new approaches to mass customization: configurations and …directory.umm.ac.id/journals/journal of...

Ž .Journal of Operations Management 18 2000 605–625www.elsevier.comrlocaterdsw

Approaches to mass customization: configurations andempirical validation

Rebecca Duray a,), Peter T. Ward b, Glenn W. Milligan b, William L. Berry b

a College of Business and Administration, UniÕersity of Colorado at Colorado Springs, Colorado Springs, CO 80933-7150, USAb Department of Management Sciences, Fisher College of Business, The Ohio State UniÕersity, Columbus, OH 43210-1399, USA

Abstract

Mass customization is a paradox-breaking manufacturing reality that combines the unique products of craft manufactur-ing with the cost-efficient manufacturing methods of mass production. Although this phenomenon is known to exist inpractice, academic research has not adequately investigated this new form of competition. In this research, we develop aconfigurational model for classifying mass customizers based on customer involvement in design and product modularity.We validate this typology through an empirical analysis and classification of 126 mass customizers. We also exploremanufacturing systems and performance implications of the various mass customization configurations. q 2000 ElsevierScience B.V. All rights reserved.

Keywords: Mass customization; Process design; Technology; Marketingroperations interface

1. Introduction

Mass customization, once considered a paradox tobe resolved in the future, has become an everydayreality for many manufacturers. Stanley Davis coinedthe term Amass customizationB in his 1987 book,Future Perfect. Davis suggested that existing tech-nology constrained possibilities for mass-customizedproducts, markets, and organizations, although hesaid that the phenomenon would prevail in the fu-ture. More contemporary researches suggest that theadvances in manufacturing, information technology

) Corresponding author. Tel.: q1-719-262-3673.Ž .E-mail address: [email protected] R. Duray .

and management methods since the publication ofFuture Perfect in 1987 have made mass customiza-

Žtion a standard business practice Kotha, 1995; Pine,.1993 . The confluence of these advances allows

producers to customize at low cost and customers toreap the benefits of customized products with rela-tively low prices.

The practice of mass customization does not fitthe conventional paradigm of manufacturing man-agement. Historically, companies chose processesthat supported the production of either customizedcrafted products or standardized mass-producedproducts. This traditional practice means that cus-tomized products usually are made using low volumeproduction processes that cope well with a greatvariety of products and with design processes that

0272-6963r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.Ž .PII: S0272-6963 00 00043-7

( )R. Duray et al.rJournal of Operations Management 18 2000 605–625606

can accommodate a high degree of customer in-volvement in specifying the product. In contrast, amass production process is chosen for making stan-dardized products in a high volume environmentwhere great attention is paid to efficiency and captur-ing scale economies. Further, product variety is rela-tively low in mass production and customer involve-ment is sought through market research only tocapture standard product design attributes that havewide appeal. In contrast to the traditional paradigm,

Ž .Davis 1987 envisioned a one-of-a-kind productmanufactured to customer specification without sac-rificing scale economies. In this way, customers areable to purchase a customized product for the cost of

Ž .a mass-produced item. Similarly, Pine 1993 definesthe goals of mass customization as providing enoughvariety in products and services so that nearly every-one finds exactly what they want at a reasonableprice.

Although these definitions provided by Davis andPine sketch the essence of mass customization, theydo not possess the specificity required to identifycompanies as mass customizers or how a companycan achieve a mass customization capability. Ac-counts of mass customization practices in companiesdescribed in the literature label a broad range ofproduction practices mass customization; however,the diversity of the practices and the companiesfurther clouds the meaning of mass customization. Inshort, extant literature has not established good con-ceptual boundaries for mass customization, nor hasthat literature presented a means to distinguish amongthe vast array of mass customization practices in away that lends clarity.

This paper addresses three important elementsmissing from the literature. First, we develop a con-ceptual model of mass customization to identify andclassify mass customizers. This model is based onthe key dimensions of mass customization and vali-dated through literature, field studies, and surveytesting. Second, we develop a classification schemeto group mass customizing companies according tothe way they achieve mass customization. Third, weexplore different approaches to mass customizationimplied by the typology by comparing the manufac-turing approach of each type. Our approach yieldsmass customization configurations that are empiri-cally validated.

We establish the external validity of the modelthrough empirical investigation of companies in sixdifferent industries. By using a number of industriesin our sample, we address the issue of whether masscustomization is a robust concept applicable across arange of industries, or whether it can be applied onlyto a limited number of special cases. On the otherhand, by limiting the study to six industries, we areable to show that mass customization is a competi-tive choice open to a number of competitors in thesame industries. In addition, we are able to controland test for industry effects.

2. Research proposition

Because mass customization is a relatively newidea; scholarly literature related to the topic is scant.In this research, we seek to uncover the importantdimensions of mass customization from an opera-tions perspective. We argue that the essence of masscustomization lies in resolving the seeming paradoxof mass producing custom products by finding effi-ciencies in two key dimensions. First, mass cus-tomizers must find a means for including each cus-tomer’s specifications in the product design. Second,mass customizers must utilize modular design toachieve manufacturing efficiencies that approximatethose of standard mass produced products. Choicesmade by mass customizers on how they approachthese two dimensions suggest a useful typology ofmass customization:

Proposition 1. Mass customizers can be identifiedand classified based on two characteristics: the pointin the production cycle of customer inÕolÕement inspecifying the product and the type of product modu-larity employed. Each mass customization configura-tion exhibits a distinct approach to the manufactureof mass-customized products.

More specifically, we will suggest a classificationwith four fundamental, mutually exclusive types ofmass customizers. We argue that this typology willalso serve to identify manufacturers that do not masscustomize.

( )R. Duray et al.rJournal of Operations Management 18 2000 605–625 607

The underlying choices with respect to customerinvolvement in design and modularity type implydifferent approaches to manufacturing processes,policies, and technologies, thereby making the classi-fications useful to manufacturers. In other words,knowing the point at which a mass customizer in-volves the customer in product design and the ap-proach to modularity taken by the mass customizersuggests the configuration of processes and technolo-gies that will be used in designing and making themass-customized product. We concede that othercharacteristics such as flexibility, agility, or serviceapproach also play important roles in mass cus-tomization viewed from an operations perspective.We argue, however, that customer involvement andmodularity are the key elements in defining masscustomization approach. This paper develops andtests this argument in detail in the following manner.

To explore this proposition, we first develop thedimensions of mass customization and discuss theimplications of the mass customization configura-tions from a theoretical perspective. Then, the modelis operationalized through an empirical evaluation of194 manufacturing plants. These data are used toestablish the mass customization dimensions in prac-tice, classify actual plants, and explore these masscustomization configurations in the context of themanufacturing systems variables employed.

3. Mass customization dimensions

The boundaries of mass customization can bemore clearly established by delineating two issues:Ž . Ž .1 the basic nature of customization and 2 themeans for achieving customization at or near massproduction costs. The first issue, the nature of cus-

Ž .tomization, has been addressed by Mintzberg 1988 .A customized product is designed specifically tomeet the needs of a particular customer. Varietyprovides choice for customers, but not the ability tospecify the product. A great deal of variety in themarketplace may satisfy most customers and, hence,substitute for customization; but customization andvariety are distinct. For example, having hundreds ofvarieties of breakfast foods on the shelf of the super-market is different from being able to specify one’s

exact breakfast food formulation from the cerealsupplier. It is important to realize that the availabilityof hundreds of varieties probably limits the marketappeal of customized products for most customers.However, variety is not customization. As WomackŽ .1993 remarks, this distinction between customizedproducts and product variety is overlooked in theexamples of companies pursuing mass customization

Ž .by Pine 1993 . This distinction is important becauseit implies that customers must be involved in speci-fying the product.

The second issue that we delineate — the methodof achieving customization at or near mass produc-tion costs — addresses the AmassB in mass cus-tomization. How can unique products be developedand manufactured in a mass production fashion?How can high volume, low cost customization be

Ž .implemented? Pine 1993 argues that modularity isa key to achieving mass customization. Modularityprovides a means for the repetitive production ofcomponents. Modularity allows part of the product tobe made in volume as standard modules with productdistinctiveness achieved through combination ormodification of the modules. Therefore, modules thatwill be used in the custom product can be manufac-tured with mass production techniques. The fact thatparts or modules are standardized allows for mass-customized products to achieve the low cost andconsistent quality associated with repetitive manufac-turing. Thus, modularity can be viewed as the criticalaspect for gaining scale or AmassB in mass cus-tomization. We further develop the issues of cus-tomization and modularity to provide a basis for amore comprehensive definition of mass customiza-tion.

3.1. Customization issues

A customized product must be designed to cus-tomer specifications. From the literature, it is appar-ent that identifying the point of initial customerinvolvement is critical to determining the degree of

Ž .customization. Mintzberg 1988 and Lampel andŽ .Mintzberg 1996 developed the idea that the level of

customer involvement in the production cycle canplay a critical role in determining the degree of

Ž .customization. McCutcheon et al. 1994 argued that


the production stage where a product is differentiatedis a key variable in process choice decisions. Byextension, the point of customer involvement inspecifying the product also may be related to choicesabout the customization process. We argue that thepoint of customer involvement in the productioncycle is a key indicator of the degree or type ofcustomization provided. For purposes of definingmass customization, we take a narrow view of theproduction cycle. Specifically, we include four pointsin the production cycle: design, fabrication, assem-bly, and use. If customers are involved in the earlydesign stages of the production cycle, a productcould be highly customized. If customer preferencesare included only at the final assembly stages, thedegree of customization will be not as great. In thismanner, point of customer involvement provides apractical indicator of the relative degree of productcustomization.

Thus, we argue that products with early customerinvolvement are relatively more customized thanthose with later involvement. The typology of

Ž .Mintzberg 1988 supports this reasoning. Mintzbergviews customization as taking one of three forms:pure, tailored, or standardized. Each form differs inthe portion of the production cycle involved and thedegree of uniqueness of the product. A pure cus-tomization strategy furnishes products designed andproduced from scratch for each individual customer.Pure customization includes the customers in theentire cycle, from design through fabrication, assem-bly and delivery and it provides a highly customizedproduct. A tailored customization strategy requires abasic design that is altered to meet the specific needsof a particular customer. In this case, the customerenters the production cycle at the point of fabricationwhere standard products are modified. In a standard-ized customization strategy, a final product is assem-bled from a predetermined set of standard compo-nents. Here, the customer penetrates the assemblyand delivery processes through the selection of thedesired features from a list of standard options. The

Ž .categorization of Mintzberg 1988 shows that thetype of customization chosen by the producer im-plies different levels of customer involvement inproduct design and different points at which thatinvolvement begins. These different customizationstrategies also imply degrees of customization, with

pure customization providing the highest degree ofcustomization with all of the products designedspecifically for the customer, and standard cus-tomization the lowest degree with only an arrange-ment of components determining the customizedconfiguration.

3.2. Modularity issues

Mass customization requires that unique productsbe provided in a cost-effective manner by achievingvolume-related economies. A number of observerssuggest that modularity is the key to achieving low

Ž .cost customization. Pine 1993 stated that true masscustomization requires modularity in production, al-though he was not specific about where or how

Ž .modularity should be used. Baldwin and Clark 1994discussed modularity in production as a means topartition production to allow economies of scale and

Ž .scope Goldhar and Jelinek, 1983 across the productŽ .lines. McCutcheon et al. 1994 suggested that mod-

ular product design is the best way to provide varietyand speed, thereby alleviating the customization re-sponsiveness squeeze, which occurs when customersdemand greater variety, and reduced delivery timessimultaneously. A modular approach can reduce thevariety of components while offering a greater rangeof end products. Flexible manufacturing systemsŽ .FMS provide for lower cost customization throughthe use of some form of modularity in their design.In the design of products for FMS manufacture,program modules for different product characteristicsare used to achieve fast set-up manufacture. Theseprogram modules are used to repeat manufacturingsequences across products and provide for modular-ity in the design of new products. Therefore, FMSproduction contains modularity.

Ž .Similarly, Ulrich 1992 argued that modularitycan help increase product variety, but he also ad-dressed the use of modularity to shorten deliverylead times, and provide economies of scope. Pine et

Ž .al. 1995 asserted that to be successful, mass cus-tomizers must employ a productionrdelivery strat-egy that incorporates modularity into componentsand processes. In essence, the literature suggests thatmodularity can facilitate increasing the number ofproduct features available while also decreasing costs.


Therefore, it follows that the successful implementa-tion of mass customization requires effective use ofmodular product designs.

Modularity is multifaceted in concept and is gen-erally described either in relative terms or as a

Ž .typology. For example, Ulrich and Eppinger 1995

viewed modularity as a relative property with prod-ucts characterized as more or less modular in design.To better distinguish types of mass customizers, arange of modularity types should be considered.Modularity can take a number of forms. The varioustypes of modularity found in production environ-

Ž .Fig. 1. Modularity types Ulrich and Tung, 1991 .


Ž .ments were discussed in Pine 1993 , although hedoes not explicitly link modularity types with masscustomization. More recently, Ulrich and TungŽ .1991 developed a similar typology of modularity.Fig. 1 depicts these types of modularity.

Modularity can represent many forms of flexiblemanufacturing. For example, Levi Straus’ custom-fitjeans are made possible through their flexible manu-facturing process, which cuts each unique patternprior to stitching and sewing. In the Ulrich and Tungtypology, the Levi’s example can be described as aAcut-to-fitB modularity. The unique patterns are builtupon one traditional style of five-pocket jeans that isaltered or Acut-to-fitB the specific dimensions of thecustomer. These Amade-for-youB jeans are available

Ž .for alteration only within a limited size range 0–18 .The concept of modularity is a basic building blockin the manufacturing situations traditionally consid-ered to be flexible.

To make the concept of modularity operational,Ulrich and Tung’s typology is adopted and inte-grated into the framework of the production cycle, asseen in Fig. 2. Using the designrproduction processas a reference point, the different types of modularitycan be assigned to the phases of the product cycle.

For example, during the design and fabrication, mod-ules can be altered or components can be fabricatedto provide for the unique requirements of the cus-tomer. Cut-to-fit and component sharing modularityrequire that components are newly designed orchanged; therefore, these types of modularity musttake place during the design and fabrication stages.With cut-to-fit modularity, components are altered tothe physical dimensions specified by the customer.This alteration requires the fabrication of a compo-nent that is standard except in a specific dimension,e.g., length, that is specified by the customer. Thiscustomization necessarily takes place during the de-signrfabrication stages. In general, component shar-ing also takes place in the design and fabricationstages. Although a standard base unit is incorporatedinto the product, additional components are fabri-cated to provide an end-product that meets customerspecification. Modularity incorporated in the stan-dard base simplifies fabrication and reduces the totalcost of the customized product.

During the assembly and use stages, modules arearranged or combined according to customer specifi-cation, but components cannot be manufactured norcan modules be altered. Component swapping, sec-

Fig. 2. Customer involvement and modularity in the production cycle.


tional, mix, and bus modularity use standard mod-ules without alteration; therefore, these types ofmodules can be combined during the assembly anduse stage of the production cycle. In each case,standard modules are combined to form an end-prod-uct that is specified by the customer. In their pureform, component swapping, sectional, mix, and busmodularity all provide customization by allowingcustomers to specify a choice among a number ofstandard modules without the option of altering anyof the modules. In particular, sectional modularitycan also be used in the post-production phases wherethe customer combines components across manufac-

Ž .turers e.g., stereo components . Sectional modular-ity may require the adoption of uniform industry

Ž .standards Garud and Kumaraswamy, 1993 .When customer involvement in specifying the

product and modularity types are combined, masscustomization can be fully realized in practice. Cus-tomer involvement provides the customization whilethe modularity restricts the range of choice to de-crease the possible variety of components, thus al-lowing for repetitive manufacturer. When modularityis employed in mass-customized products, productdistinctiveness is a result of either the combinationof standard modules into a finite number of permuta-tions or the alteration of prescribed modules into alimited range of products. In contrast, purely cus-tomized products are infinite in permutations result-ing from craft manufacture. Modularity bounds thedegree of customization of the product and distin-guishes mass customization from pure customizedproducts. The fact that these parts or modules arestandardized allows for mass-customized products toachieve the low cost and consistent quality associ-ated with repetitive manufacturing.

4. Mass customization configurations

In Section 3, we argue that the model of masscustomization that emerges from the literature usestwo critical identifiers: customer involvement in theproduction cycle and modularity type. Bringing theseconcepts together, mass customization can be de-fined as building products to customer specificationsusing modular components to achieve economies of

scale. Distinctions can be made among mass cus-tomizers based on the point at which the customerbecomes involved in the design process and the typeof modularity employed by the producer. These twoattributes are interrelated and when taken togethersuggest mass customization archetypes. Fig. 3 showsthe dimensions juxtaposed, with point of customerinvolvement in design and type of modularity form-ing the archetypes.

4.1. Classification matrix

As shown in Fig. 3, mass customizers can beidentified and classified based on customer involve-ment and modularity type. Mintzberg’s definitions ofcustomization provide a good beginning for describ-ing the degree of customer involvement and can beseen down the left-hand column. When customersare involved at the design stage, products can bealtered to fit customers’ expectations with infinitevariety. In the fabrication stage, customer involve-ment means specifying relatively incremental changesto a standard design. In the assembly stage, customerrequirements must be met from a finite set of com-ponents. These two stages represent a time in theproduction cycle when customer preferences requirephysically altering existing components or cons-tructing unique components. Building further onMintzberg’s ideas, customer involvement in thepost-production or use stage should also be consid-ered. A product that can be adjusted or manipulatedby the consumer to provide customization at thepoint of delivery can also be considered mass-custo-

Ž . Ž .mized. Both Davis 1987 and Pine 1993 discussthis type of post-production customization which thatwe refer to as Apoint of sale customizationB.

Modularity provides the basis for repetitiveness inproduction or the AmassB in mass customization.

Ž .Baldwin and Clark 1994 use phases in a product’sdevelopment to circumscribe the type of modularityemployed. They argue that the type of modularitydiffers at different points in the production cycle.

Ž .The typology of Ulrich and Tung 1991 identifiestypes of modularity that can be employed that, bydefinition, fit the stages of the production processesof design and manufacturing. The model, depicted inFig. 3, overlays Baldwin and Clark’s production


Fig. 3. Matrix grouping of mass customization configurations.

phases with Ulrich and Tung’s modularity types.Modularity is addressed across the top row of themodel. When modules are designed to provide theability to modify components, modularity will beutilized in the design and fabrication stages of theproduction cycle. In the later stages of the productioncycle, assembly and use, modules are added or inter-changed, but not altered.

4.2. Archetypes

The juxtaposition of customer involvement andmodularity create four groups or mass customizationtypes. Group 1 includes both the customer involve-ment and modularity occurring during the design andfabrication stages. Since in this instance both the

customer involvement and modularity require fabri-cating a customized component, we name this groupthe Fabricators. Fabricators involve the customersearly in the process when unique designs can berealized or major revisions can be made in theproducts. Fabricators closely resemble a pure cus-tomization strategy, but employs modularity to gaincommonality of components. An example of a Fab-ricators is Bally Engineered Structures, a manufac-turer of walk-in coolers, refrigerated rooms, and

Ž .clean rooms described by Pine et al. 1993 . Productmodules are cut-to-fit specific dimensions of thecustomer, providing unique rooms manufacturedfrom modular components. Modular components arealtered in fabrication to AfitB the specific building. Inaddition, unique components may be designed for


specific application. Customers are involved in thedesign and fabrication stage of the production cycle,and component sharing and cut-to-fit modularity areused to provide the mass-customized product.

Group 2 incorporates customer involvement inproduct design during the design and fabricationstages but uses modularity during the assembly anddelivery stages. Because customer involvement pre-cedes the use of modularity, we refer to this group asInÕolÕers. With InÕolÕers, customers are involvedearly in the process although no new modules arefabricated for this customer. Customization isachieved by combining standard models to meet thespecification of the customer. Perhaps, the earlyinvolvement of the customer imbues the customerwith a greater sense of customization or ownershipof the product design, although no customized com-ponents are fabricated. Because they do not fabricatecustomized components to customer specification,InÕolÕers capture greater economies of scale thanFabricators while maintaining a high level of cus-tomer involvement. An example of this type of masscustomizer is Andersen Windows. Andersen uses adesign tool that helps customers develop the specificdesign of their windows. However, products areproduced from 50,000 possible window componentsŽ .Pine et al., 1995 . Components are not designed orfabricated for the specific application. However, cus-tomers specifications are AdesignedB and then, thecomponents are selected by the manufacturer to fitthis design. The sheer number of components pro-hibits the customer from simply choosing from aprescribed list, as with component swapping modu-larity. The customer is involved in the specificationduring the design and fabrication stages, but theproduct is assembled from modular components inthe assembly and use stages of the production cycle.

Group 3 involves the customer during assemblyand delivery but incorporates modularity in the de-sign and fabrication stages. Group 3, which we callModularizers, develops a modular approach in thedesign and fabrication stages, although customers donot specify their unique requirements until the as-sembly and use stage. Modularizers use modula-rity earlier in the manufacturing process than whencustomization occurs. This modularity may beconsidered component commonality. In this type,Modularizers may not gain maximum customization

advantages from modularity. For example, a masscustomizing upholstered furniture manufacturer usesmodularity in the design of a sofa frame which is

Ž .used in many product lines component sharing .This modularity provides for component commonal-ity, but is not used for customization. In the assem-bly stage, a customer chooses a fabric or wood finish

Ž .from a prescribed list component swapping , provid-ing some degree of customization. Modularizers in-corporate both customizable modularity in the laterstages of the production cycle and non-customizablemodularity in the design and fabrication stages of theproduction cycle.

Group 4 brings both customer involvement andmodularity to bear in the assembly and use stages.We call this group Assemblers. Assemblers providemass customization by using modular components topresent a wide range of choices to the customer.Assemble-to-order manufactures can be consideredmass customizers if customers specify products froma pre-determined set of features. Assemblers moreclosely resemble the operations of mass productionthan the other configurations of mass customers.Assemblers differ from mass producers in that theproducts have been designed so that the customercan be involved in specifying the product. Becausethe range of choices made available by Assemblersis large relative to mass producers, customers per-ceive the product to be customized. Motorola pagers,a recognized leader in mass customization, can beconsidered a Amass standardB customizer. Pagers canbe designed to a customer’s specification from awide range of options that are added at the produc-

Ž .tion phase Pine, 1993; Donlon, 1993 .In short, mass customizers can be typed on the

basis of two key dimensions: modularity and cus-tomer involvement. These same dimensions can beused to identify companies that do not possess masscustomization capabilities. Manufacturers that do notinvolve the customer in the design process or do notemploy modularity should not be considered masscustomizers. Without some degree of customer in-volvement in the design process, a product cannot beconsidered as customized. Companies that do in-volve the customer in the design process, but do notexhibit modularity in manufacturing, are also ex-cluded. These manufacturers should be consideredthe traditional customizers, a producer of one–off


goods without the economies of repetitive manufac-turing.

It is interesting to note that applying this typologyallows us to conclude that some widely cited exam-ples of mass customization do not fit the bill. For

Ž .example, Davis 1987 used Cabbage Patch dolls asan example of gaining a competitive edge throughmass customization. While it is true that each doll isa unique end-item which customers select at a retailoutlet, the customer does not participate in the designof the doll. To be mass-customized, the producerwould have to offer customers a means of having adoll made to their specifications, such as eye or hair

Ž .color. Similarly, Pine 1993 used Swatch watches asan example of mass customization. Swatch offerscustomers an extraordinarily wide selection of prod-ucts. Although this provides great variety, customersdo not have the ability to specify the design in anyway; therefore, this example misses the mark. Theseexamples illustrate that the two-dimensional opera-tional definition of mass customization lends clarityto the more casual definitions found in the literature.

5. Empirical validation

The conceptual typology presented above has beenvalidated using both secondary and primary data andboth case studies and surveys. In using multiple

Ž .methods, we follow the advice of Harrigan 1983who argued for using multiple research methodolo-

Ž .gies or granularities for testing business strategymodels. Case studies provide the ability to capturenuances of a company’s strategy, while surveys oflarger samples allow more confidence in generalconclusions.

The typology validation process itself provides forŽ .three levels of granularity Harrigan, 1983 . Initially,

the model was validated using case examples frompractitioner journals which discuss mass customiza-tion examples. Fifteen companies were identifiedthat appear, based on the information provided in theliterature, to exhibit mass customization character-istics. The next level of validation included inter-views with managers from companies randomly se-lected from an APICS directory to determine theextent of product customization. Based on the infor-mation obtained in the interview, more than half of

the companies selected at random could be consid-ered mass customizers. These interviews suggestedthat mass customization might be practiced in someform by a fairly large portion of companies drawn atrandom. A more intensive study of mass customiza-tion, including plant visits and phone interviews, wasthen conducted in the furniture industry. The furni-ture industry was selected for plant visits since thisindustry has traditionally provided customization ofend-products. This more in-depth study of furnituremanufacturers was used to better illustrate mass cus-tomization characteristics prior to the survey, as wellas validate the survey instrument. The third level ofvalidation included a survey of 639 companies inindustries anticipated to include mass customizersbased on evidence from the literature. This three-levelvalidation process provides both coarse- and fine-grained looks at mass customization configurationsto substantiate the conceptual model. In this paper,we concentrate on the model validation through thesurvey and use the data collected from the 194respondents to validate the model.

5.1. SurÕey methods

The sample was drawn from the Society of Manu-Ž .facturing Engineers SME membership data. Re-

Žspondents were selected based on title Vice Presi-dent of Manufacturing, Manufacturing Manager,

.Plant Manager to assure that respondents repre-sented a high level of responsibility. Executives inthe sample were selected randomly but their compa-

Ž .nies were limited by size more than 50 employees ,Žindustry furniture and fixtures, fabricated metal

products, machinery except electrical, electric andelectronic equipment, transportation equipment, and

. Žinstruments and related products and geography In-diana, Massachusetts, Michigan, North Carolina,

.Ohio, and Pennsylvania . Size and geographic limita-tions were imposed in the interest of homogeneityand efficiency. Because we seek to understand masscustomization practice rather describe its prolifera-tion, industries were selected on the basis of externalevidence suggesting that mass customization wasfairly common in these industries. Using the methods

Ž .suggested by Dillman 1978 and Salant and DillmanŽ .1994 , the questionnaire was sent to 639 plants with

Ž .194 responding 30.4% .


To determine if the respondents differed signifi-cantly from those that did not respond, the jobclassification of the respondents and total samplewere compared. The portion of respondents in eachcategory was compared to the expected number ofrespondents based on the percentage of each cate-gory represented in the total sample. A Chi-squaredtest of the expected and actual number of respondentwas not significant at the level of as0.05. Thisfinding supports the assertion that there was nosystematic difference between those companies re-sponding to the survey and those that did not.

In addition, a similar test was performed usingindustry classification. The portion of respondents ineach category was compared to the expected numberof respondents based on the percentage of eachcategory represented in the total sample. A Chi-squared test of the expected and actual number ofrespondent was significant at the level of as0.05,indicating that respondents may differ from non-re-spondents in industry representation. This result isnot surprising as the questionnaire was directed atplants producing customized products and standardproduct manufacturers may have been reluctant torespond. The level of customization of products willmost likely differ between the industries represented.Since customizers are more likely to respond, thismay explain the respondent bias based on industrycodes.

Multiple respondents from the same plant arecompared to assess the degree of agreement andthereby appraise the reliability of responses from the

Ž .primary informants. James et al. 1984 developed amethod to assess the degree of agreement amongraters. Data for two respondents were collected for47 plants in the present study. Threshold values havebeen established to determine AgoodB reliability,where values closer to 1.0 represent better agree-ment. All values reported for the scales used exceed0.70 and, therefore, are judged to have AgoodBagreement.

To focus respondents on the customized portionof their product lines that may be produced at aplant, the following definition of customization wasincluded in the survey:

Customized products are those products that aredesigned, altered, or changed to fit the specifica-tions of an end-user. Please answer the following

questions regarding only your AcustomizedB prod-ucts. Component or intermediate products are onlyconsidered to be customized if the user of thefinished product dictates or influences the specifi-cations of the component.

5.2. Instrument deÕelopment

To the extent possible, established scales wereused to enhance validity, reliability and generaliz-ability of measures. Established scales were usedextensively for the contextual variables and will bedescribed in Section 5.2.2. When proven establishedscales were not available, survey questions weredeveloped based on existing literature. The classifi-cation variables, customer involvement and modular-ity were developed from literature and are discussedin detail in Section 5.2.1.

5.2.1. Classification ÕariablesBecause of the paucity of empirical research pub-

lished on mass customization, two key scales wereconstructed: customer involvement in the design pro-cess and product modularity. Exploratory factor anal-ysis using Principle Components and a Varimaxrotation was used for the two scales representingcustomer involvement and modularity, respectively,as the initial determinant of factor composition fol-lowing the criteria recommended by Hair et al.Ž .1992 . These authors suggested that for a simplestructure factor solution, only one loading on anyfactor for each variable should be significant, andthat the lowest factor loading to be considered signif-icant would, in most instances, be 0.30. These crite-ria were upheld for both the customer involvementand modularity scales. All items had significant fac-tor loadings on at least one factor as described by

Ž .Hair et al. 1992 .Following the logic recommended by Carmines

Ž .and Zeller 1979 , the number of factors was deter-mined using an external validation technique. Gerb-

Ž . Ž .ing and Anderson 1988 and Anderson et al. 1987suggested a similar technique using the correlation offactors with external variables to test for unidimen-sionality of factors. External validation techniquesand inter-item analysis were used to determine thenumber and composition of the factors and to vali-date the unidimensionality of these factors. These

Ž .methods are supported by Flynn et al. 1990 , Ander-


Ž .son and Gerbing 1982 , and Carmines and ZellerŽ .1979 . Further factor simplification and reliabilityassessments were made using Cronbach’s reliabilitycoefficients. The resulting factors were developedusing the factor analysis to produce individual stan-dardized factor scores.

5.2.1.1. Customer inÕolÕement. To determine thepoint of customization in the production cycle, thissurvey addressed the stages where an end-user cus-tomer participates in specifying the product. Using

Ž .the definition of Mintzberg 1988 as a guideline,items were included that address the various waysproducts can be specified from customers. Thesequestions were designed to measure the point ofcustomer involvement in the design process as oneof the following: design, fabrication, assembly, oruse. Respondents were asked to indicate the level ofagreement, from strongly disagree to strongly agree,with each customer-involvement-related statementusing a seven-point Likert scale.

The initial exploratory factor solution was simpli-fied using inter-item analysis and validated followingthe logic of construct validation, as described above.One item was dropped from each of the scales. Infactor one, Acustomers can assemble a product fromcomponentsB was omitted to increase the Cronbach’salpha coefficient from 0.7422 to 0.7657. In factortwo, Acustomer specifications are used to alter stan-dard components for each orderB was dropped toimprove the Cronbach’s alpha coefficient from0.6255 to 0.6404. Both these factors’ Cronbach’scoefficient alpha exceed the 0.60 threshold often

Ž .cited for exploratory work Nunnally, 1978 to assessinter-item reliability. The resulting two factors repre-sent customer involvement in the design process andsupport the constructs proposed in the original con-ceptual model. These two factors are intuitively ap-pealing as they place customer involvement into oneof two stages of the production cycle: design andfabrication or assembly and use. The specific itemsthat support each factor are shown below.

5.2.1.2. Customer inÕolÕement in the design and( )fabrication stages CI DESFAB .–

Ø Customer’s requests are uniquely designed intothe finished product;

Ø Each customer order requires a unique design;Ø Customers can specify new product features;Ø Each customer order requires the fabrication of

unique components prior to assembly; andØ Customers can specify size of products.

5.2.1.3. Customer inÕolÕement assembly and use( )CI ASMUSE .–

Ø Each customer order is assembled from compo-nents in stock;

Ø Customers can select features from listings;Ø Customer orders are filled from stock; andØ Customers can assemble a product from compo-

nents.

Ž .The first factor CI DESFAB represents cus-–tomer involvement in the design and fabricationstages. Customers can change the actual design ofthe product or introduce new features rather thanselecting features from a listing as specified in thefactor, CI ASMUSE. This involvement requires the–design or fabrication of a unique component for suchcustomers.

Ž .In the second factor CI ASMUSE , all items–relate to the involvement of the customer through theselection of standard components or products from aprescribed listing of features. This involvement doesnot allow for new designs or features to be produced;therefore, the customer is not involved in design andfabrication, but instead in the assembly or use stageof the production cycle.

In combination, these two factors, CI DESFAB–and CI ASMUSE, accurately depict the role of the–customer in the design process as seen in Fig. 3.CI DESFAB represents the involvement of the cus-–tomer at the beginning stages of the productioncycle, when product are designed and componentsare fabricated. CI ASMUSE represents involvement–of the customer in the assembly and use stages of theproduction cycle. Positive values for these scalesshow that the respondent agrees that a particularmode of customer involvement is used in their cus-tomized product line. Negative values show that therespondents disagree that customers are involved at aparticular point in the production cycle.


5.2.1.4. Modularity. The type of modularityemployed is suggested to be a critical issue in under-standing a manufacturer’s approach to mass cus-tomization. Items addressing the modularity of cus-tomized product lines are based on the definition of

Ž . Ž .modularity types of Ulrich and Tung 1991 Fig. 2 .The initial exploratory factor solution was simplifiedusing inter-item analysis and validated following thelogic of construct validation of Carmines and ZellerŽ .1979 . The resulting two-factor solution is presentedbelow.

Ž .The first factor MOD FAB includes four items–that reflect modularity issues involving design orchanges to the components for a specific customer.MOD FAB can be considered a measure of modu-–larity in the design or fabrication of a product. The

Ž .second factor MOD STD contains five items that–address modularity in the form of options to standardproducts or interchangeability of components. Thistype of modularity most likely will be utilized in theassembly stages of a manufacturing process. Thesetwo factors represent two distinct approaches tomodularity. The components of these factors arelisted below.

(5.2.1.5. Modularity through fabrication MOD–)FAB .

Ø Components are designed to end-user specifica-tions;

Ø Components are sized for each application;Ø Components are altered to end-user specifica-

tions; andØ Component dimensions are changed for each

end-user.

(5.2.1.6. Modularity through standardization MOD–)STD .

Ø Products have interchangeable features and op-tions;

Ø Options can be added to a standard product;Ø Components are shared across products;Ø New product features are designed around a stan-

dard base unit; andØ Products are designed around common core tech-

nology.

Reliability assessments were made using Cron-bach’s reliability coefficients. MOD FAB achieved–a Cronbach’s alpha of 0.7887 and MOD STD–yielded a coefficient of 0.6901. These alpha valuesboth exceed the suggestion of 0.60 for exploratory

Ž .research by Nunnally 1978 .When taken together, these two factors provide a

measure of the types of modularity in use. MOD–FAB, when positive, indicates that the respondentagrees that some components are fabricated or sizedto provide customization in their products. MOD–STD, when positive, indicates that the respondentagrees that features and options are added to stan-dardized components or base technologies to achievecustomization of end-products.

5.2.2. Contextual ÕariablesThe typology is explored using manufacturing

decision variables that reflect the paradoxical natureof mass customization. Mass customizers can chooseto develop manufacturing systems that are based onthe traditional manufacturing practices of AcustomBcraft or standard AmassB produced products. Threecategories of structure and infrastructure manufactur-

Žing decision variables Hayes and Wheelwright,.1984 were used to represent the manufacturing sys-

tem variables that are implicit in the tactical opera-tion’s concepts of AmassB and AcustomB. To explorethe nature of AmassB in a manufacturing system, thestructural variables of process choice and technologywere used. Inherent in the process choice decision isan implication of product volumes as seen in the

ŽProduct–Process matrix variables Hayes and.Wheelwright, 1984 . If high volumes are anticipated,

as in standard product production, line processes areselected over job shop processes that are reserved forcraft production. Therefore, the selection of lineprocesses may imply that mass customizers areutilizing a large volume of modules or standardcomponents. Process choice was represented by therespondent assessing the usage rates of the traditionalprocess forms: job shop, batch, line, continuous. Tocapture the usage of purchased components,Apurchased from suppliersB was added. Respondentswere asked to identify, on a seven-point Likert scale,the appropriate level of usage expressed as a percent-age of the products from ANo Products — 0%B,


through ASome Products — 50%B, to AAll Products— 100%B.

To update the traditional process choice alterna-tive, technology usage also was selected as an indica-tor of AmassB manufacturing techniques. Newermanufacturing technologies may provide similar in-dications of higher volume production of standardparts or modules. Technology variables were devel-oped based on a set of items in the Boston Univer-

Žsity Manufacturing Futures Project Miller and Voll-.mann, 1985; Ward et al., 1988 . These items have

been used effectively and have been deemed reliableŽ .De Meyer and Ferdows, 1985; Boyer et al., 1996 .These scales develop three technology variables: de-sign, manufacturing and administrative. Administra-tive technologies are used to represent process con-trol methods. Variables were included to representdesign and manufacturing technologies to better ex-amine the nature of the manufacturing approach ofeach type. Design technologies include: Computer-

Ž .Aided Design CAD , Computer-Aided EngineeringŽ .CAE , and Computer-Aided Process PlanningŽ .CAPP . The manufacturing technologies include:

Ž .Computer Numerical Control CNC , Computer-Ž .Aided Manufacturing CAM , Robotics, Real-time

Ž .process control system, Group Technology GT ,FMS, and bar codingrautomatic identification. Therespondent was asked to APlease indicate the extentto which the following are used for your ‘custo-mized’ productsB. The seven-point Likert scale wasanchored at 1 with Anot used — 0%B, at 4 withAused for some orders — 50%B, and at 7 with Ausedfor all orders — 100%B.

Customization of products can also be assessed bythe usage of variables representing production plan-ning and material control methods as described by

Ž .Hayes and Wheelwright 1984 . Administrative tech-nologies can be used to assess these methods. In

Ž .addition, Vickery et al. 1999 used a firm’s made-Ž .to-order MTO capability to capture the extent to

which a company customizes. The use of productionplanning methods to facilitate MTO manufactu-ring was used as a tactical representation of cus-tomization in manufacturing systems. Variables wereincluded to explore administrative technology andproduction control methods. Administrative tech-nologies were developed from the scales described

Žabove De Meyer and Ferdows, 1985; Boyer et al.,

.1996 . Administrative technologies include: Elec-Ž .tronic Data Interchange EDI , Material Requirement

Ž . Ž .Planning MRP , Decision Support Systems DSS ,and Knowledge-Based Systems. These technologiescan be used to facilitate the production planning andmaterial control functions.

Production planning techniques also were mea-sured by assessing the usage of different methods:

Ž .MTO, made-to-stock MTS , assemble-to-orderŽ .ATO and JIT. The respondent was asked to APleaseindicate the degree to which the following produc-tion planning techniques are used for my ‘custo-mized’ product lineB. The seven-point Likert scalewas anchored with Astrongly disagreeB and AstronglyagreeB.

5.3. Financial performance

Business performance is a crucial indicator for allŽ .strategic configurational works Ketchen et al., 1993 .

To obtain a relative measure of performance whilepreserving privacy, this study used perception ofperformance in relation to competitors. Return on

Ž .investment, return on sales profit margin , and mar-ket share were used to measure performance relativeto competitors. In addition, growth in these mea-sures, as well as sales growth, was used to capturetrend in performance. These measures have been

Ž .used most recently as a group in Boyer et al. 1997Ž .and Vickery et al. 1994 . This performance factor

achieved a Cronbach’s alpha of 0.914, which ex-ceeds suggestions for exploratory research by Nun-

Ž .nally 1978 .

6. Operationalized model

Customer involvement in the production processdetermines the degree of uniqueness of the product,and is represented by two variables, CI DESFAB–and CI ASMUSE. CI DESFAB represents involve-– –ment in the design and fabrication stage while CI–ASMUSE represents involvement during the assem-bly or use phases. Modularity follows a similarpattern. Modularity in the design and fabrication of aproduct is represented as MOD FAB, and MOD– –STD represents the use of standardized componentsin the assembly or use phases of a production.


6.1. Classification criteria

To classify any company, the specific combina-tion of customer involvement and modularity vari-ables must be examined. In practice, it is reasonableto assume that once a customer is involved in theprocess, or modularity is employed, that involvementor modularity would carry throughout the productioncycle. For example, if a customer’s initial point ofinvolvement is in the design stage of the productioncycle, the customer’s preference would be incorpo-rated throughout the remaining stages of fabrication,assembly, and use. With regard to modularity, asimilar situation exists. If a product were manufac-tured from modular components in the design pro-cess, these modular components would be includedin the product throughout the production cycle. Sinceonly one measure is required of each company foreach construct, the earliest point of involvement inthe production process, either customer or modular-ity, will be used to represent the respective variablefor that company. For each case, a set of two values,one that corresponds to the customer involvementaxis and one that represents the modularity axis, isneeded.

To operationalize this concept, the factor scorescorresponding to the earliest point of involvement,design and fabrication, of the product will be consid-ered first. If a respondent company scores positivelyin the design and fabrication variable, this variablewill be used to represent the construct on the axis. Ifa company’s response yields a negative score, thedesign and fabrication variable is excluded and theassembly and use variable is examined. A positivevalue on assembly and use is used to represent theconstruct, while negative values are excluded. Ifnegative values occur for both the design and fabri-cation variable and the assembly and use variable, novalue is assigned. Table 1 shows the mass customiza-tion groups and the corresponding variable values.

Once each company has been assigned one valuefor each of the variables, customer involvement andmodularity, the classification process is simplified.Each respondent is assigned to a specific cell of thematrix based on the values used to create the vari-able. If a company has customer involvement andmodularity variables assigned from the design andfabrication stages, these variables would be assigned

Table 1Group classification scheme — value of variables

Group Modularity Customer involvement

Designr Assemblyr Designr Assemblyrfabrication use fabrication use

1, Fabricators q " q "

2, InÕolÕers y q q "

3, Modularizers q " y q4, Assemblers y q y q

to Group 1, Fabricators. If the variables assigned forboth customer involvement and modularity were fromthe assembly and design stages, the case would beassigned to Group 4, Assemblers. The classificationof 194 companies resulted in 126 mass customizers:77 Fabricators, 15 InÕolÕers, 17 Modularizers and17 Assemblers.

6.2. Industry effects

The groups were tested to determine the effects ofindustry on group membership. Using SIC codes,plants were grouped into similar SIC classifications.SIC data were available for 116 of the 126 masscustomizers. A Chi-square test of group membership

Ž .and SIC category was not significant ps0.56 .Therefore, group membership is not related to indus-try classification.

7. Manufacturing context of typology

After the mass customization types were explic-itly identified, the groups were examined to fulfillthe third purpose of the study — to explore thedifferent manufacturing approaches to mass cus-tomization implied by the typology. The conceptualmodel allows for different implementations of a masscustomization strategy. As described in previous sec-tions, three categories of variables were selected torepresent the AmassB and AcustomB components of amanufacturing system: process choice, planningtechniques and technology. For descriptive purposes,an ANOVA with post-hoc Scheffe’s test was used to´determine the differences of the variables between


groups. Table 2 shows the group means and ANOVAsignificance levels of these variables. Significant dif-ferences exist between the groups for at least onevariable in each of the categories of process choice,technology, production planning, and performance atthe 0.05 alpha level. This implies that there aredifferences in the manufacturing implementation ofmass customization between the groups.

7.1. Process choice

The mass customization types use different pro-cesses to achieve their mass customization capabil-ity. Significant differences exist between the groupsfor percent of components manufactured using lineprocesses and the percent of components that arepurchased, but not for job shop and batch processes.

Table 2Summary of ANOVA four-group matrix — significance of differences between group means of manufacturing system variables

Variables p-value, Group 1, Group 2, Group 3, Group 4, ns126one-way Fabricators InÕolÕers Modularizers Assemblers

Ž . Ž . Ž . Ž .ANOVA ns77 ns15 ns17 ns17

Process choiceJob 0.119 3.5 4.27 3.5 2.59 124Batch 0.881 3.57 3.67 3.31 3.82 124

) ) ) )Ž . Ž . Ž .Line 0.000 2.72 3 , 4 2.86 4.44 1 4.56 1 120) )Ž . Ž .Purchased 0.012 2.93 3.50 3 1.75 2 2.59 123

Process control†Ž . Ž . Ž .MTS 0.013 2.70 4 3.73 2.23 4† 4.00 1†, 3† 113

ATO 0.309 5.89 6.33 5.54 5.20 118) )Ž . Ž .MTO 0.008 6.41 4 6.33 6.36 5.13 1 121

JIT 0.503 4.17 4.47 3.33 4.20 111

TechnologyDesign

) )Ž . Ž .CAPP 0.011 2.96 4.13 3 1.67 2 2.47 123) ) ) ) ) )Ž . Ž . Ž .CAD 0.000 5.93 3 6.00 3 4.05 1 , 2 5.06 126

CAE 0.430 3.84 4.07 3.29 3.06 125

ManufacturingCAM 0.394 3.74 3.71 3.00 3.00 124CNC 0.109 4.28 3.20 3.33 3.47 124

Ž . Ž .Robotics 0.009 1.67 2.43 1.25 3 2.59 4 122Group Technology 0.143 2.36 3.47 2.20 2.31 119Real-time process controls 0.071 2.79 3.27 1.56 2.88 122

) )Ž . Ž . Ž .FMS 0.011 3.09 2† 4.47 1†, 3 2.13 2 3.29 121Bar coding 0.173 3.09 3.33 1.88 3.31 122

AdministrativeŽ . Ž .EDI 0.046 3.20 4.07 4† 2.56 2.19 2† 122

) )Ž . Ž .MRP 0.042 4.97 2 6.53 1 4.75 5.29 124DSS 0.872 2.28 2.60 2.47 2.63 116Knowledge-based 0.687 2.43 2.47 2.50 2.53 117

) )Ž . Ž . Ž .Performance 0.000 y0.162 4 0.557 y0.002 4† 0.813 1 , 3† 126

)Significant differences between groups at as0.05.))Significant differences between groups at as0.01.†Significant differences between groups at as0.10.


Fabricators exhibit a significantly lower level usageof line processes in the manufacturer of componentsthan Modularizers and Assemblers. Fabricators arethe mass customization type that most closely mir-rors pure customization. Therefore, this limited useof line processes for component manufacturer in theFabricators is not surprising. Fabricators, by defini-tion, provide customization through fabrication ofdistinct components. This custom fabrication wouldincrease the need for more flexible manufacturingmethods than the traditional line manufacture ofcomponents.

Assemblers have been described as the mass cus-tomizer that most closely resembles mass producers.Assemblers have the highest usage of line processes,which is consistent with their similarity to massproducers. Therefore, the use of line manufacture byAssemblers is expected. Modularizers also incorpo-rate line manufacturer into their manufacturing sys-tems. Modularizers and Assemblers share customerinvolvement in the late stages of the productioncycle, but Modularizers utilize modularity in theearly stages of the production cycle. Modularizersmay use modules to provide component commonal-ity without providing customization until the cus-tomer is involved in the later stages.

InÕolÕers have the highest usage of purchasedcomponents and show significant difference withModularizers. These two groups do not share cus-tomer involvement or modularity dimensions, butneither group has matched involvement of customerand modularity dimensions in the production cycle.InÕolÕers have customer involvement early in theprocess and modularity later in the production cyclewhile Modularizers exhibit the opposite character-istics. Perhaps for InÕolÕers, the early customer in-volvement allows time for the purchase of compo-nents for specific customer.

7.2. Production control

The use of process control techniques mirrors theanticipated usage by mass customization type. As-semblers, the mass customizer that most closelyresembles mass producers, have the highest usage of

Ž .make-to-stock MTS planning systems. Assemblersdiffer significantly in their usage levels of MTS fromFabricators and Modularizers, both of which utilize

modularity early in the production process. In addi-tion, Fabricators, the mass customizers that mostclosely resemble pure customizers, have the highestusage in MTO planning systems, which differs sig-nificantly from the Assemblers. The use of processcontrol techniques follows the anticipated usage forAssemblers and Fabricators. It should be noted,however, that if the raw scores are examined, allmass customizers utilize high levels of MTO andATO planning technique and moderate levels of JITin their manufacturing system. This high usage levelof MTOrATO control techniques is expected. Bydefinition, all mass customers should have customerinvolvement specifying the product’s attributes atsome point in the production cycle. This customerinvolvement would require the use of production toorder and not to stock.

7.3. Technology

Mass customizers differ in their usage of technol-ogy. At least one variable differs significantly acrossthe groups for design, manufacturers and administra-tive technologies. For design technologies, the groupModularizers is significantly different from Fabrica-tors and InÕolÕers. Modularizers have significantlylower usage of CAD technologies than Fabricatorsand InÕolÕers. The conceptual model suggests thatFabricators and InÕolÕers have customer involve-ment in the design and fabrication stages of theproduction cycle while Modularizers have customerinvolvement in the assembly and use stages. It ap-pears that mass customizers employ CAD technol-ogy when customers are involved in the early stagesof production. The use of design technologies, suchas CAD, to manage the customer specifications inthese stages should be expected. When customers areinvolved early in the process, more resources may berequired to manage the design function.

However, use of CAPP differs between InÕolÕersand Modularizers with InÕolÕers exhibiting thehighest usage of CAPP technology. Once again, thisfinding is expected since InÕolÕers have early cus-tomer involvement that must be planned and man-aged through the assembly stages of production. Thisarrangement would require the use of CAPP tech-nologies. Unexpectedly, the third design technology

Ž .variable CAE did not reflect any differences among


the groups. The raw scores for these variables reflectlow levels of usage of CAE. Groups using modular-ity in the early production stages require modules tobe altered and, therefore, may rely on CAD fordesign purposes. Perhaps, the emphasis on CADtechnologies supplants the use of CAE technologiesfor these mass customizers.

For manufacturing technologies, the most signifi-cant differences between the groups can be seen inthe usage of flexible manufacturing technologiesŽ .FMS . The highest users of FMS technology areInÕolÕers — the mass customizer with customerinvolvement early in the production stage and modu-larity employed at later stages. The InÕolÕers havesignificant difference in FMS usage than both Fabri-cators and Modularizers, both which have modular-ity early in the process. This indicates that if cus-tomers are involved prior to the usage of modularity,mass customizers are adopting a higher level of FMSusage. This result is not unexpected. FMS usage mayreplace the more traditional forms of modularity inthe early stages of production. FMS modularity isgained through modularity of programs and replica-tion of design aspects between products. The flexibil-ity of this technology to quickly change part typesallows customer to specify more aspects of a productthan may not be captured with the concepts ofmodularity presented in this study.

Robotics usage also differed across the mass cus-tomizing groups. Assemblers have the highest usagewhile Modularizers have the lowest usage. How-ever, the raw scores for both groups are less thanthree, which would indicate Robotics usage on less

Žthan of 32% of products. Note: The questions an-chored the raw scale score responses to percentagesof usage. The seven-point Likert scale was anchoredat 1 with Anot used — 0%B, at 4 with Aused forsome orders — 50%B, and at 7 with Aused for allorders — 100%B. A score of 3 is approximately

.32%. This does indicate a greater usage of Roboticsin Modularizers than Assemblers, although no sig-nificant conclusions should be drawn from this fact.The other manufacturing technology variable, CAM,CNC, Group Technology and Bar coding, do notshow significant differences between the groups.However, these variables can give us a richer pictureof the mass customizers. The low level of usage ofthese technologies by all the mass customizers in the

study may signify their lack of importance to theimplementation of mass customization.

Administrative technology usage also differsamong mass customizers. Assemblers have signifi-cantly lower usage levels of EDI than InÕolÕers.Both these groups have modularity in the later stagesof the production process, but differ on the customerinvolvement. The higher usage level of EDI byInÕolÕers may correspond to an earlier involvementof customers. This finding would not be surprising ifthe electronic communication medium is used forcustomer interaction. However, this study cannotconfirm the usage of EDI with customer as opposedto suppliers.

The usage of MRP system also differ among thegroups with InÕolÕers exhibiting the highest usagewhile Modularizers and Fabricators exhibit lowerlevels of usage. InÕolÕers have the highest usagelevel, and although not significant, Assemblers havethe second highest usage levels. InÕolÕers and As-semblers share modularity in the later stages ofproduction. It is not surprising that these mass cus-tomizers would use MRP technologies to plan for theuse of the modules in the assembly and use stages ofthe production cycle.

7.4. Financial performance

Business performance varies both within andacross mass customizer configurations. Both highand low performers are practicing mass customiza-tion in each of the groups. Significant differences inthe financial performance factor exist between thegroups, with Assemblers displaying a significantlyhigher group mean for performance than Modulariz-ers and Fabricators. InÕolÕers also exhibits highperformance when measured as a group mean. BothAssemblers and InÕolÕers have modularity in thelater stages of production, but differ on point ofcustomer involvement in specifying the product. Al-though no significant difference is seen between theperformance means of Modularizers and InÕolÕers,differences do exist between Modularizers and As-semblers. The use of modularity in the later stages ofproduction may point to increased performance formass customizers.

The findings with respect to business performanceof companies adhering to each of the mass cus-


tomization types provide valuable insights for com-panies pursuing mass customization. There appearsto be a performance difference among companiespursuing mass customization based on the type ofmodularity employed. Companies using modularityin the assembly and use stages exhibit higher perfor-mance than those using modularity in the design andfabrication stages. This suggests that those masscustomizers that are closest to mass producers inmanufacturing approach are most likely to reap thebenefits of mass customization. Perhaps, these masscustomizers that best able to achieve scale economieswhile delivering a customized product will exhibitbetter financial performance than those that do not.However, this alone may not guarantee high perfor-mance. Companies should be aware that high-perfor-ming companies are found across the spectrum of alltypes of mass customization.

8. Summary and conclusions

This study develops a conceptual typology ofmass customization that provides an explicit meansfor identifying and categorizing mass customizersfrom the perspective of operations. We suggest thattwo variables are key in classifying mass customiz-ers: the point in the production cycle where thecustomer is involved in specifying the product andthe type of modularity used in the product. Wevalidate the conceptual types empirically using datafrom a sample of mass customizers.

We also suggest and demonstrate broader config-urations of mass customization. Specifically, the dif-ferent approaches to mass customization implied bythe typology were examined to provide a richerpicture of the manufacturing systems employed. Pro-cess choice, planning techniques, and technologyvariables were examined as well as business perfor-mance. We demonstrate that mass customizers dodiffer on each of these dimensions based on theirmass customization type. Most particularly in thistypology, those mass customizers that theoreticallyresemble mass producers, Assemblers, choose manu-facturing systems that use line processes, incorporatethe highest levels of MTS planning methods andutilize MRP. Fabricators, those mass customizersthat most closely resemble craft producers, have the

highest levels of MTO planning systems and highusage levels of CAD. The commonality of the manu-facturing systems among mass customizers is alsonoteworthy. All mass customizers use some form ofa MTO or assemble-to-order planning system, incor-porate some form of batch processing and a majorityof the products are made using CAD technology.This finding is not surprising. The use of CADtechnology may be required to manage the AcustomBportion of the order in an efficient cost-effectivemanner. Batch processing may provide the AmassBproduction of common modules. Then, the overallproduction process is managed using MTO and as-semble-to-order planning systems. The differencesand commonalties among these mass customizationgroups provide a rich description of mass customiza-tion manufacturing systems.

Although both high and low performers are foundamong all mass customization types, we do discernbetter business performance among the types that usestandard modules and employ modularity in the laterstages of the production cycle.

8.1. Limitations and future research

This study has only begun to explore mass cus-tomization as a manufacturing phenomenon. Thisstudy takes a step forward in mass customizationresearch by providing a conceptual model of masscustomization and substantiating this model throughan empirical investigation. However, this empiricalexploration provides a one-time snapshot of com-pany practices. A natural extension of this research,and most empirical work in manufacturing strategy,is a longitudinal examination of companies. Manu-facturing systems are ever changing and a time lagmay exist between making a decision on manufactur-ing priorities and realizing the related manufacturingcapability. A longitudinal look at mass customizerswould clarify the issues relating to the implementa-tion of this strategy in practice.

This study neglects to include the use of serviceas a mass customization technique. In addition tocustomizing product attributes, products may bemass-customized through the availability of cus-tomizable services. Manufacturers are increasinglylooking to expand their product offerings through the

Žaddition of service to the product package Wise and


.Baumgaretener, 1999 . Services may also be modu-larized and may provide another avenue for masscustomization. Future research may wish to includeservices as part of the mass customization model.

In addition, this study only investigates the cus-tomized portion of company’s product lines. Theintegration of customized and standardized productsis not examined. Mass customization may enhanceoverall firm performance, including the design andproduction of standard products, through informationgained regarding customer preferences. The relation-ship of mass customization to the entire organizationmay play a critical role in the success of a masscustomization strategy. Therefore, the scope of masscustomization research should be broadened to in-clude both mass-customized and standard products.

This paper has broadened the conceptual model ofmass customization and its manufacturing implica-tions, but has neglected to make any specific valuejudgments to the inherent worth of mass-customizedproducts. Future studies may wish to explore themarket implications of mass customization, the ac-companying customer benefits, the effects of choiceon customer satisfaction, and the costs associatedwith the implication of mass customization practices.

References

Anderson, C., Gerbing, W.R., 1982. Some methods for respecify-ing measurement models to obtain unidimensional construct

Ž .measurement. Journal of Marketing Research 19 4 , 186–192.Anderson, J.C., Gerbing, D.W., Hunter, J.E., 1987. On the assess-

ment of unidimensional measurement: internal and externalconsistency and overall consistency criteriarsome further re-marks on measurement-structure interaction and the unidimen-

Ž .sionality of constructs. Journal of Marketing Research 24 4 ,432–444.

Baldwin, C.Y., Clark, K.B., 1994. Modularity-in-design: an analy-sis based on the theory of real options. Working Paper,Harvard Business School, Boston, MA.

Boyer, K., Ward, P.T., Leong, G.K., 1996. Approaches to thefactory of the future — an empirical taxonomy. Journal of

Ž .Operations Management 14 4 , 297–313.Boyer, K., Leong, K., Ward, P., Krajewski, L., 1997. Unlocking

the potential of advanced manufacturing technologies. JournalŽ .of Operations Management 15 4 , 331–347.

Carmines, E.G., Zeller, R.A., 1979. Reliability and Validity As-sessments. Sage Publications, Beverly Hills, CA.

Davis Stanley, M., 1987. Future Perfect. Addison-Wesley Publish-ing, Reading, MA.

De Meyer, A., Ferdows, K., 1985. Integration of informationsystems in manufacturing. International Journal of Operations

Ž .and Production Management 5 2 , 5–12.Dillman, D.A., 1978. Mail and Telephone Surveys: The Total

Design Method. Wiley, New York.Donlon, J.P., 1993. The six sigma encore; quality standards at

Ž .Motorola. Chief Executive 90 , 48–54.Flynn, B.B., Sakakibara, S., Schroeder, R.G., Bates, K.A., Flynn,

E.J., 1990. Empirical research methods in operations manage-Ž .ment. Journal of Operations Management 9 2 , 297–326.

Garud, R., Kumaraswamy, A., 1993. Technological and organiza-tional designs for realizing economies of substitution. StrategicManagement Journal 16, 93–109.

Gerbing, D.W., Anderson, J.C., 1988. An updated paradigm forscale development incorporating unidimensionality and its as-

Ž .sessment. Journal of Marketing Research 25 2 , 186–192.Goldhar, J., Jelinek, M., 1983. Plans for economics of scope.

Ž .Harvard Business Review 61 6 , 141–148.Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1992.

Multivariate Data Analysis with Readings. 3rd edn. Macmil-lan, New York, NY.

Harrigan, K.R., 1983. Research methodologies for contingencyapproaches to business strategy. Academy of Management

Ž .Review 8 3 , 398–405.Hayes, R.H., Wheelwright, S.C., 1984. Restoring Our Competitive

Edge: Competing Through Manufacturing. John Wiley andSons, New York.

Ketchen, D.J., Thomas, J.B., Snow, C.C., 1993. Organizationalconfigurations and performance: a comparison of theoretical

Ž .approaches. Academy of Management Journal 36 6 , 1278–1313.

Kotha, S., 1995. Mass customization: implementing the emergingparadigm for competitive advantage. Strategic ManagementJournal.

James, L.R., Demaree, R.G., Wolf, G., 1984. Estimating within-group inter-rater reliability with and without response bias.

Ž .Journal of Applied Psychology 69 1 , 85–96.Lampel, J., Mintzberg, H., 1996. Customizing customization.

Ž .Sloan Management Review 38 1 , 21–30.McCutcheon, D.M., Raturi, A.S., Meredith, J.R., 1994. The cus-

tomization responsiveness squeeze. Sloan Management Re-Ž .view 35 2 , 89–99.

Miller, J.G., Vollmann, T.E., 1985. The hidden factory. HarvardŽ .Business Review 63 5 , 142–150.

Mintzberg, H., 1988. Generic strategies: toward a comprehensiveframework. Advances in Strategic Management 5, 1–67.

Nunnally, J.C., 1978. Psychometric Theory. 2nd edn. McGraw-Hill, New York.

Pine, B.J. II, 1993a. Mass Customization: The New Frontier inBusiness Competition. Harvard Business School Press, Boston,MA.

Pine, B.J. II, 1993b. Mass customizing products and services.Ž .Planning Review 21 4 , 6.

Pine, B.J. II, Peppers, D., Rogers, M., 1995. Do you want to keepŽ .your customers forever? Harvard Business Review 72 3 ,

103–114.


Pine, B.J. II, Victor, B., Boyton, A.C., 1993. Making masscustomization work. Harvard Business Review 71, 108–119.

Salant, P., Dillman, D.A., 1994. How to Conduct Your OwnSurvey. Wiley, New York.

Ulrich, K., 1992. The role of product architecture in the manufac-turing firm. Working Paper, MIT, Sloan School of Manage-ment.

Ulrich, K., Eppinger, S., 1995. Product Design and Development.McGraw-Hill, New York, NY.

Ulrich, K., Tung, K., 1991. Fundamentals of product modularity.Proceedings of the 1991 ASME Winter Annual Meeting Sym-posium on Issues in DesignrManufacturing Integration, At-lanta.

Vickery, S., Droge, C., Germain, R., 1999. The relationship

between product customization and organizational structure.Ž .Journal of Operations Management 17 4 , 377–391.

Vickery, S.K., Droge, C., Markland, R.E., 1994. Strategic produc-tion competence: convergent, discriminant, and predictive va-

Ž .lidity. Production and Operations Management 3 4 , 308–318.Ward, P.T., Miller, J.G., Vollmann, T.E., 1988. Mapping manu-

facturing concerns and action plans. International Journal ofŽ .Operations and Production Management 8 6 , 5–18.

Wise, R., Baumgaretener, P., 1999. Go downstream: the newprofit imperative in manufacturing. Harvard Business Review

Ž .77 5 , 133–141.Womack, J.P., 1993. A book review of mass customization. Sloan

Ž .Management Review 34 3 , 121–122.

new approaches to mass customization: configurations and …directory.umm.ac.id/journals/journal of...

Documents