Collaboration Forms and Differencies Article

International Journal of Production Research,Vol. 44, No. 16, 15 August 2006, 3207–3226

Collaboration forms, information and communication

technologies, and coordination mechanisms in CPFR

P. DANESE*

Department of Management and Engineering, University of Padua,

Stradella S. Nicola, 3, 36100 Vicenza, Italy

(Revision received December 2005)

Collaborative Planning, Forecasting, and Replenishment (CPFR) programmesseek to improve the ability to anticipate and satisfy future demand by enhancingcollaboration among companies within the supply network. Despite the existenceof a detailed and comprehensive process model—published by the VoluntaryInterindustry Commerce Standards Committee—in practice CPFR can takea number of different forms. This paper aims to investigate differences in CPFRimplementation as to the type of inter-company collaboration put into practice,and the Information and Communication Technologies (ICTs) and coordinationmechanisms adopted to perform CPFR. Moreover, it seeks to analyse therelationships between these dimensions in order to comprehend and explain therationale behind the managerial choices that lead companies to implementdifferent CPFR configurations. The multiple-case study method is adoptedto investigate the research questions. In particular, the implementation of CPFRin seven supply networks is examined. This research identifies six types ofcollaboration that can be performed to implement CPFR. Then, using thistaxonomy as its starting point, it accounts for differences in the adoptionof the ICTs and coordination mechanisms necessary to support CPFR. Thepaper’s conclusions summarize the research’s main theoretical and managerialcontributions.

Keywords: Supply-chain management; Collaborative Planning, Forecasting, andReplenishment; Multiple-case study; Coordination mechanisms; Information andCommunication Technologies; Supply-chain collaboration

1. Introduction

Recent years have witnessed growing excitement on the subject of CollaborativePlanning, Forecasting and Replenishment (CPFR) as a consequence of theimpressive results achieved by successful CPFR programmes in supply networkscoordinated by large, high-performing focal firms, such as Wal-Mart (Hill 1999,Parks 1999, 2001, Songini 2001), Nabisco & Wegmans (Parks 1999), Procter& Gamble (Schachtman 2000), and Kmart (Songini 2001). CPFR refers to‘the collaborations where two or more parties in the supply chain jointly plana number of promotional activities and work out synchronized forecast, on the

*Email: [email protected]

International Journal of Production Research

ISSN 0020–7543 print/ISSN 1366–588X online � 2006 Taylor & Francis

http://www.tandf.co.uk/journals

DOI: 10.1080/00207540600557991

basis of which the production and replenishment processes are determined’

(Larsen et al. 2003, p. 532). Several authors describe the pilot project between

Wal-Mart, Warner-Lambert, Benchmarking partners, SAP, and Manugistics in 1995

as the first CPFR programme ever implemented (Cooke 1998). Afterwards, in 1998,

the Voluntary Interindustry Commerce Standards (VICS) Committee—a group

dedicated to the adoption of bar-coding and EDI in department store/mass

merchandise industries—developed a nine-step process model as a guideline for

CPFR implementation (VICS 1998).Despite the existence of this process model that suggests how to design and

implement CPFR, some authors (ECR Europe 2001, 2002, Larsen et al. 2003,

Seifert 2003) maintain that in practice, CPFR can take a number of different forms.

First, the type of collaboration among companies involved in CPFR can vary

significantly. According to Larsen et al. (2003), for instance, the level of integration

(e.g. the degree of discussion, data exchanged, plan synchronization, etc.) or the

business processes involved in the CPFR collaboration can differ. In addition,

by comparing CPFR cases that exist in the literature, it emerges that different types

of Information and Communication Technologies (ICTs) and inter-firm coordina-

tion mechanisms can be used to support CPFR. Simply as an example, the adoption

of ICTs ranges from the use of simple tools, such as a fax, to more advanced

Internet-based CPFR solutions. Similarly, coordination mechanisms can encompass

temporary task forces, personnel dedicated to customer-supplier interfaces, and even

more complex mechanisms.The objectives of this research are to:

(1) investigate differences in CPFR implementation as to the type of collabora-

tion put into practice, and the ICTs and coordination mechanisms adopted

to perform CPFR;(2) analyse the relationships between these dimensions in order to comprehend

and explain the rationale behind the managerial choices that lead companies

to implement different CPFR configurations.

As to the coordination mechanisms adopted to support CPFR, it is worth noting

that this paper focuses on analysing only a particular group of coordination

mechanisms that seems to be of crucial importance in assuring coordination during

CPFR, encouraging mutual adjustment among units. Mintzberg (1979) calls them

liaison devices. Those of particular importance include: liaison positions, meetings,

and integrating managers.The paper is organized as follows. First, I examine the existing literature on

CPFR in order to highlight what researchers have found on the subject and

what theories have been put forward. A discussion of the research design follows.

Then, I analyse the differences in CPFR implementation and extract the key

propositions explaining the relationships between the type of collaboration put

into practice, the ICTs, and liaison devices adopted to perform CPFR. Finally,

conclusions summarize the main theoretical and managerial contributions of the

research.

3208 P. Danese

2. Literature review

Much of the research on CPFR is based on the VICS process model (see Barratt

and Oliveira 2001, Harrington 2003, Seifert 2003, Crum and Palmatier 2004)

(figure 1).The VICS Guide (1998) suggests that participating companies first formally

commit to a programme of demand forecast collaboration—including an agreement

on sponsors within each organization and key metrics (step 1)—and develop

a joint plan (step 2) that establishes item-management profiles, such as order

minima and multiples, lead times, reorder frequency and promotions for those

Stock-Keeping Units (SKUs) upon which they will collaborate. On the basis

of consumption data, participating companies create sales forecasts to support the

joint business plan (step 3), and cooperate in identifying and resolving those

exceptions/deviations (e.g. forecast inaccuracies, over/under stock situations, execu-

tion problems) that require the readjustment of the joint sales forecast (steps 4

and 5). Then, by combining the sales forecast, inventory strategies, and other

information, it is possible to generate a specific order forecast that allows the seller to

allocate production capacity against demand while minimizing safety stock (step 6).

Again, in this case, the participating companies jointly set order forecast constraints

and cooperate in identifying and resolving exceptions, thus creating new adjusted

order forecasts (steps 7 and 8). The final step establishes the replenishment plan

by transforming the order forecast into a committed order (step 9).However, some authors, such as Larsen et al. (2003) look at CPFR by

considering different degrees of collaboration rather than a slavish step-by-step

model. They state that CPFR can be implemented in various ways, as it can be

differentiated in terms of both the scope of the collaboration—indicating the number

of business processes involved—and the depth of the collaboration—measuring the

integration of business processes. A similar concept also emerges from the ECR

guide on CPFR (ECR Europe 2001). It suggests that a company may decide to

collaborate by implementing only few steps of the VICS model, for instance, only by

managing order forecasts (steps 6, 7, and 8 of the VICS model). Again, Seifert (2003)

recognizes that different forms of CPFR collaboration can exist among customers

and suppliers. He believes it is due to the fact that CPFR is implemented by means

of a phased approach. For example, several companies adopting the Vendor

Managed Inventory (VMI) approach are trying to modernize it by adding

collaborative technologies to facilitate greater communication and improve results.

GenerateOrder

ResolveExceptionsto OrderForecast

IdentifyExceptionsto OrderForecast

CreateOrderForecast

ResolveExceptionsto SalesForecast

IdentifyExceptionsto SalesForecast

CreateSalesForecast

CreateJointBusinessPlan

DevelopFront EndAgreement

PLANNING FORECASTING REPLENISHMENT

Figure 1. Activities in the CPFR process.

Collaborative Planning, Forecasting, and Replenishment 3209

VMI is an integrated approach for customer-supplier coordination, by which thesupplier decides on the appropriate inventory levels for each product and on theinventory policies to maintain these levels. As trust builds and technologyinfrastructure is acquired, the collaborative VMI process can be extended to takeon additional aspects of the CPFR standards and achieve greater benefits.

CPFR implementation cases described in the literature seem to confirm thatCPFR can take a number of different forms across supply networks. Various typesof CPFR partnerships are being experimented with. Wal-Mart and Warner-Lambertembarked on the first CPFR pilot, involving Listerine products. They used specialCPFR software to exchange forecasts. Supportive data, such as past sales trends,promotion plans and even the weather, were often transferred in an iterative fashionto allow them to converge on a single forecast should their original forecasts differ.The pilot scheme was very successful, resulting in an increase in Listerine sales andbetter fill rates, and in a reduction of inventory investment (Cooke 1998, Hill 1999).Other famous examples of CPFR pilots include Sara Lee’s Hanes and Wal-Mart,involving 50 stock-keeping units of underwear supplied to almost 2500 Wal-Martstores (Hill 1999, Parks 1999, 2001, Songini 2001). In 1996, Heineken USA employedCPFR to cut its order-cycle time and is currently providing collaborativeplanning and replenishment software to its top 100 distributors (Aviv 2001).Procter & Gamble has several active CPFR pilots under way (Schachtman 2000).Levi Strauss & Co. incorporates certain aspects of the CPFR business processinto its retail replenishment service (e.g. by creating joint business plans andidentifying exceptions) (Aviv 2001). Additionally, in the ECR report entitled‘European CPFR Insights’, several CPFR pilots are described, including:Unilever–Sainsbury’s–GNX, Condis–Henkel–Cartisa, Kraft–Sainsbury’s–GNX,Carton–Scholler, and Vandemoortele–Delhaize (ECR Europe 2002).

By comparing the CPFR cases mentioned above, the existence of commonalitiesas well as significant differences in CPFR implementation emerge. In particular,one of the main differences concerns the collaboration approach—businessprocesses included in the collaboration and the way of collaborating (e.g. degreeof discussion, communication/synchronization, etc.) may differ. Moreover, althoughseveral CPFR cases involve manufacturer-retailer dyads, it is worth noting thatin some cases, a company can collaborate with numerous other supply networkmembers both upstream and downstream in the supply network (ECR Europe2001, 2002).

Finally, cases described in the literature confirm the ECR Europe’s (2002) beliefthat different types of technologies and tools can be used to support CPFR.Companies can sometimes engage in CPFR through low-tech approaches such asface-to-face planning meetings, sending daily sales information via fax, spreadsheetsof sales, ordering and promotional data via email or by using special interfacesfor data transfer, such as Electronic Data Interchange (EDI).

Additionally, Advanced Planning and Scheduling (APS) tools can sometimesbe used to support decision-making during CPFR collaboration. They providequantifiable information to illustrate various possible solutions, thus allowingthe decision-maker to decide which one is most appropriate, based on other,non-quantifiable factors. Moreover, APS tools allow the decision-maker to analysethe consequences of their decisions, depending on different possible scenarios.This kind of what-if analysis can help avoid problems before they occur.

3210 P. Danese

Finally, some of the most sophisticated technologies adopted to supportCPFR include the Internet-based CPFR solutions (Attaran and Attaran 2002a, b,Sparks and Wagner 2003). According to Seifert (2003), these include:

(1) Web-based collaboration, designed to allow process and information sharingamong multiple trading partners. Information that can be shared includesforecasts, promotional activities, inventory plans, POS data, transportationrequirements, and changes to previously agreed-to plans.

(2) Event management and analysis, allowing participants to engage in exceptionmanagement. The aim is to monitor and alert participants to exceptions,status changes, and discrepancies. When unexpected variations are foundamong sets of data, the application issues a notification, or alert, to theappropriate participants so they can get online, review the exception, and takeaction as required.

(3) Tracking and reporting, providing the capability to analyse performanceagainst key indicators and to generate management reports.

CPFR solutions can sometimes be available in a hub or private trading network,allowing multiple tiers of suppliers and customers to collaborate. They can interactvia a neutral intermediary—the (e-)exchange. Internet trading exchanges enable theexistence of many-to-many collaborative relationships using the same standards,communication links, systems, and metrics (Seifert 2003).

As well as Information and Communication Technologies (ICTs), another aspectseems to be particularly important when it comes to designing a CPFR process: theliaison devices activated to align different supply network activities. ECR Europe(2002) underlines the importance of adopting liaison devices, such as meetingsand cross-company teams to ensure collaboration. According to this guide, thesecoordination mechanisms are particularly useful during the establishment of thefront-end agreement and when companies collaborate in resolving order/forecastexceptions. In the Unilever–Sainsbury’s–GNX case, reported in ECR Europe (2002),collaboration takes place among the logistics representatives every Friday, withdiscussion of key findings and agreement on changes, where appropriate, to bringforecasts into line.

Frankel et al. (2002) maintain that meetings are important, especially during theearly stages of the collaborative initiative, when representatives from the firmsinvolved often hold formal daily, weekly, and monthly meetings to discuss goals andaccomplishments. However, they also emphasize that most companies implementingCPFR continue to hold periodic meetings in which representatives from all areasof the collaborative firms participate.

3. Research methodology

The multiple-case study method is adopted to investigate the research questions.The literal and theoretical replication issues guided the selection of the cases.In particular, they were assessed by ascertaining that both similarities and differencesexisted among the cases with regard to the type of collaboration put intopractice, and the ICTs and coordination mechanisms adopted to perform CPFR.

Collaborative Planning, Forecasting, and Replenishment 3211

The case-selection process resulted in the identification of seven cases of supplynetworks whose central firms operate in different sectors (table 1).

All data were gathered through company visits made between 2001 and 2004.Triangulation was used to ensure research reliability obtaining the same piece ofinformation from different sources: semi-structured interviews, documentation,archival records, and direct observations (McCutcheon and Meredith 1993).

Data analysis involves two steps:

(1) within-case analysis;(2) cross-case analysis.

In the within-case analysis, I break down case-study data and group it accordingto the macro-categories illustrated in table 2, namely the collaboration put intopractice to perform the CPFR (i.e. CPFR collaboration), the ICTs and liaisondevices adopted to support CPFR.

With regard to the CPFR collaboration, collected data concern:

(1) the number and type of business processes involved in the collaboration;(2) the level of integration (e.g. degree of discussion, coordination/

synchronization, etc.);(3) the number of units with which a company collaborates.

Data on the ICTs encompass information about the type of technologies adopted forcollecting, collating, processing, and disseminating information during CPFR (faxand/or email; electronic data exchange and integration; APS tools; Internet-basedCPFR solutions).

Finally, data on liaison devices regard the types of liaison devices adoptedto encourage mutual adjustment among companies. From the cases, it emerges thatliaison devices of particular importance are: liaison positions; task forces/standingcommittees and integrating managers. According to Mintzberg (1979), liaisonpositions are jobs created to directly coordinate the work of two units withouthaving to pass through managerial channels. These positions carry no formalauthority per se; rather, those who hold them must use their power of persuasion,negotiation, and so on to bring the two sides together. Instead, task forces andstanding committees are institutionalized forms of meetings which bring membersof a number of different units together to deal respectively with a temporary or morepermanent and regular issue. Finally, integrating managers are essentially liaisonmanagers with formal authority, not over the units they link, but over somethingimportant to those units.

The within-case analysis makes it possible to compare and contrast the CPFR inthe seven analysed networks. This step is necessary in order to discuss the cross-caseanalysis where axial and selective coding techniques are used to make connectionsamong categories to explain the phenomenon of interest.

4. CPFR practices across supply networks

By comparing the analysed cases, it emerges that CPFR collaboration, and theICTs and liaison devices adopted to support CPFR, can vary across companiesimplementing CPFR for different reasons. In this section, a taxonomy is proposed

3212 P. Danese

Table

1.

Overview

ofthecases.

Centralcompany

Product

managed

throughCPFR

Supply-network

mem

bersinvolved

inthecollaboration

AStarter

batteries

.CompanyA:MU

producingstarter

batteries

.Twoowned

centralDCs;severalinternationalDCs

BInjectable

cephalosporins

.CompanyB:MU

producinginjectable

cephalosporins

.Owned

DCslocatedworldwideanddirectlyreplenished

bycompanyB

.Owned

andindependentsuppliers

CAntibiotics

andanaesthetics

.CompanyC:MU

producingananaesthetic

andresponsible

forthefinal

packagingofanantibiotic

.Owned

DCslocatedworldwide

.Productionandpackagingplants

(owned

andindependent)

DIndustrialsystem

components

andairconditioners

.CompanyD:MU

producinganddistributingcomponents

forindustrial

system

sandairconditioners

.SomeOEMs,distributors

andcontractors

EConsumer

andsoftgoods

.CompanyE:salescompanysellingandmarketingconsumer

andsoft

goods

.Largecustomers(retailers)

FFats

andmargarines,sw

eets,

toilet

paper

andfeminine

hygieneproducts

.CompanyF:subsidiary

ofafoodretailer

locatedin

Belgium,responsible

forestablishingpromotionalplansforthelocalsupermarketsandfor

managingthereplenishmentoftheBelgianDC

.Threesuppliersproducingfats

andmargarines;sw

eets

andfeminine

hygieneproducts

GCorrugatedcardboardsfor

product

transport

.CompanyG:MU

producingcorrugatedcardboards

.Somecustomers(foodproducers)

Collaborative Planning, Forecasting, and Replenishment 3213

Table

2.

Within-case

analysis.

Macro-category

Data

tobeincluded

CPFR

collaboration

Number

andtypeofbusinessprocesses

involved

inthecollaboration

CPFR

phases;collaborationin

definingpromotional,salesandorder

forecast

plans

Level

ofintegration

Supply-network

mem

ber

rolesin

conductingCPFR

phases;how

each

phase

iscarriedout(data

tobeexchanged,synchronizationofplans,

meetings,identificationofexceptions,etc.)

Number

ofinteractingmem

bers

Number

ofCPFR

partnerscollaboratingwiththecentralcompany

ICT

Fax,em

ail

Data

sentbyfaxand/orem

ail

Electronic

data

exchangeandintegration

Adoptionoftoolsfacilitatingtheexchangeofdata

andtheirintegration

withthelocalsystem

s;types

ofdata

exchanged

bythesetools;when

they

havebeenadopted;whattheadvantages

oftheiradoptionare

APStools

AdoptionofAPStools;types

ofdecisionssupported

byAPS;when

they

havebeenadopted;whattheadvantages

oftheiradoptionare

Internet-basedCPFR

solution

AdoptionofInternet-basedCPFR

solutions;theirfunctions;applications;

when

they

havebeenadopted;whattheadvantages

oftheiradoptionare

Liaisondevices

Liaisonpositions

Inform

ationonpeople

responsible

forassuringcoordinationamong

companies;whattheirfunctionis

Meetings

When

meetingsare

organized;whoparticipates;whatisdecided

during

thesemeetings,etc.

Integratingmanagers

Inform

ationonpeople

responsible

forassuringcoordinationamongunits

andtakingdecisionsduringCPFR

(whattheirfunctionis,whatthey

candecide,

etc.)

3214 P. Danese

in which CPFR collaborations are classified into six groups. Then, using thistaxonomy as the starting point, differences in ICT and liaison device adoption areexplained.

4.1 CPFR collaborations

Differences in CPFR collaborations can be grasped by considering the followingtwo variables:

(1) the number of interacting units;(2) the depth of the collaboration (figure 2).

The ‘number of interacting units’ variable indicates the number of supply-network members with whom the central company collaborates in performingthe CPFR. The number of interacting units can be high or low. In fact, from theanalysed cases, a clear distinction between a group of companies collaboratingwith only a few partners (fewer than four)—assigned to the low-class—and a groupof companies collaborating with several partners (more than 20)—assigned to thehigh-class—seems to emerge. For instance, central company G collaborates withfour customers, while central company B collaborates with about 50 distributioncentres (DCs) located worldwide.

As to the depth of the collaboration, three levels have been identified:

(1) communication,(2) limited collaboration, and(3) full collaboration,

by comparing (1) the number of business processes involved in the collaboration and(2) the level of integration. According to Barratt and Oliveira (2001), CPFR includes

DE

PT

H O

F C

OL

LA

BO

RA

TIO

N

Communication

LimitedCollaboration

FullCollaboration

Low High

NUMBER OF INTERACTING UNITS

TYPE 4

Company B - DCsCompany B - suppliers Company C - DCs Company A - DCs

TYPE 2

Company E - customersCompany F - suppliers

TYPE 5

Company B - suppliersCompany D - customers

TYPE 3

Company G - customers

TYPE 1

Company E - customers Company F - suppliers

TYPE 6

Company A - DCsCompany C - subcontractors/MU

Figure 2. Types of CPFR collaboration.

Collaborative Planning, Forecasting, and Replenishment 3215

four major sub-processes: (1) issuing of the front-end agreement; (2) developmentof the joint business plan; (3) management of sales forecasts; and (4) managementof order forecasts (see figure 1). By analysing the cases, it can be seen that the CPFRis not always extended to all these sub-processes. Moreover, the level of integrationamong companies can be very different, confirming the literature. In some cases,for example, two or more companies only exchange data/information withoutcoordinating or synchronizing their plans. In other cases, companies collaborateto agree upon how they should coordinate their plans through joint decision-makingor to jointly solve sales or order forecast ‘exceptions’.

On a communication level, companies often collaborate simply by exchangingdata and information with trading partners. The types of data exchanged can differ.For example, central company D receives order forecast plans from its customersand then uses them to elaborate sales forecasts. Instead, central company A receivesstock level and consumption data from the DCs. In both cases, the collaborationis simply a sort of data communication. Indeed, parties neither coordinate norsynchronize business, sales, or order forecast plans.

Limited collaboration differs from communication by taking the collaborationa little further than mere data exchange. Parties must also synchronize their plansand manage exceptions. Central company B, for example, receives data on stocklevels and sales forecast plans from the DCs. On the basis of these data, the centralsystem proposes the replenishment plans, suggesting dates to the central companyfor the deliveries of final products to each distribution centre. The deliveries aredecided in order for the stock level at the DCs’ facilities to fall within a jointlyestablished range (called the ‘VMI min–max’ range). Replenishment planshave then to be confirmed by the planners within both the central companyand the DCs. If a DC does not confirm the plans, new replenishment plans areproposed. Moreover, if a DC asks for additional orders that fall within thefrozen planning horizon, the central company can propose—on the basis of awhat-if analysis—alternative delivery plans by estimating the impact of any ordertime/volume change on the plans of the downstream supply network members.

Compared with limited CPFR collaboration, full CPFR collaboration ischaracterized by an increased number of areas in which companies collaborate.The collaboration includes the synchronization and coordination of business,sales, and order forecast plans. For example, central company F collaborates withits suppliers in defining, as well as order forecast plans, joint business, and salesforecast plans.

It is worth noting that in two cases—B and C—the central company collaboratesdifferently in the upstream and downstream networks (see figure 2). For instance,central company C collaborates with the DCs on a limited collaboration level, whileits collaboration with the other production/packaging plants currently consistsin just the exchange of data/information on stock levels and available capacity(i.e. communication level).

Finally, the arrows in figure 2 highlight the fact that certain changes aretaking place (or going to take place) in the way central firms are implementing(or going to implement) CPFR. Central companies E and F intend to extendcollaboration to several customers and suppliers, respectively. Moreover, company Ais planning to align and synchronize its replenishment plans with those of theDCs, through an approach that is able to make its supply chain more responsive.

3216 P. Danese

This explains the arrow shifting the CPFR collaboration between company Aand its DCs from type 6 to type 4. Finally, until only a few years ago, centralcompany B simply communicated its raw-material order-forecast plans to a fewsuppliers (i.e. communication level). Recently, it has begun to collaborate througha ‘VMI min–max approach’ with its active agent supplier. In future years,company B intends to extend this type of collaboration to several other suppliers.

4.2 Relationships between CPFR collaborations and ICTs

From the cross-case analysis, it emerges that technologies used to support CPFRcollaboration can vary significantly. Some companies communicate via fax or email,while others exchange data electronically (e.g. via Intranet or Extranet). CompaniesE and F use Internet-based CPFR solutions for exchanging data and managingsales/order forecast exceptions. Finally, in some cases, analytical tools, such as APSsolutions, are adopted to facilitate decision-making in solving order-forecastexceptions.

Based on collected data and CPFR technology literature (Helms et al. 2000,ECR Europe 2001, 2002, Frankel et al. 2002, Seifert 2003), the ICTs supportingCPFR may be classified into four groups forming an ordinal scale of sophistication.The sophistication of each group of ICTs is related to its capability of ensuringeffective coordination in performing CPFR and can be evaluated by consideringits capability of supporting fast and accurate data/information exchange and CPFRcollaboration (e.g. event management).

The first group includes non-sophisticated tools used for data/informationexchange. For instance, data/information can be sent via fax and/or email. Thesecond encompasses ICTs supporting electronic data exchange and its integrationinto companies’ local systems. The third includes APS tools, as well as tools allowingcompanies to exchange data electronically. Finally, the fourth group involvesInternet-based CPFR solutions that are able to support electronic data exchange,monitor and alert CPFR participants when an exception occurs, and supportcollaboration during all CPFR phases. The first group represents the leastsophisticated ICTs, while Internet-based CPFR solutions are the most sophisticated.

By classifying the ICTs adopted in the analysed cases into the groupsdescribed above while simultaneously considering the types of CPFR collaboration(cf. previous section), an interesting relation seems to emerge (figure 3).

Figure 3 shows that when a company interacts with only a few partners, it is notnecessary to implement sophisticated ICTs to support CPFR. Companies D and G,for example, communicate with their customers via fax and email; similarly,company B does so with its suppliers.

In a different way, when the number of interacting units is high, moresophisticated technologies have to be adopted to manage the CPFR collaborationeffectively. Interestingly enough, the level of sophistication seems to depend on thedepth of the collaboration.

When companies collaborate with several partners, and CPFR collaboration islimited to data communication (i.e. communication level), ICTs that allowelectronic data/information exchange and its integration into companies’ localsystems are adopted. Adoption of these types of ICTs becomes essentialto guaranteeing the speed of information transfer and accuracy of information.

Collaborative Planning, Forecasting, and Replenishment 3217

Indeed, when a company interacts with several units, data exchange via fax and/oremail does not guarantee the accurate and timely updating of data. In fact, personnelwould have to devote considerable time to interpreting files or faxes received andto inserting data into the company’s local system. For these reasons, company Ccommunicates with subcontractors via Extranet and with MUs owned by thepharmaceutical group via Intranet. Data and information can be extracted fromand imported into local systems. Similarly, company A can read and extractDC stock and sales data thanks to the Data Import module included in theDistribution Planning Model (DPM) software.

When companies collaborate on a limited collaboration level, and thenumber of interacting units is high, the sophistication of the adopted ICTs increases.In fact, as well as using tools to electronically exchange data/informationand integrate it in companies’ information systems, companies also tend to adoptAPS tools to support the collaboration (e.g. to manage order-forecast exceptions).For instance, in the collaboration between company B and its DCs, stock-leveldata and sales forecast plans are used by the central planning system to elaboratethe Distribution Requirements Planning (DRP). The software adopted isManugistics. Company B can access DRP results via Intranet. Moreover, to solveorder-forecast exceptions, it can use Cyberplan, an APS tool produced by Cybertec,allowing planners to choose alternative replenishment plans when an additionalproduct quantity is required by one or more DCs during the frozen period. Similarly,company C accesses DC data and DRP plans elaborated by the central systemvia Intranet and uses an APS tool to solve exceptions. It is also worth noting that

TYPE 5

� Company B – suppliers (FAX, EMAIL)� Company D – customers (FAX, EMAIL)

Lim

ited

Col

labo

ratio

nFu

llC

olla

bora

tion

DEPT

H O

F CO

LLAB

ORA

TIO

N

TYPE 4

� Company B - DCs (ELECTRONIC DATAEXCHANGE AND DATA INTEGRATION + APS)

� Company B – suppliers (ELECTRONIC DATAEXCHANGE AND DATA INTEGRATION + APS)

� Company C – DCs (ELECTRONIC DATAEXCHANGE AND DATA INTEGRATION + APS)

� Company A – DCs (ELECTRONIC DATAEXCHANGE AND INTEGRATION + APS)

TYPE 2

� Company E – customers (INTERNET-BASEDCPFR SOLUTION)

� Company F – suppliers (INTERNET-BASEDCPFR SOLUTION)

TYPE 3

� Company G – customers (FAX, EMAIL)

TYPE 1

� Company E – customers (FAX, EMAIL; TESTINGOF INTERNET-BASED CPFR SOLUTION)

� Company F – suppliers (FAX, EMAIL; TESTINGOF INTERNET-BASED CPFR SOLUTION)

Com

mun

icat

ion TYPE 6

� Company A – DCs (ELECTRONIC DATAEXCHANGE AND INTEGRATION)

� Company C - subcontractors/MU (ELECTRONICDATA EXCHANGE AND INTEGRATION)

Low High

NUMBER OF INTERACTING UNITS

Figure 3. Relationships between types of CPFR collaboration and ICTs.

3218 P. Danese

in the future, company A intends to collaborate with its local DCs using a

‘VMI min–max’ approach (see arrow in figure 3 shifting CPFR collaboration

between company A and its DCs from type 6 to type 4). In this case, the DPM will be

useful to solve order-forecast exceptions through what-if analyses. Company B is

also testing the possibility of collaborating with its suppliers using a ‘VMI min–max’

approach (cf. figure 3). Managers intend to exchange data/information electronically

with independent and owned suppliers via Extranet and Intranet, respectively.

Moreover, an APS tool will be used to solve order-forecast exceptions. It will allow

companies to establish the final order-forecast plans taking into account the order

status of the active agent and packaging suppliers.Finally, when companies collaborate on a full collaboration level, and the

number of interacting units is high, Internet-based CPFR solutions are adopted

to exchange data/information and support collaboration in defining business plans,

and sales and order forecasts. Company E does not currently use sophisticated

ICTs to collaborate with its customers. However, given that, in the future, it intends

to extend collaboration to several partners, it is testing the Internet-based CPFR

solution powered by Syncra Systems that will be useful to manage collaboration

with several companies. Similarly, until a few years ago, company F exchanged

data/information with its suppliers via email. Today, as it is company F’s intention

to involve several other suppliers in the CPFR collaboration, CPFR implementation

is based on exploitation of technology offered by the WWRE e-exchange. Among

services offered by WWRE, the Internet-based Collaborative Planner solution

is used by company F and its suppliers to manage the entire CPFR process.

Company F’s supply-chain manager states:

[. . .] WWRE gives companies the opportunity to develop and use standards

to enhance communication and thus collaboration. Standards allow the

suppliers to see customers’ data/information and plans in a standard format

(and vice versa). Moreover, the Collaborative Planner solution gives the

possibility to dramatically reduce the time spent in solving sales/order forecast

exceptions.

Thus, I can advance the following proposition.

Proposition 1: Adoption of the ICTs supporting CPFR varies across the CPFR

collaboration types.

. Proposition 1a: When the number of interacting units is low, low sophisticated

ICTs, such as fax or email, are adopted to support CPFR collaboration,

independently of the depth of collaboration;. Proposition 1b: When the number of interacting units is high, more

sophisticated ICTs are adopted to support CPFR collaboration. Their degree

of sophistication depends on the depth of collaboration: as the depth of

collaboration moves away from communication towards limited and full

collaboration, ICT sophistication increases.

Figure 4 offers a graphical representation of the relationship between ICTs and types

of CPFR collaboration synthesized in Proposition 1.

Collaborative Planning, Forecasting, and Replenishment 3219

4.3 Relationships between CPFR collaborations and liaison devices

Cross-case analysis demonstrates that the mix of liaison devices a company canput into practice to manage inter-company relationships can vary significantly.Figure 5 summarizes the types of liaison devices adopted to support CPFRcollaboration in each of the analysed cases and, by simultaneously consideringthe adopted liaison devices and types of CPFR collaboration, supports the existence

Fax, email

Electronic dataexchange and data

integration

Electronic dataexchange and dataintegration + APS

Internet-based CPFRsolutions

Degree ofsophistication

Degree ofsophistication

Communication Limitedcollaboration

Fullcollaboration

Low number of interacting units

Fax, email

Electronic dataexchange and data

integration

Electronic dataexchange and dataintegration + APS

Internet-based CPFRsolutions

High number of interacting units

Depth ofcollaboration

Communication Limitedcollaboration

Fullcollaboration

Depth ofcollaboration

Figure 4. Relationship between ICT degree of sophistication, depth of collaboration, andnumber of interacting units.

3220 P. Danese

of a relationship between these two variables. In fact, if no relationship existed,

one would expect an equal dispersion of the different mixes of liaison devices within

the matrix.This research result can be explained by considering a concept widely recognized

in organizational literature (Galbraith 1973, Nadler and Tushman 1987, Gupta

and Govindarajan 1991). It states that the liaison devices can be seen as formingan ordinal scale of complexity. The more complex the liaison devices, the greater

its information-processing capability, and thus its capability to ensure effective

coordination among interdependent units. In particular, the adoption of theintegrating managers, task forces/standing committees, and liaison positions

provides for a high, medium, and low information-processing capability, respec-

tively, as the degree of complexity differs. As the types of CPFR collaborationare different because of the number of interacting units and depth of collaboration,

it is logical to expect that each type of CPFR collaboration requires a different

coordination capability and, as a result, that the adopted liaison devices vary acrossthe types of CPFR collaboration.

DE

PT

H O

F C

OL

LA

BO

RA

TIO

N

TYPE 4

� Company B - DCs (MEETING +INTEGRATING MANAGER + LIAISONPOSITION)

� Company B – suppliers (MEETING +INTEGRATING MANAGER + LIAISONPOSITION)

� Company C – DCs (MEETING +INTEGRATING MANAGER + LIAISONPOSITION)

� Company A – DCs (MEETING +INTEGRATING MANAGER + LIAISONPOSITION)

TYPE 2

� Company E – customers (MEETING +INTEGRATING MANAGER + LIAISONPOSITION)

� Company F – suppliers (MEETING +INTEGRATING MANAGER + LIAISONPOSITION)

TYPE 3

� Company G – customers (MEETING +LIAISON POSITION )

TYPE 1

� Company E – customers(MEETING + LIAISON POSITION )

� Company F – suppliers (MEETING + LIAISON POSITION )

TYPE 6

� Company A – DCs (LIAISONPOSITION)

� Company C - subcontractors/MU(LIAISON POSITION)

Low High

NUMBER OF INTERACTING UNITS

TYPE 5

� Company B - suppliers (LIAISONPOSITION)

� Company D - customers (LIAISONPOSITION)

Lim

ited

Col

labo

ratio

nF

ull

Col

labo

ratio

nC

omm

unic

atio

n

Figure 5. Relationships between types of CPFR collaboration and liaison devices.

Collaborative Planning, Forecasting, and Replenishment 3221

In particular, figure 5 highlights that when two or more companies communi-cate only (type 5 and 6 CPFR collaborations), liaison positions are adoptedas coordination mechanisms. Indeed, it is necessary to entrust a person or a groupof people with the issue of managing data exchanges among units participatingin the CPFR stages. As this person has in fact no formal authority, they representa liaison position, usually with the task of collecting/organizing information andmanaging inter-firm relationships. For example, in case D when different customerscommunicate their order forecast plans to the MU, a person is responsiblefor collecting data by customers and gathering documents and files. Similarly,in company C, planners are responsible for guaranteeing the accurate exchangeof data on time with the subcontractors and other MUs.

In a different way, in the case of companies collaborating on a limitedor full-depth collaboration level, as well as liaison positions, widely adoptedcoordination mechanisms are meetings involving members from the interactingunits. Meetings are defined by Mintzberg (1979, p. 163) as ‘the prime vehicle usedin the organization to facilitate mutual adjustment’. In CPFR collaboration type 1,2, 3, and 4, meetings are adopted to exchange data and information or to discussplans when exceptions occur (e.g. when a DC asks company B for an additionalproduct quantity during a frozen period or when a customer requires company Gto rush a delivery). The use of meetings to achieve coordination in the case of limitedor full collaboration derives from the necessity to manage a two-way interactionthat implies the exchange of data, information, and knowledge, rather than a simpleone-way communication, on the basis of which final order/sales forecast plansare defined.

In CPFR collaboration types 2 and 4, as the number of involved units increases,as well as meetings, integrating managers are useful to coordinate units. In fact, thesetypes of CPFR collaboration represent complex situations in which the organizationof meetings allows managers to discuss order/sales forecast plans, and contributesto solving exceptions by maintaining an open and encouraging atmosphere sothat conflicts are neither intentionally avoided nor resolved through the use of forceby one side. Smoothing over conflicts is in fact ineffective because it may leavediscord in people’s minds and thus undermine the quality of the relationship.However, integrating managers are also needed because when the amount of contactincreases, it may be useful to entrust a person with the responsibility of managingthe CPFR collaboration. For example, when company C’s managers exchangedata/information and discuss order-forecast plans with the DCs, the productmanager, as the integrating manager, has the authority and influence to be ableto establish the final forecast, should there be any conflict. Moreover, in the future,when companies E and F collaborate with several customers and suppliers,respectively, the Customer Supply-chain manager in company E and the categorymanagers in company F will be responsible for directing meetings and determiningfinal sales/order forecast plans (see arrows shifting CPFR collaborations betweencompany E and its customers and company F and its suppliers from type 1 totype 2 in figure 5).

Finally, it is worth noting that in two other cases, the conditions of equilibriumare changing. Companies A and B are planning to modify their way of collaboratingand, as a result, the mix of liaison devices adopted to assure coordination (see arrowsin figure 5). Company B currently communicates its order-forecast plans to only

3222 P. Danese

a few suppliers, but in the future it intends to collaborate with several suppliers undera limited collaboration approach. Purchasing managers have the role of liaisonpositions, guaranteeing the accurate exchange of order-forecast plans. At present,company B only collaborates with the active agent supplier under a ‘VMI min–max’approach (i.e. limited collaboration level). In this case, meetings are also important

Liaisonpositions

Meetings

Integratingmanagers

Degree ofcomplexity

Low High

Depth of collaboration at acommunication level

Liaisonpositions

Meetings

Integratingmanagers

Degree ofcomplexity Depth of collaboration at a

limited/full collaborationlevel

Number ofinteracting units

Low High Number ofinteracting units

Figure 6. Relationship between liaison device degree of complexity, depth of collaboration,and number of interacting units.

Collaborative Planning, Forecasting, and Replenishment 3223

to manage order-forecast exceptions. In the future, when company B will collaborateon a limited collaboration level with several suppliers, an integrating managerwill be required to solve exceptions. Similarly, company A has planned to collaborateon a limited collaboration level with its DCs. Managers agree that meetings andintegrating managers will be useful to coordinate units.

Results of the relationship between CPFR collaborations and liaison devicescan be summarized in the following proposition:

Proposition 2: The adoption of liaison devices supporting CPFR varies across theCPFR collaboration types.

. Proposition 2a: When the depth of collaboration is at a communicationlevel, low-complexity liaison devices, such as liaison positions, are adoptedto support CPFR collaboration, independently of the number of interactingunits.

. Proposition 2b: When the depth of collaboration is at a limited/fullcollaboration level, more complex liaison devices are adopted to supportCPFR collaboration. Their degree of complexity depends on the numberof interacting units: when it is low, medium-complexity liaison devices areadopted; when it is high, high-complexity liaison devices are adopted.

Figure 6 provides a graphic illustration of the relationship summarized inProposition 2.

5. Conclusions

Using data from seven case studies of supply networks, this research investigatesdifferences in CPFR implementation with regard to the type of collaboration putinto practice, as well as the ICTs and liaison devices adopted to perform CPFR.Moreover, this study accounts for differences in ICT and liaison device adoptionby referring to the different types of CPFR collaboration (characterized by dissimilardepths of collaboration and numbers of interacting units). These research findingsmay offer an original contribution to the debate on CPFR from both academicand managerial perspectives.

From an academic point of view, this research might contribute to advancingtheory on CPFR in two ways: (1) by proposing a CPFR model, and thus byidentifying the macro-building blocks upon which CPFR is based and by seekingto establish relationships between them, and (2) by explaining the rationalebehind the managerial choices that lead companies to implement different CPFRconfigurations.

Most of the studies on CPFR focus on identifying the advantages andfuture developments of CPFR or explaining CPFR implementation phases.In doing so, few authors devote their attention to rigorously defining CPFRvariables or proposing relationships among variables and measures. This research,instead, aims to clarify differences in CPFR by analysing the different formsof CPFR collaborations, as well as the ICTs and liaison devices used to supportCPFR, thus contributing to the definition of measures for these variables.Moreover, it seeks to explain these differences by looking at the relationships

3224 P. Danese

between the different types of CPFR collaboration, ICTs and liaison devices, thuscontributing to the advance of substantive research on CPFR.

In addition, research is still at an early stage of investigating the reasons behindthe existence of different CPFR configurations. In the literature on CPFR, a theorywidely diffused to explain differences in ICT adoption states that the implementationoptions vary according to the number of interacting units (ECR Europe 2001, 2002,Seifert 2003). For instance, Seifert (2003) maintains that newer web-based collab-oration applications have been designed to allow the sharing of processes andinformation among multiple trading partners. There is no reason to disagree withthis theory. Nevertheless, from this research, it emerges that another variableis important when it comes to explaining different ICT options, such as the depthof collaboration, depending on the number of business processes involved in thecollaboration and the level of integration.

The theoretical development presented here also has interesting practicalimplications as it may help managers in the decision-making process to select themost appropriate action to be taken to implement CPFR.

In particular, one simple but important implication of this research is thatCPFR is characterized by two dimensions—one technical (the ICTs), the otherorganizational (the liaison devices). Given the dominant technical inclinationof most professionals in the CPFR field, this study suggests that one danger mightbe that technical solutions are prescribed for companies when organizationalsolutions are needed. As notably argued by Van Dierdonck and Miller (1980), pastexperience in production planning and control systems adoption suggests that thismight lead to several problems.

A second implication of this research is that it provides managers with aframework for anticipating changes in ICT and liaison device adoption, as theyanticipate changes in CPFR collaboration. The relationships that have emergedfrom this study (cf. propositions 1 and 2) are in fact dynamic given that links betweenthe CPFR collaborations, ICTs, and liaison devices can not only vary acrosscases but also change over time within each case.

However, although research findings are encouraging, the opportunity to usethe relationships found in this research as a managerial tool calls for the testingof research findings within larger samples of firms, representative of a broader rangeof industries. The analysed case studies are in fact limited to a relatively smallsample and only a few industries. Future research should evaluate a wider sampleof networks involving both large and small companies in several industries viacase-based and/or survey-based research designs.

References

Attaran, M. and Attaran, S., Collaborative computing technology: the hot new managingtool. J. Manage. Dev., 2002a, 21, 598–609.

Attaran, M. and Attaran, S., Collaborative computing technology: the hot new managingtool. Team Perform. Manage. Int. J., 2002b, 8, 13–20.

Aviv, Y., The effect of Collaborative Forecasting on Supply Chain Performance. Manage.Sci., 2001, 47, 1326–1343.

Barratt, M.A. and Oliveira, A., Exploring the experiences of collaborative planning initiatives.Int. J. Phys. Distrib. Logist. Manage., 2001, 31, 266–289.

Collaborative Planning, Forecasting, and Replenishment 3225

Cooke, J.A., VMI: Very mixed impact?. Logist. Manage. Distrib. Rep., 1998, 37, 51–54.Crum, C. and Palmatier, G.E., Demand collaboration: what’s holding us back. Supply Chain

Manage. Rev., 2004, January/February, 54–61.ECR Europe. A Guide to CPFR Implementation (ECR Europe facilitated by Accenture), 2001.ECR Europe. European CPFR Insights (ECR Europe facilitated by Accenture), 2002.Frankel, R., Goldsby, T.J. and Whipple, J.M., Grocery Industry Collaboration in the wake

of ECR. Int. J. Logist. Manage., 2002, 13, 57–72.Galbraith, J.R., Designing Complex Organizations, 1973 (Addison-Wesley: Reading, MA).Gupta, A.K. and Govindarajan, V., Knowledge flows and the structure of control within

multinational corporations. Acad. Manage., 1991, 16, 768–792.Harrington, L., 9 Steps to success with CPFR. Transport. Distrib., 2003, April, 50–52.Helms, M.M., Ettkin, L.P. and Chapman, S., Supply chain forecasting: Collaborative

forecasting supports supply chain management. Bus. Process Manage. J., 2000, 6,392–407.

Hill, S., CPFR builds the united partnerships of apparel. Apparel Ind., 1999, 60, 54–58.Larsen, T.S., Thernøe, C. and Andresen, C., Supply chain collaboration: theoretical

perspective and empirical evidence. Int. J. Phys. Distrib. Logist. Manage., 2003, 33,531–549.

McCutcheon, D.M and Meredith, J.R., Conducting case study research in operationsmanagement. J. Oper. Manage., 1993, 11, 239–256.

Mintzberg, H., The Structuring of Organizations, 1979 (New York: Prentice-Hall).Nadler, D.A. and Tushman, M.L., Strategic Organization Design, 1987 (IL: Glenview:

Foresman).Parks, L., CPFR programs facilitate inventory management. Drug Store News, 1999, 21, 27.Parks, L., Wal-Mart gets onboard early with collaborating planning. Drug Store News, 2001,

23, 14–15.Schachtman, N., Trading Partners Collaborate to Increase Sales. Inform. Week, 2000,

807, 182–185.Seifert, D., Collaborative Planning, Forecasting and Replenishment: How to Create a Supply

Chain Advantage, 2003 (AMACOM: New York).Songini, M., Pharmavite to launch collaborative system with Kmart. Comput. World, 2001,

35, 7–9.Sparks, L. and Wagner, B.A., Retail exchanges: a research agenda. Supply Chain Manage.

Int. J., 2003, 8, 17–25.Van Dierdonck, R. and Miller, J. G., Designing production planning and control systems. J.

Oper. Manag., 1980, 1, 37–46.VICS, 1998. CPFR Guidelines. Available online at: www.cpfr.org (accessed February 2005).

3226 P. Danese