain of fresh perishable products.Quantitative results provided were obtained by means of real data, grounded on theRLP2 pilot study. The major benefits gained through RFID deployment, referred tothe sales turnover, can be summarized as follows: (i) OOS reduction can reach 0.28%;(ii) efficient shelf inventory management scores 0.33% and, (iii) shrinkage reductionscores 1.14%. Therefore, as a global result, the overall RFID values a +1.75% increasein sales turnover. From a qualitative point of view, we have also observed a relevantimprovement in product freshness (+18%).

Qualitatively, it is well known from the literature that increasing the shelf stocklevel helps preventing the OOS occurrence; conversely, the side effects are relatedto higher shrinkage and invested capitals, while product freshness is worsened. Onthe other hand, reducing the shelf stock level increases the probability of OOS occur-rence, as well as the need of replenishment operations, although shrinkage and capitalinvestment are more efficient, as shown in Table 8. Conversely, by exploiting RFIDtechnology, OOS can be reduced through timely and punctual shelf replenishmentwhen a NOOS situation is observed. Moreover, shelf stock levels can be optimizedthanks to full visibility of shelf inventory levels, resulting in a reduced capital hold-ing costs. Last, but not least, higher inventory accuracy and proper management ofproduct shelf life during shelf replenishment provides room for further significantsavings.

From a theoretical perspective, this study contributes to the existing knowledge bydemonstrating and quantifying the economic benefits of RFID case-level implemen-tation in the fresh perishable fast moving products supply chain, enabling quantitativecosts/benefits analyses. The originality of the paper can thus be found in the quan-titative validation of theoretical assumptions thanks to the results of an in-field pilotin FMCG supply chain. Moreover, to our knowledge, this is the first study where theRFID impact has been assessed not only on OOS reduction, but also on shrinkageand capital holding costs optimization.

Limitations of this study could be found both in the number and type of storesinvolved in the pilot, which are among the Auchans best performing hypermarkets,


as well as in the products chosen for the pilot, which are limited to a specific categorywith a low sale velocity. To this extent, we are planning to extend the research througha further pilot study, which will include stores where OOS occurrence should besignificantly higher, as well as products with higher sales velocity.

References

Agarwal, V. (2001). Assessing the Benefits of Auto-ID Technology in the Consumer GoodsIndustry. Cambridge University, Auto-ID Center. Available at /http://www.autoidlabs.org.uk/camwhitepapers.html (accessed July 2005).

Aloysius, J.A., Goyal, S. & Hardgrave, B.C. (2008). Does RFID improve inventory accuracy? A preliminaryanalysis. International Journal of RF Technologies: Research and Applications, 1(1), 4456.

Aloysius, J.A., Goyal, S. & Hardgrave, B.C. (2011). Improving inventory management in the retail store:The effectiveness of RFID tagging across product categories. Operations Management Research,4(12), 613.

Bertolini, M. (2009). Performance measurement and analysis for an RFID technology application tocommercial products. International Journal of RF Technologies: Research and Applications, 1(4),279305.

Bertolini, M., Bottani, E., De Vitis, A. & Butcher, T. (2010). A pilot project for evaluating the impact ofAUTO-ID technologies on out-of-stock. Supply Chain Forum: An International Journal, 11(4), 2434.

Bertolini, M., Bottani, E., Rizzi, A. & Volpi, A. (2009). The benefits of RFID and EPC in the supply chain:Lessons from an Italian pilot study. In D. Giusto, et al. (eds.). The Internet of Things: 20th TyrrhenianWorkshop on Digital Communications, Pula (Italy), vol.1, pp. 293302. DOI 10.1007/978-1-4419-1674-7 28, ISBN: 978-1-4419-1673-0.

Bertolini, M., Ferretti, G., Montanari, R., Rizzi, A. & Vignali, G. (2012). A quantitative evaluation of theimpact of the RFID technology on shelf availability. International Journal of RF Technologies, 3, 122.

Bharadwaj, S., Corsten, D. & Gruen, T. (2002). Retail Out-of-Stocks: A Worldwide Examination of ExtentCauses and Consumer Responses. The Food Institute Forum (CIES, FMI, GMA).

Bigliardi, B., Colacino, P. & Dormio, A.I. (2011). Innovative characteristics of small and medium enter-prises. Journal of Technology Management and Innovation, 6(2), 8393.

Bottani, E., Montanari, R. & Rizzi, A. (2009). The impact of RFID technology and EPC system on stock-out of promotional items. International Journal of RF Technologies: Research and Applications, 1(1),622.

Bottani, E. & Rizzi, A. (2008). Economical assessment of the impact of RFID technology and EPC systemon the fast moving consumer goods supply chain. International Journal of Production Economics,112(2), 548569.

Corsten, D. & Gruen, T. (2003). Desperately seeking shelf availability: An examination of the extent, thecauses, and the efforts to address retail out-of-stock. International Journal of Retail & DistributionManagement, 31(12), 605617.

ECR Europe. (2003). Optimal Shelf Availability Increasing Shopper Satisfaction at the Moment of Truth,ECR Europe Publications, available at www.ecrnet.org

Fernie, J. & Sparks, L. (2004). Logistics and retail management: Insights into current practice and trendsfrom leading experts, 2nd edition, Kogan Page Business Books.

Frazier, G.V., Prater, E. & Reyes, P.M. (2005). Future impacts of RFID on e-supply chains in groceryretailing. Supply Chain Management: An International Journal, 10(2), 134142.

Handfield, R.B. & Zahay, D.L. (2004). The role of learning and technical capabilities in predicting adoptionof B2B technologies. Industrial Marketing Management, 33(7), 627641.

Hardgrave, B.C., Miller, R. & Waller, M. (2005). Does RFID reduce out of stocks? A preliminary analysis.available at http://itri.uark.edu/ITRI-WP058-1105


Hardgrave, B.C., Miller R. & Waller, M., (2006). RFIDs impact of on out of stocks: A sales velocityanalysis. available at http://itri.uark.edu/ITRI-WP068-0606

Hardgrave, B.C. (2009). Item level RFID for apparel: The Dillards RFID initiative. ITRI-WP146-0409.Available at http://itri.uark.edu/ (Accessed March 2012)

Krotov, V. (2008). RFID: Thinking outside of the supply chain. Available at http://www.cio.com/article/-278174108/RFID Thinking Outside of the Supply Chain (Accessed June 2009).

Mitchell, A. (2012). Improving On-Shelf Availability. It Matters More. White paper, Symphony IRI Group,Technology Services.

Papakiriakopoulos, D. (2005). Towards the development of an algorithm to discover out-of-shelf situations.In G. Doukidis, A. Vrechopoulos (Eds.). Consumer driven electronic transformation, Springer, Berlin,pp. 167177. ISBN 978-3-540-22611-6.

Rekik, Y., Sahin, E. & Dallery, Y. (2008). Analysis of the impact of the RFID technology on reducingproduct misplacement errors at retail stores. International Journal of Production Economics, 112(1),264278.

Taylor, R.R., Morgan, R.M. & Morton, A.R. (2003). Efficient versus responsive supply chain choice:An empirical examination of influential factors. Journal of Product Innovation Management, 20(6),430443.

The Global Retail Theft Barometer. (2010). Making Progress in Preventing Shrink. Available athttp://www.globalretailtheftbarometer.com/pdf/grtb-2010-summary.pdf

28_2013_IJRFT_Reducing Out of Stock, Shrinkage and Overstock Through (1) (1)

Documents

shelves stock levels

pilot project

backroom stock levels

stock oos problem

fresh products account

impact of rfid technology

fresh food supply chain

shelves replenishment