Download as pdf or txt
Download as pdf or txt
You are on page 1of 34

Decision Sciences 

C 2012 The Authors

Volume 43 Number 1 Decision Sciences Journal 


C 2012 Decision Sciences Institute

February 2012

Order Management in the


Customization-Responsiveness Squeeze∗
Antti Tenhiälä† and Mikko Ketokivi
IE Business School, Calle de Marı́a de Molina 12—5, 28006 Madrid, Spain,
e-mail: antti.tenhiala@ie.edu, mikko.ketokivi@ie.edu

ABSTRACT
Make-to-order (MTO) products may be either customized or standard, and customization
can occur either at the configuration or component level. Consequently, MTO production
processes can be divided into three customization gestalts: non-customizers, custom as-
semblers, and custom producers. In this article, we examine how the multilevel nature of
customization affects order management in processes that produce complex MTO prod-
ucts. We first empirically validate the existence of the three customization gestalts and
subsequently, analyze the order management challenges and solutions in each gestalt
in a sample of 163 MTO production processes embedded in seven different supply
chains. In the analyses, we follow a mixed-methods approach, combining a quantitative
survey with qualitative interview data. The results show that important contingencies
make different order management practices effective in different gestalts. Further quali-
tative inquiry reveals that some seemingly old-fashioned practices, such as available-to-
promise verifications, are effective but commonly neglected in many organizations. The
results also challenge some of the conventional wisdom about custom assembly (and
indirectly, mass customization). For example, the systematic configuration management
methods—conventionally associated with project business environments—appear to be
equally important in custom assembly. [Submitted: March 22, 2011. Revision received:
June 1, 2011. Accepted: July 1, 2011.]

Subject Areas: Decision Support Systems, Field Studies, Manufacturing-


Sales Interface, Mass Customization, Modular Design, and Operations
Management-Information Systems Interface

INTRODUCTION
How can make-to-order (MTO) manufacturers of complex products—heavy-
duty industrial cranes, high-technology weapons systems, and special-purpose

∗ The authors gratefully acknowledge Editor Asoo Vakharia, the anonymous Associate Editor, and the
anonymous reviewers for constructive comments. We are also grateful to Kari Tanskanen, Fabrizio Salvador,
Gopesh Anand, Suzanne de Treville, and Kenneth Boyer for their comments on earlier drafts of the paper. We
further gratefully acknowledge financial support from Tekes—The Finnish Funding Agency for Technology
and Innovation (PROLOG and SSOC projects) and the Spanish Ministry of Science and Innovation (Grant
No. ECO2010–18293: Micro-Foundations of Organizational Capabilities). Finally, we thank the employees
of the participating organizations for sharing their expertise with us and critically contributing to the success
of this research.
† Corresponding author.

173
174 Customization-Responsiveness Squeeze

elevators—cope with customers that demand not only swift and punctual deliv-
eries but also increasingly sophisticated and customized product features? In this
article, we answer this question with an in-depth field study of how variance in the
level of product customization influences the ways in which manufacturing firms
can effectively manage their day-to-day processes.
Our contribution unfolds in two ways. On the one hand, we address one
part of the practically topical question of how manufacturers can tackle multiple
and often conflicting competitive priorities. In customized production, this conflict
has been labeled the customization-responsiveness squeeze, caused by production
times being often considerably longer than delivery times desired by the customer
(McCutcheon, Raturi, & Meredith, 1994). The topic has received research attention
from multiple perspectives: mass customization (Pine, 1993), product architectures
(Salvador, Forza, & Rungtusanatham, 2002), process design (Tu, Vonderembse,
Ragu-Nathan, & Ragu-Nathan, 2004), supplier relationships (Krajewski, Wei, &
Tang, 2005), customer involvement (Duray, Ward, Milligan, & Berry, 2000), and
supply chain structures (Randall & Ulrich, 2001). What is missing, however, is a
theory that explains the variance in the effectiveness of different order management
practices. The aim in this article is to formulate and test such a theory.
On the other hand, we contribute to the more formal theoretical discussions
on organizational information processing by looking at the implications of the
multilevel aspects of customization. In particular, according to the information
processing view of organization design (Galbraith, 1973; Tushman & Nadler,
1978), the requisite information processing capacity of an organization depends
on multiple factors. Thus far, however, the effect of product customization has been
either neglected or at best modeled through rather crude proxies (e.g., Daft, 2004).
This study suggests how product customization can be modeled more accurately
as a two-dimensional construct and how this construct influences the requisite
information processing capacity of an organization.
In MTO manufacturing, the order management practices of interest are the
use of product configurator (PC) software, available-to-promise (ATP) verifica-
tions, and configuration management (CM). In contrast with the extant, primarily
analytically based research on order management (e.g., Ramdas, 2003), we con-
duct an empirical study of the effectiveness of the different order management
practices. To be sure, analytical research has helped us understand how to opti-
mize different tasks in order management, such as the delivery date promising
(Bertrand, Zuijderwijk, & Hegge, 2000; Barut & Sridharan, 2005; Venkatadri,
Srinivasan, Montreuil, & Saraswat, 2006). For some reason, however, the applica-
tions that have been found effective in analytical research have been scarcely
implemented in practice (e.g., Krishnan & Ulrich, 2001). This leads to the
question: have researchers really modeled the application context in adequate
detail?
In an attempt to enhance the empirical and theoretical grounding as well
as the practical relevance of this research stream, our research effort draws
from a multisource data set on the order acquisition and fulfillment practices
in 163 MTO production processes embedded within seven different supply
chains. We first develop a two-dimensional measurement framework for prod-
uct customization and subsequently use it to analyze the contingency effects of
Tenhiälä and Ketokivi 175

customization on the effectiveness of different order management practices (PC,


ATP, and CM). Finally, the quantitative results are elaborated using further quali-
tative inquiry.
Some of our findings are context specific, others are more generalizable.
The first key finding is that customization is, both theoretically and empirically, a
multilevel concept, which manifests itself in various customization gestalts. Cus-
tom assembly, for instance, is a mixture of non-customized mass production and
fully customized production, as suggested in the mass customization literature
(Pine, 1993). Our findings remind that the practices of both mass production and
fully customized production have to be mastered in order to succeed as a custom
assembler of complex products; implementing a PC and otherwise emulating the
order management practices of mass production is not sufficient. In addition, it is
necessary to master the basic project management practices of the CM methodol-
ogy, conventionally associated with one-off production systems. Our results also
show that MTO manufacturing does not always require the implementation of
familiar mass customization concepts such as postponement, modular products,
and standardized components.

LITERATURE REVIEW AND BASELINE HYPOTHESES


The essential information in MTO manufacturing is recorded in customer orders
that specify what must be produced and when the products must be finished. Or-
der management involves creating and maintaining the information about product
specifications and the promised delivery dates. Thus, order management links
directly to two dimensions of manufacturing performance, namely product con-
formance and delivery performance. Effective order management ensures that
product specifications match customers’ needs and thus links to product confor-
mance. Delivery performance is in turn facilitated due to the enhanced ability
to obtain feasible production schedules and accurate and reliable delivery date
promises.
On a general level, order management can be divided into two phases: order
acquisition and order fulfillment (Forza & Salvador, 2002b). The challenges of the
first phase are in the customer interface, where the task is to configure producible
solutions that conform to the heterogeneous customer needs (Salvador & Forza,
2004). In addition to ensuring that the promised product is technically feasible, the
first phase includes the determination of delivery dates that are feasible in terms
of available materials and capacity (Bixby, Downs, & Self, 2006; Zorzini, Corti,
& Pozzetti, 2008). In the order fulfillment phase, the challenge lies in coping with
potential modifications to product specifications and delivery dates, both of which
are endemic to most MTO manufacturing environments (Danese & Romano, 2004;
Van Wezel, Van Donk, & Gaalman, 2006).
In the following, we structure the literature review and theory development
according to these two generic phases of order acquisition and fulfillment. In the
theory development, we first formulate hypotheses that reflect the conventional
wisdom on order management. We then extend these baseline hypotheses with our
contingency-theoretical arguments.
176 Customization-Responsiveness Squeeze

Order Acquisition
MTO manufacturing has already begun to manifest its idiosyncratic characteristics
by the time of the sales transaction. The goal of the transaction is to elicit customer
needs and to communicate the available options. Typical risks include customers
becoming confused with the offered variety and the manufacturer making mis-
takes in configuring the products (Huffman & Kahn, 1998; Hegde, Kekre, Rajiv,
& Tadikamalla, 2005). From an organizational perspective, functional integration
is crucial: sales personnel must be able to ensure both the technical viability of
the offered configurations and the feasibility of the promised delivery dates. This
requires effective flow of information between sales, engineering, and manufac-
turing functions (Lawrence & Lorsch, 1967; Salvador & Forza, 2004). Without
appropriate investments in information processing tools, order acquisition is prone
to errors and waste of resources (e.g., Forza & Salvador, 2006).
PC tools offer one solution to the information processing challenges of the
order acquisition phase. They help ensure technical feasibility by formalizing the
rules about how products can be configured and by providing user interfaces that
help sales personnel translate desired features into technical specifications (Forza
& Salvador, 2002b). These tools can also help customers as they are trying to
understand and articulate their needs to the supplier (Huffman & Kahn, 1998).
The PC tools are often embedded in enterprise resource planning (ERP) systems
(e.g., SAP, 2011a), but they can be also purchased as stand-alone software (e.g.,
Configure One, 2011). Case studies have suggested that configurators are indeed
effective in ensuring products’ technical performance, quality, and conformance
to customer requirements (Forza & Salvador, 2002a,b). In sum, we can formulate
the baseline hypothesis on PCs as follows:

H1a: Use of product configurator software in MTO production processes is


positively associated with product conformance.

While the PC tools help the sales personnel with the product specifications, they are
of little use in trying to determine feasible delivery dates (Forza & Salvador, 2006).
Yet, systematic determination of delivery dates is particularly crucial in MTO
manufacturing, where many materials are often order specific and the capacity
utilization at the time of the order acquisition has considerable influence on the
delivery lead times (Zorzini et al., 2008). In the MTO context, the use of fixed
lead-time quotes always results in ineffective delivery date promises (Proud, 2007).
Under high capacity utilization, this promise is too optimistic and under low
utilization, less responsive than it could be.
Various ATP techniques enable dynamic delivery date determination based
on material availability and current capacity utilization. These techniques are some-
times called capable-to-promise or advanced available-to-promise (a-ATP) verifi-
cations, depending on their specific features (Pibernik, 2005). Following Framinan
and Leisten (2010), we use the generic label ATP in reference to all of these tech-
niques. Although empirical research on the use of ATP tools is scarce, the tools
themselves are widely applied in practice (Kilger & Schneeweiss, 2005; Stadtler,
2005). For example, ERP systems typically feature several alternative techniques
for conducting ATP verifications (e.g., SAP, 2011b). The pioneering case study of
Tenhiälä and Ketokivi 177

Bixby et al. (2006) showed that the use of ATP verifications can facilitate integra-
tion of sales and manufacturing and thus enhance delivery performance. In sum,
the baseline hypothesis on ATP verifications is:
H1b: Use of available-to-promise verifications in MTO production processes
is positively associated with delivery performance.

Order Fulfillment
Once the order fulfillment phase begins, the primary challenge in order manage-
ment is to respond to the changes in product specifications and delivery dates.
Research on MTO industries indicates that customers frequently request amend-
ments to product configurations and delivery dates after the initial placement of
orders (Riley, Diller, & Kerr, 2005). Unfortunately however, such mid-process
changes are often credulously accepted at the cost of other orders. Hasty approvals
of change requests lead to capacity and materials shortages that in turn negatively
affect product conformance and on-time delivery (Hanna, Camlic, Peterson, & Lee,
2004). Despite the risk of such adverse performance effects, MTO manufacturers
feel pressure to approve the changes because freezing the product specifications
and delivery schedules in the order acquisition phase is generally considered un-
acceptable customer service (Danese & Romano, 2004).
In order to avoid the negative effects of customers’ change requests, all
amendments should be evaluated based on their effects on the overall production
plans (Lyon, 2004). For example, the delivery date of an order with a pending
specification change request from the customer may have to be postponed to ensure
that the execution of other orders is not disturbed (e.g., Guess, 2002; PMI, 2006).
The evaluations of requested changes necessitate effective procedures where the
order documents are systematically updated and transferred between the sales and
the manufacturing functions.
To be sure, there are MTO manufacturers for whom change requests from cus-
tomers are not a major concern. The ability to accommodate mid-process changes
is, however, important to these manufacturers as well. This is because changes
may also result from causes other than the changing customer needs, such as pro-
curement delays, machine breakdowns, and quality problems (Koh, Gunasekaran,
& Saad, 2005). In general, MTO manufacturers tend to be considerably more vul-
nerable than make-to-stock manufacturers to typical manufacturing uncertainties
(Koh & Simpson, 2005). Managing uncertainties is more difficult in MTO man-
ufacturing because materials are often order specific and consequently, there are
limited possibilities to replace missing, deficient, or scrapped parts. Use of inven-
tories as buffers against temporary capacity shortages is constrained for the same
reason. These difficulties can, however, be alleviated with diligent management
of order documents: if manufacturing glitches are systematically communicated
by updating the order documents, sales personnel will get advance information
about the forthcoming delivery problems and consequently, will have extra time
to negotiate alternative delivery dates and arrangements with the customers.
The practices of recording and approving changes in the order documents
of customized products are discussed under the rubric of CM (Guess, 2002; PMI,
2006). The CM literature postulates that the initial configurations and delivery
178 Customization-Responsiveness Squeeze

schedules of all orders are documented in standardized forms. In addition, all


changes to the initial documents are subject to approval in formal review proce-
dures that are triggered by issuing standardized order-deviation documents. The
principles of CM originate in project business environments, but they have been
successfully applied in the manufacturing context as well (Lyon, 2004). Formal
management of order documents is beneficial in MTO manufacturing because it
provides a safeguard against accepting changes that jeopardize the feasibility of the
product specifications or the promised delivery date of the order being changed,
or any other order. In sum, we present the following baseline hypotheses on
CM:
H1c: Use of configuration management practices in MTO production pro-
cesses is positively associated with product conformance.
H1d: Use of configuration management practices in MTO production pro-
cesses is positively associated with delivery performance.

A CONTINGENCY THEORY OF ORDER MANAGEMENT


Hypotheses H1a–H1d summarize the general-level conventional wisdom on order
management. Our argument is that the boundaries of applicability of each order
management practice are more specific than these general propositions suggest. In
order to better understand the effectiveness of each practice, one must take into
account the multilevel nature of product customization.

Configuration-Level and Component-Level Customization as Key


Contingencies
Customization of MTO products can occur on two levels. First, products can be
tailored to customer needs by switching and swapping the constituent components
(Salvador et al., 2002); we label this configuration-level customization. Second,
individual components comprising the configurations may be customized as well;
we label this component-level customization. The separation of the dimensions
became important after the introduction of modular product architectures (Starr,
1965), and since then, configuration-level customization has assumed an important
role in industrial practice (Pine, 1993; McCutcheon et al., 1994).
Limiting product customization to the configuration level increases respon-
siveness through three different mechanisms. First, when components are not order
specific, their procurement lead times do not influence delivery times of end prod-
ucts (Sheu & Wacker, 1997). Second, processing times of all component-level
manufacturing operations become irrelevant to delivery lead times (Feitzinger &
Lee, 1997; Su, Chang, & Ferguson, 2005). Third, component commonality reduces
the required level of production resource specialization, which in turn reduces the
need for internal buffers in the manufacturing processes (Fisher & Ittner, 1999).
We use the label custom assemblers in reference to manufacturers that produce
customized configurations from standardized components.
Standardization of components may not, however, be desirable or even pos-
sible in all MTO environments. Customer needs may be so idiosyncratic and their
Tenhiälä and Ketokivi 179

applications for the products so diverse that customization must be extended to the
component level (e.g., Robertson & Ulrich, 1998; Hegde et al., 2005). This applies
in our empirical context in particular. Complex machinery consists of hundreds
or thousands of tightly coupled components, which must interact seamlessly with
the other equipment that the customer uses in conjunction. Although manufactur-
ers of such products typically seek to restrict customization to the configuration
level as much as possible, they often find configuration-level customization to
be insufficient to meet all customer requirements. An industrial robot is a good
example of a product that cannot be completely modularized. This is because the
robot must be integrated, for instance, to the equipment that feeds and unloads it,
as well as to the software that controls the automation systems in the customer’s
factory. Furthermore, the grabbers and jigs of the robot, and often also its oper-
ating dimensions, must be customized according to idiosyncratic customer needs.
Similar component-level customization is often also required in many other capital-
intensive goods, such as industrial instruments, construction materials, and luxury
craft products. Component-level customization is probably the most prevalent in
those job shops that are subcontractors to other manufacturers and thus produce
items that are designed by their customers. Their concerns about the proliferation
of components may receive little attention among the product designers of the con-
tracting firm. In sum, there are many valid reasons for some MTO manufacturers
to remain purely custom producers, who assemble customized configurations from
(at least partly) customized components.
Finally, we must also consider the alternative that MTO products are sim-
ply not customized at all. Both configurations and components may be standard
for all customers, but the products are still made to order simply because they
are so costly that the manufacturer cannot commit to producing them until a firm
customer order is received. Producing to stock is not an option in business environ-
ments where demand is sporadic and products are expensive and perishable or have
very short life cycles (Weng & Parlar, 2005). We use the label non-customizers
for MTO manufacturers that produce standard configurations from standard
components.
Although much of the extant literature has treated product customization as
a unidimensional variable (e.g., Safizadeh, Ritzman, & Mallick, 2000; Sousa &
Voss, 2001; Sousa, 2003; Thirumalai & Sinha, 2009), there are also studies that
recognize multidimensionality, albeit only implicitly (Duray et al., 2000; Swafford,
Ghosh, & Murthy, 2006; Klein, 2007). Contemporary literature is limited in the
sense that the different ways of customizing products have not been used to explain
the boundaries of applicability of various operations management practices (Sousa
& Voss, 2008). This is in stark contrast with the classical organization theory
literature, where product variety—a construct similar to product customization—
has been incorporated as one of the most important contingency variables (e.g.,
Woodward, 1965; Hickson, Pugh, & Pheysey, 1969; Hayes & Wheelwright, 1979).
In this article, we incorporate the logic of the classical organization theory literature
by using the two dimensions of customization to arrive at contingency hypotheses
on the context-dependent applicability of order management practices in the MTO
context. In the following, we formulate these more elaborate hypotheses from the
baseline hypotheses using the customization gestalts.
180 Customization-Responsiveness Squeeze

Product Configurator Hypothesis


PC tools should be effective among custom assemblers, but since non-customizers
do not configure their products, PC tools should obviously not affect their product
conformance. However, according to the extant literature, PC tools should not be
effective for custom producers either, because their nonstandard components may
have a virtually unlimited variety of interfaces and thus, it should be impossible
to maintain reliable configuration rules (Forza & Salvador, 2006). The baseline
hypothesis H1a can thus be elaborated into the following contingency hypothesis:

H2a: Use of product configurator software is positively associated with


product conformance among custom assemblers.

Available-to-Promise Verifications Hypothesis


Information systems that feature ATP verifications require that lead-time parame-
ters be known in advance and maintained diligently within the software (Vollmann,
Berry, Whybark, & Jacobs, 2005; SAP, 2011b). This cannot be done with sufficient
accuracy if some components are unique. Even if a product has only a few unique
components, these components are the most likely to have the longest lead times.
This is due not only to the required additional engineering work but also because
the unique components may have to be ordered from noncustomary suppliers and
handled outside of the standard procurement processes. Consequently, the unique
components often determine the total lead time of the product, and ATP verifi-
cations may be completely ineffective even if the lead time parameters of 99%
of the components were known at the time of the order acquisition. Therefore,
ATP verifications should not be as effective for custom producers as they are for
non-customizers and custom assemblers. The baseline hypothesis H1b can thus be
elaborated into the following contingency hypothesis:

H2b: Use of available-to-promise verifications is positively associated with


delivery performance among (i) non-customizers and (ii) custom as-
semblers.

Configuration Management Hypothesis


In order for CM practices to be effective, product specifications should be diverse
across the customer base (Guess, 2002; PMI, 2006). If the only customer-specific
attributes are straightforward, such as order quantities, systematic CM practices
should not add much value. On the contrary, they might indeed constitute an
unnecessary burden on the order management system. Consequently, CM prac-
tices should not be effective for non-customizers whose product specifications are
standard. The baseline hypotheses H1c and H1d can thus be elaborated into the
following contingency hypotheses:

H2c: Use of configuration management practices is positively associated


with product conformance among (i) custom assemblers and (ii) custom
producers.
Tenhiälä and Ketokivi 181

H2d: Use of configuration management practices is positively associated


with delivery performance among (i) custom assemblers and (ii) custom
producers.

The three product customization gestalts and the contingency hypotheses are sum-
marized in Figure 1. We hypothesize the southeast quadrant to be empirically void;
it is difficult to fathom a context in which standard product configurations are built
from customized components.

RESEARCH DESIGN
We sought empirical insight on the hypotheses through a two-pronged field study.
First, we collected and analyzed quantitative survey data in the MTO manufac-
turing context. In addition to the quantitative data, we sought further insight by
interviewing order management experts in the MTO context. This was to ensure
that we had not overlooked any important contextual considerations that might
have an effect on the use of the order management practices in the three gestalts.
As the focus of this paper is clearly on the development of context-dependent
middle-range theory (Bourgeois, 1979), we had to take proper action to incorpo-
rate all important idiosyncrasies of the MTO environment. Operating exclusively
in a hypothetico-deductive mode was deemed inappropriate because such an ap-
proach might overlook important contextual idiosyncrasies.

Sample
In order to test the hypotheses empirically, we first identified an industry where
MTO manufacturing is relevant: manufacturing of complex industrial machinery
appeared to be a suitable option. Within this industry, we approached seven different
corporations. In collaboration with senior executives in these corporations, we
identified a number of MTO factories in their supply chains that would suit our
research needs. We ended up with a total of 73 factories (6–14 factories per supply
chain). Within these factories, we further identified a total of 180 MTO production
processes that were organized and managed separately from one another. These
production processes were chosen as the units of empirical analysis; both the order
management practices and the levels of product customization can be meaningfully
evaluated at the level of production process, not at the level of the entire factory,
let alone the corporation. In the spring of 2007, the managers responsible for the
operations of the selected production processes were invited to respond to a Web-
based survey. Due to the strong senior executive support, we obtained 163 valid
responses (response rate of 91%). Table 1 describes the sample.
Because the sample is carefully selected from a very specific theoretically
relevant industry, we wish to emphasize the context-dependence of the hypotheses,
the measurement instrument, as well as the eventual findings. Our approach should
be understood as an attempt to formulate and test what Glaser and Strauss (1967)
labeled substantive theory. In contrast with formal theory, substantive theory links
intimately to a specific context, and indeed, may not even be adequately understood
outside this specific context. Thus, the type of generalizability we seek in this article
Figure 1: Hypothesized contingency effects of order management practices.
182
Customization-Responsiveness Squeeze
Tenhiälä and Ketokivi 183

Table 1: Sample overview.


Supply Chain Geographic Studied Production Inter-
Products Scope Factories Processes views
Air defense artillery Global 14 20 4
Cannons, fire control units, and
related electronics
Aero-derivative power turbines Global 11 27 3
Turbines and auxiliary equipment
for power generation and secondary
recovery of oil and gas
Factory automation Europe & 10 27 3
Flexible manufacturing systems, the U.S.
robotized production cells, and
loading/deburring/measuring
stations for machine tools
Heavy-capacity industrial cranes Global 14 33 5
Process cranes for waste
management, paper, and steel indus-
tries, gantry and container cranes for
shipyards and harbors
Reactive power compensation systems Global 6 12 3
Capacitor banks, static
compensators, and harmonic filters
Remote-refrigerated display cabinets Northern & 11 25 5
for grocery retailers Eastern
Refrigerators, freezers, combination Europe
cabinets, and deli bars
Special-purpose elevators Global 7 19 7
Elevators for skyscrapers and ships,
and luxury elevator cars

Note: Unit of analysis: production process; total sample in the survey = 180; usable
responses = 163 (91%).

is dissimilar to the kind of generalizability (and formal theory) typically sought in


statistical research, in particular in research that spans multiple industries.

Measures
The dependent variables of our analysis, product conformance and delivery perfor-
mance, represent the two primary market requirements in MTO production. These
two variables also reflect operational capabilities at the production process level,
whereas performance in terms of many other important market requirements, such
as low price and global availability, depends more on the overall capabilities of
the entire firm. As the order management practices are typically specific to indi-
vidual production processes, we consider these two variables the most appropriate
measures for the effectiveness of order management.
Product conformance is measured as the ability of a production process to
satisfy varying customer requirements, which is a crucial capability in our context.
184 Customization-Responsiveness Squeeze

Many of the manufacturing processes in our sample manufacture capital-intensive


machinery, and thus, the deliverables have to be aligned with the quality criteria and
technical standards of the customer. In addition, the products have to be physically
compatible with other equipment that the customers use in tandem.
Delivery performance is measured through the main elements of the respon-
siveness of a production process (Szwejczewski, Mapes, & New, 1997). In our
context, delivery performance is just as important as product conformance, be-
cause even minor delays can be very costly to the customers whose own value
creation processes may depend in a fundamental way on the swift and timely
supply of equipment (Yeo & Ning, 2002).
We used the familiar factor-analytic approach to bridge the theoretical con-
structs to their respective empirical measures. Both performance dimensions as
well as the variables for configuration-level and component-level customization
were operationalized with three reflective indicators. The validity and reliability of
this measurement model were tested in a confirmatory factor analysis with Mplus
5.2 software using the robust maximum likelihood (MLR) estimator. Table 2 shows
the MLR estimates along with their robust standard errors (Yuan & Bentler, 2000).
The results appear adequate. First, the statistics are satisfactory in terms of con-
vergent validity; all items load significantly on their hypothesized factors, and the
standardized loadings are reasonable. Second, the composite reliability indices
indicate no problems with measurement reliability. Third, Fornell and Larcker’s
(1981) condition for discriminant validity is satisfied because the proportions of
average variance extracted are greater than the highest squared correlations be-
tween the variables. Fourth, the nonsignificant chi-square statistic and the other fit
indices indicate that the overall model fit is also adequate.
The three order management practices were operationalized using seven
indicators. The questionnaire items are listed in Table 3. In creating these variables,
we used the formative instead of the factor-analytic reflective indicator mode
(Bollen & Lennox, 1991). The reason for this is the fact that the indicators do not
reflect the underlying construct as much as they constitute it. Further, the individual
indicators that constitute the construct are not necessarily intercorrelated, that is,
the tools are not necessarily used in conjunction with one another. But the fact that
they are not necessarily used in conjunction with one another does not mean they are
unreliable or invalid measures. If the conventional factor-analytic reflective mode
were used, one would, however, ascribe such lack of correlation to unreliability and
invalidity of measurement. Two measures need not be correlated with one another
to be valid measures of a construct (Curtis & Jackson, 1962). For instance, the item
that addresses the confirming of delivery dates based on available manufacturing
capacity is a valid measure of ATP verifications, no matter how strongly or weakly
it correlates with the other items, say for instance, considerations of suppliers’
capacity utilization.
The formative-reflective distinction is important here, because it is crucial
not to impose unidimensionality, internal consistency, or convergent validity re-
quirements in the case of formative indicators. But as far as the empirical results
and hypothesis testing are concerned, we find there to be little difference: modeling
order management as reflective would lead to similar substantive results. This is
not to be taken as a general conclusion, however. The reason for observing identical
Tenhiälä and Ketokivi 185

Table 2: Validity and reliability analysis of the reflective measurement scales.


Standardized item loadings
“Evaluate the performance of your production process in comparison to [three most
important competitors]. . .” 1: much worse, 2: somewhat worse, 3: about similarly, 4:
somewhat better, 5: much better
Product conformance (CR = 0.79, AVE = 0.56)
“. . .in the quality of products” 0.78∗∗∗
“. . .in the technical performance of the products” 0.79∗∗∗
“. . .in the ability to satisfy customers’ requirements for the products” 0.67∗∗∗
Delivery performance (CR = 0.86, AVE = 0.68)
“. . .in the ability to confirm delivery dates for customers’ first 0.86∗∗∗
requirement dates”
“. . .in the ability to deliver on the confirmed delivery date” 0.85∗∗∗
“. . .in the average lead time from order acquisition to delivery” 0.75∗∗∗
“How do the following statements describe the operations in your production process?”
1: very poorly, 2: somewhat poorly, 3:moderately, 4: quite well, 5: very well
Component-level customization (CR = 0.76, AVE = 0.51)
“Our products are made of standard components” (reverse coded) 0.67∗∗∗
“Our products are configured from modules” (reverse coded) 0.78∗∗∗
“We can use inventories as buffers against peaks in demand” (reverse 0.69∗∗∗
coded)
Configuration-level customization (CR = 0.67, AVE = 0.41)
“Our production planning is based on product-specific customer orders” 0.54∗∗∗
“Our products are designed according to customers’ specifications” 0.81∗∗∗
“Our products have to be adapted to customers’ applications” 0.54∗∗∗
Latent variable correlations
Delivery performance–Product conformance 0.24∗
Delivery performance–Component-level customization 0.21
Delivery performance–Configuration-level customization 0.10
Product conformance–Component-level customization 0.31∗∗
Product conformance–Configuration-level customization 0.28∗
Component-level customization–Configuration-level customization 0.17

Notes: χ 2 = 64.04, degrees of freedom (d.f.) = 48, p = .061, χ 2 /d.f. = 1.33. Comparative
fit index (CFI) = 0.962. Non-normed fit index (NNFI) = 0.948. Root mean square error of
approximation (RMSEA) = 0.050 [90% confidence interval: 0.000–0.080]. Standardized
root mean square residual (SRMR) = 0.061. CR: composite reliability; AVE: average
variance extracted.
∗∗∗
p < .001; ∗∗ p < .01; ∗ p < .05.

results is that in our data, we find that the indicators of the formative constructs
do indeed correlate rather strongly with one another (Table 3). But from the point
of view of assessing the measurement properties of the indicators, this finding is
ancillary: the choice between the formative and the reflective mode is to be made
on logical not empirical grounds.

Evaluation of Perceptual Measures


As the empirical measures are based on perceptions, we performed two kinds of
analysis to further evaluate their validity and reliability. First, in order to estimate
186

Table 3: Descriptive statistics and intercorrelations of the order management measures.


“How do the following statements describe the order management in your production
process?” 1: very poorly, 2: somewhat poorly, 3: moderately, 4: quite well, 5: very well μ σ PC ATP1 ATP2 ATP3 CM1 CM2
PC “We use product configurator software” 2.62 1.55
ATP1 “Delivery dates are confirmed on the bases of available manufacturing capacity” 3.49 1.35 −.19∗
ATP2 “Availability of raw materials is ensured before delivery dates are confirmed” 2.69 1.32 −0.15 0.51∗∗∗
ATP3 “Suppliers’ capacity utilization is considered when delivery dates are confirmed” 2.77 1.43 −0.12 0.61∗∗∗ 0.70∗∗∗

CM1 “Order fulfillment processes are managed as projects” 3.55 1.26 0.20 0.06 0.27∗∗ 0.21∗
CM2 “All changes to project plans are documented” 3.20 1.23 0.19∗ −0.10 0.14 0.14 0.58∗∗∗
CM3 “Deviations from project plans are managed in a variation management process” 3.19 1.30 0.26∗∗ 0.07 0.21∗ 0.17 0.54∗∗∗ 0.66∗∗∗

p < .05; ∗∗ p < .01; ∗∗∗ p < .001.
Customization-Responsiveness Squeeze
Tenhiälä and Ketokivi 187

interrater agreement (IRA), we used two informants for one production process in
each supply chain. The IRA ratios (James, Demaree, & Wolf, 1984) were 0.75 for
the PC tool variable, 0.75 for ATP verifications, 0.67 for CM practices, 0.96 for
configuration-level customization, 0.68 for component-level customization, 0.98
for product conformance, and 0.98 for delivery performance.
Second, we also had access to objective data on the on-time delivery per-
formance, which enabled an examination of criterion validity. Comparing the
perceptual measure with the objective measure was possible in 38 production
processes. In this subsample, the correlation between the two measures was 0.61
(p < .001), which provides criterion validity for the perceptual delivery perfor-
mance measure. Because objective data were available only for 38 observational
units, we had to use the perceptual measure in the final analysis. At the same time,
the perceptual measure may actually be more valid than the objective measure
because organizations in the sample had slightly different definitions for on-time
delivery; some calculated it at the level of items and others at the level of deliveries
or orders. These semantic differences would make the validity of the objective
measure questionable. The perceptual measure—where informants are asked to
evaluate their relative position to their competitors—does not suffer from the same
validity problem. Further, the objective percentage-based timeliness measure is
not as content valid because it does not register the length of the delay of a late
delivery, which is obviously an extremely important component of timeliness.

Qualitative Data
To further enhance the grounding of the theory, we wanted to give a voice to
the practitioners of the empirical context. Toward this end, we conducted 30
semistructured interviews lasting 1–2 hours each. The interviewees worked in sim-
ilar positions as the informants of the survey (e.g., master schedulers and process
managers). The objective was to gain a deeper understanding of how practitioners
actually configured products, determined delivery dates, and managed changes to
customer order documents. The interviews uncovered reasons behind choices to
implement specific tools, as well as the challenges these practitioners had faced in
using them. In addition to the interviews, we obtained further qualitative data from
internal documents as well as 34 site visits, which gave us an opportunity to observe
the daily operating routines. We also obtained feedback from three workshops ar-
ranged to discuss the results of the survey with representatives of the sample firms.
Overall, the goal of the qualitative part was to complement the quantitative hypoth-
esis testing with rich contextual data and to uncover potential contingency factors
that the hypothetico-deductive method might have overlooked.

SURVEY RESULTS
The contingency hypotheses were tested using the fit-as-gestalts operationalization
discussed by Venkatraman (1989). Specifically, the sample was grouped into three
gestalts according to the product customization scores of each observational unit.
We used two methods in forming the gestalts: theoretical and empirical (Miller,
1981). In the former, we split both dimensions of customization at the middle of
188 Customization-Responsiveness Squeeze

their scales so that those who scored low on both dimensions were categorized as
non-customizers and those who scored high on the configuration-level customiza-
tion were divided into custom assemblers and custom producers based on their
scores in the component-level customization. In the latter approach, we used the
IBM SPSS 19 software to create three gestalts with a hierarchical cluster analysis
based on the squared Euclidean distances of the between-group average linkages of
each observational unit’s scores on the two scales. The choice of the method only
influenced the gestalt membership of four observational units, and the methods
yielded substantively similar results in the hypothesis testing. Thus, we will only
report the results based on the theoretical gestalt formulation.

Product Customization Gestalts


Figure 2 summarizes the product customization gestalts and gives examples of
each. As expected, the quadrant of customized components and standard config-
urations is empty. Another observation is the imbalance in the distribution of the
data points: only 14% of the data is located in the non-customizer quadrant. This,
we concluded, is simply a characteristic of the sample: while MTO operations
are not necessarily associated with configuration-level customization, there seems
to be a drift towards configuration-level customization in this specific sample. In
contrast, we observed more variance in the component-level dimension: 87 obser-
vational units have a low and 64 a high level of component customization. In sum,
the vast majority of observational units in our sample are either custom assemblers
or custom producers.

Hypothesis Testing
Table 4 provides the descriptive statistics and the intercorrelation matrix of the
variables. All theoretical variables were constructed by calculating the average of
their respective indicators. For control variables, we tried a number of demographic
measures as proxies of technological and organizational complexity, but the only
measure to have any effect on the dependent variables was the production volume
of the process.
Due to the nested nature of the sample, we tested the hypotheses in a lin-
ear mixed model with fixed effects for the independent variables and a random
intercept for the seven supply chains within which the observational units were
nested. We used the restricted maximum likelihood estimation method (Patterson
& Thompson, 1971), embedded in IBM SPSS 19, to accommodate the uneven
group sizes (Harville, 1977); there were 12–33 processes per supply chain. We
first estimated a null model without any predictors to partition the variance in the
dependent variables into a process-level (within-groups) and supply chain-level
(between-groups) components. The former would then be used in the subsequent
steps as a benchmark in calculating a surrogate for the R2 statistic (Snijders &
Bosker, 1999). The null models indicated that, on average, supply chain-level vari-
ability accounted for 12% of the variance in product conformance and 30% of
the variance in delivery performance. These percentages are significant enough to
warrant a multilevel analysis to ensure unbiased estimation.
Table 5 shows the results of the analyses on the product conformance variable.
The first step shows the null models, the second the effect of the control variable,
Figure 2: Contents of the product customization gestalts.
Tenhiälä and Ketokivi
189
190 Customization-Responsiveness Squeeze

Table 4: Descriptive statistics and intercorrelations of all variables.


Variables μ σ 1 2 3 4 5 6 7
1 Product 3.78 0.71
conformance
2 Delivery 2.99 0.78 0.20∗
performance
3 Production 3498 6429 −0.03 −0.02
volumea
4 Product 2.62 1.55 0.12 0.07 0.13
configurator
tool
5 Available-to- 3.02 1.21 0.21∗∗ 0.32∗∗∗ −0.24∗∗ −0.14
promise
verification
6 Configuration 3.29 1.11 0.10 0.50∗∗∗ −0.09 0.26∗∗ 0.20∗
management
practices
7 Component-level 3.19 0.87 0.24∗∗ −0.11 −0.01 0.03 −0.03 −0.05
customization
8 Configuration- 4.06 0.87 0.14 0.11 0.09 −0.19∗ 0.18∗ −0.06 0.09
level
customization

Note: a Logarithmic transformation is used in the analyses but means and standard deviations
of this variable are shown untransformed.

p < .05; ∗∗ p < .01; ∗∗∗ p < .001.

and the third, the effects of the theoretically relevant variables. The models are
compared on the basis of two statistics: (i) the likelihood information criteria show
deviance scores, where a statistically significant difference indicates that the added
variables significantly improve the model fit; (ii) the surrogate for the R2 statistic
describes the reduction in the within-group residual variance.
The results in Table 5 show that the baseline hypothesis H1a on the universal
positive effect of the PC tools is empirically supported, but so is its elaborated form
H2a. In addition to the expected contingency effect among custom assemblers, the
configurators have a nonhypothesized positive effect in the custom producer gestalt.
Neither hypothesis on the product conformance effect of the CM practices (H1c
and H2c) is supported, but the ATP verifications turn out to have a nonhypothesized
positive effect in the full sample and the custom assembler gestalt.
Table 6 shows the results on delivery performance. The baseline hypothesis
H1b on the universal effect of ATP verifications is rejected, but its contingency
hypothesis H2b is supported. The delivery performance effects of the CM practices
are supported both universally (H1d) and in the gestalts of custom assemblers and
custom producers (H2d).
We also compared the coefficients of the three contingency effects that were
significant in two different gestalts. For this purpose, we used the unstandardized
coefficients and their standard errors to calculate z-test statistics, as described by
Paternoster et al. (1998). The difference in the effect of the CM practices on delivery
performance among custom assemblers and custom producers was approaching
Tenhiälä and Ketokivi 191

Table 5: Results on product conformance.


Model 1 Model 2 Model 3
Entire sample
Intercept 3.76∗∗∗ (0.11) 3.36∗∗∗ (0.21) 2.74∗∗∗ (0.36)
Production volume 0.07∗∗ (0.03) 0.08∗ (0.03)
Product configurator tool (H1a) 0.12∗∗ (0.04)
Available-to-promise verifications 0.14∗ (0.06)
Configuration management −0.04 (0.06)
practices (H1c)
−2 restricted log likelihood 334 282 [52∗∗∗ ] 245 [37 ∗∗∗ ]
[difference]
Surrogate R2 .19 .19
Non-customizers
Intercept 3.56∗∗∗ (.20) 2.82∗∗∗ (0.49) 0.34 (0.84)
Production volume 0.19∗ (0.08) 0.65∗∗ (0.12)
Product configurator tool −0.03 (0.12)
Available-to-promise verifications 0.06 (0.11)
Configuration management 0.10 (0.18)
practices
−2 restricted log likelihood 59 34 [25∗∗∗ ] 21 [13∗∗ ]
[difference]
Surrogate R2 .62 .89
Custom assemblers
Intercept 3.73∗∗∗ (0.11) 3.56∗∗∗ (0.30) 2.23∗∗∗ (0.48)
Production volume 0.02 (0.04) 0.05 (0.05)
Product configurator tool (H2a) 0.15∗ (0.07)
Available-to-promise verifications 0.20∗ (0.08)
Configuration management 0.04 (0.09)
practices (H2ci )
−2 restricted log likelihood 131 128 [3] 114 [14∗∗ ]
[difference]
Surrogate R2 0.04 0.19
Custom producers
Intercept 3.85∗∗∗ (0.13) 3.63∗∗∗ (0.26) 2.94∗∗∗ (0.53)
Production volume 0.05 (0.04) 0.04 (0.04)
Product configurator tool 0.16∗∗ (0.06)
Available-to-promise verifications 0.12 (0.08)
Configuration management −0.03 (0.07)
practices (H2cii )
−2 restricted log likelihood 145 126 [19∗∗∗ ] 108 [18∗∗∗ ]
[difference]
Surrogate R2 .19 .25

Note: Unstandardized coefficients, standard errors in the parentheses.



p < .05; ∗∗ p < .01; ∗∗∗ p < .001.

significance (p < .10), but the other two were not. Despite this, the value of the
contingency perspective is evident in the results. Relying solely on the universal
effects would have led to a false conclusion about the effectiveness of the practices
in specific product customization gestalts.

Possible Effects of Common Method Variance


Because we collected the data on both the dependent and the independent vari-
ables from the same informants, it is possible that common method variance
192 Customization-Responsiveness Squeeze

Table 6: Results on delivery performance.


Model 1 Model 2 ]Model 3
Entire sample
Intercept 2.97∗∗∗ (0.17) 2.94∗∗∗ (0.26) 1.64∗∗∗ (0.35)
Production volume 0.01 (0.03) 0.00 (0.03)
Product configurator tool 0.02 (0.04)
Available-to-promise verifications 0.06 (0.05)
(H1b)
Configuration management 0.34∗∗∗ (0.05)
practices (H1d)
−2 restricted log likelihood 332 303 [29∗∗∗ ] 227 [76∗∗∗ ]
[difference]
Surrogate R2 0.03 0.29
Non-customizers
Intercept 2.77∗∗∗ (0.20) 2.65∗∗∗ (0.52) 0.83 (1.14)
Production volume 0.03 (0.09) 0.14 (0.16)
Product configurator tool 0.02 (0.16)
Available-to-promise verifications 0.40∗ (0.13)
(H2bi )
Configuration management 0.04 (0.24)
practices
−2 restricted log likelihood 51 36 [15∗∗∗ ] 22 [14∗∗ ]
[difference]
Surrogate R2 0.34 0.64
Custom assemblers
Intercept 3.12∗∗∗ (.16) 3.21∗∗∗ (0.36) 1.49∗∗∗ (0.49)
Production volume −0.02 (0.05) 0.03 (0.05)
Product configurator tool −0.03 (0.07)
Available-to-promise verifications 0.23∗∗ (0.09)
(H2bii )
Configuration management 0.23∗ (0.09)
practices (H2di )
−2 restricted log likelihood 137 134 [3] 116 [18∗∗∗ ]
[difference]
Surrogate R2 .02 .24
Custom producers
Intercept 2.93∗∗∗ (0.25) 2.89∗∗∗ (0.36) 1.94∗∗∗ (0.44)
Production volume 0.00 (0.04) −0.04 (0.03)
Product configurator tool 0.07 (0.04)
Available-to-promise verifications −0.12 (0.06)
Configuration management 0.41∗∗∗ (0.05)
practices (H2dii )
−2 restricted log likelihood 150 144 [6∗ ] 80 [64∗∗∗ ]
[difference]
Surrogate R2 0.00 0.62

Note: Unstandardized coefficients, standard errors in the parentheses.



p < .05; ∗∗ p < .01; ∗∗∗ p < .001.

(CMV) is present. The data passed the commonly used Harman’s single factor
test (Podsakoff & Organ, 1986), and the check against the objective measure of
delivery performance further alleviates the concern for method bias. However,
CMV can never be proved completely absent (Podsakoff, MacKenzie, Lee, &
Podsakoff, 2003), and thus it is important to understand its possible effects. For
Tenhiälä and Ketokivi 193

this purpose, we used the heuristic of Siemsen et al. (2010, p. 467) to deter-
mine whether the effect of the possible CMV in the data had inflated or deflated
the estimated coefficients. The point in which the inflating effect turns into a
deflating one depends on the correlations among the studied variables and the
number of independent variables. With three independent variables, the aver-
age correlations between the independent and the dependent variables should
be at least 0.33 to ensure that the effect is deflating. In addition, the correlations
among the independent variables should be less than 0.30 for the heuristic to be
reliable.
We studied the correlations in each gestalt to see whether the required con-
ditions hold. The requirement for the correlations of the independent variables
to be below 0.30 holds in each gestalt. The adherence to the other requirement
depends on the gestalt and the dependent variable. Three of the hypothesized con-
tingency effects readily pass the 0.33 threshold, and thus in the possible presence
of CMV, our results are actually conservative estimates of the true relationships.
These results are the effects of ATP verifications on delivery performance among
non-customizers (H2bi ) and custom assemblers (H2bii ) as well as the effect of
CM practices on delivery performance in the custom assembler gestalt (H2di ).
The correlations of the variables with regard to three other contingency effects are
between 0.30 and 0.31, falling just short of the 0.33 threshold. Thus, in the possible
presence of CMV, the coefficients may be slightly inflated. These are (i) the effect
of PC tools on product conformance among custom assemblers (H2a), (ii) the non-
hypothesized effect of ATP verifications on product conformance in the custom
assembler gestalt, and (iii) the effect of CM practices on delivery performance
among custom producers (H2dii ). The only significant contingency effect to fall
further from the required threshold is the nonhypothesized relationship between
PC tools and product conformance in the custom producer gestalt (correlation =
.15). Thus, if the amount of CMV has been considerable, then this finding is at
risk of Type I error.

INTERPRETATION OF SURVEY RESULTS AND


COMPLEMENTARY OBSERVATIONS FROM
QUALITATIVE INQUIRY
The statistical results have several implications. The finding that different order
management practices are effective in different product customization gestalts is
particularly important. The fact that different practices influence different dimen-
sions of performance is also relevant. In the following section, we further discuss
and interpret the results of the statistical analyses and augment the interpretations
in light of the qualitative data.

Observations about Product Configurators


The most important finding with regard to the use of PC tools was that they
were not only very often used by custom producers but they also exhibited a
nonhypothesized positive effect in that gestalt. The reason for this outcome was
discovered in the interviews: notwithstanding the fact that the components of the
194 Customization-Responsiveness Squeeze

one-off products were not standardized and thus the selection of the configurations
could not be completely automated, the tools were indeed useful in conducting
feasibility checks at a higher level. Specifically, instead of using the configurator
to ensure the compatibility of individual components, it can be used to check the
compatibility of certain “bundles of features.” This finding is consistent with the
literature on design rules (Baldwin & Clark, 2000); even if rules cannot be defined
for direct choices between explicit alternatives (i.e., product components), they
can be employed as “metaroutines” of sorts, which may prove useful in choices
between groups of alternatives (i.e., product features).

Observations about Available-to-Promise Verifications


The quantitative analyses led to three observations about ATP verifications. First,
the survey data showed that ATP verifications were used less frequently than the
academic literature suggests. This observation was supported by the interviews.
Half of the interviewees (15 out of 30) admitted that they relied on fixed de-
livery lead-time quotes. Most interestingly, 8 of these 15 who did not use ATP
verifications were custom assemblers. This is somewhat surprising, given that in
this gestalt, ATP verifications were both hypothesized and found to be effective.
Could it be that some practices, while seemingly old-fashioned, are indeed ef-
fective in the contemporary business environments? As an analogy, organization
theorists have made similar observations about functional organization structures:
out of fashion perhaps, but still very effective in many environments (Galbraith,
2005).
The second interesting finding was that ATP verifications had a nonhypoth-
esized positive effect on product conformance in the custom assembler gestalt
(which was also reflected as an effect in the entire sample). This could be due to
the fact that when ATP verifications are used to ensure feasibility of the promised
delivery date, it is much less likely that the order fulfillment process will have
to face unexpected time pressures. The interviewees from processes that used
ATP verifications indeed complained less about time pressures. Site visits further
supported our impression that processes were better under control when ATP ver-
ifications were used. Since mistakes are easily made under time pressure, it is
actually understandable that ensuring the feasibility of the delivery dates has a
positive effect on product conformance.
The third interesting observation was that many custom producers used ATP
verifications even though we neither hypothesized nor found them to be effective
in that gestalt. The interviewees explained that it is indeed extremely difficult to
promise accurate delivery dates in fully customized production; at the time of
sales negotiations with the customers, no-one in the organization has a perfect
understanding of the exact amounts of the work required. The interviewees did
maintain, however, that reliable delivery date promises are as crucial in one-off
production as in any other MTO environment. Hence, the observed ineffectiveness
of ATP verifications in custom production should not be interpreted as evidence
that verifications are futile. On a positive note instead, software providers should
be encouraged to develop verification methods that help overcome the ambiguities
of the one-off production. It could also be that certain kinds of “metaroutines,”
Tenhiälä and Ketokivi 195

similar to the ones that custom producers used in PC tools, could also be applicable
in ATP verifications.

Observations about Configuration Management Practices


Two observations about the CM practices merit attention. First, the practices have a
significant influence on delivery performance both in the custom producer and the
custom assembler gestalts. Extant literature has often considered custom assembly
to be linked to mass customization and as such, to be an extension of mass produc-
tion. At the same time, CM practices have been associated with and solely studied
within the context of project business (e.g., Harter, Krishnan, & Slaughter, 2000;
Shenhar, 2001). Consequently, custom assemblers seem to have downplayed the
importance of managing product configurations and delivery schedules systemat-
ically in the order fulfillment processes. The interviewees, however, specifically
emphasized the importance of the systematic maintenance of order documents in
both gestalts. Out of the 27 interviewees that belonged either to custom assembler
or custom producer gestalts, over half (14) were satisfied with their CM practices;
those remaining (13) thought that their practices were not systematic enough.
Interestingly, 10 out of the 13 unsatisfied were custom assemblers.
The importance of CM practices arises from two characteristics that are
common to all kinds of customized production: product specifications depend on
customer needs and there is always some lead time between order acquisition and
product delivery. Their combined effect is that customer needs may change during
the order fulfillment phase. If order documents are not well maintained, those who
receive change requests will have trouble evaluating the impact of these changes
on both the feasibility of the configuration as well as the promised delivery date.
The key is to strike a balance, because automatically declining all change requests
will have negative effects on product conformance, but at the same time, accepting
all requests will jeopardize delivery performance.
Interviewees argued that change requests were indeed often accommodated
without sufficient understanding of their implications. This in turn had an ad-
verse effect on planned delivery schedules in particular. The underlying reason for
accommodating change requests was also familiar: sales and customer service per-
sonnel are pressured to exhibit flexibility toward the customers, and in the absence
of compelling reasons to decline, change requests will usually be accepted. But if
these personnel had at their disposal a systematic procedure for managing order
amendments and evaluating their consequences, they would be able explicitly to
assess the consequences of the requested changes, to revise delivery schedules
accordingly, and to communicate the new delivery dates to the customers.
This last observation also explains the second important finding regarding
the CM practices, namely, their nonsignificant effect on product conformance.
The interviewees’ comments on the general tendency to accept change requests
at the cost of delivery performance suggest that the adherence to the changing
specifications is so important that it will be ensured, with or without systematic
CM procedures. Thus, the only benefit of the CM practices is that it is easier to
preserve delivery performance when processing the changes that would be done
meticulously even in the absence of systematic procedures.
196 Customization-Responsiveness Squeeze

General Observations on Technological Constraints and “Best Practices”


We also asked the interviewees why someone might use a practice that is intuitively
ineffective (e.g., a PC tool in the non-customizer gestalt). The answers generally
related to technological constraints. For example, if a PC tool is embedded in
the ERP system, the system simply enforces its use. Of course, it is possible to
define multiple procedures in a single ERP system so that processes with different
needs can use the system differently, even within the same factory. However,
the interviewees appeared to be generally unaware of such opportunities. The
interviewees further pointed out that the uniformity of practices is often enforced
by promoting the procedures embedded in the ERP systems—like many other
concepts and tools in operations management—as “best practices.” Consequently,
they are considered universally beneficial, and any deviations from their uniform
implementations are strongly discouraged in the name of standardization and
reduction of variation.
We made two interesting further observations in the survey data regarding
potential technological and institutional constraints in multigestalt facilities, that
is, facilities that hosted processes from at least two different gestalts. For example,
one factory in the air defense artillery supply chain housed one process for the
production of cannons and another for the turrets into which the cannons were
installed. Although the cannon itself was completely non-customized, the turrets
had to be fully customized to make them fit the countless different platforms and
vehicles on which the end customers mounted the turret. The processes located in
these kinds of multigestalt facilities seemed to run a special risk of using nonfitting
order management practices.
We found, on the one hand, that there were a total of 45 processes that used
at least one practice, which—according to our theory and empirical evidence—did
not fit the specific gestalt. Interestingly, 20 of these 45 processes were located in
a multigestalt facility where another process used the same practice and further,
represented a gestalt in which the practice was indeed effective. On the other
hand, a significant number of processes exhibited the same phenomenon, but in
reverse: in 40 processes, we identified at least one practice that would have been
effective for the focal process but ineffective for another process hosted by the
same multigestalt facility. Out of these 40 processes, 27 did not use this practice,
even though it would have been effective. These two observations suggest that
processes in multigestalt facilities may be subject to technological and institutional
conformity pressures, which may come in the face of process- and gestalt-specific
effectiveness considerations. At a general level, this finding is consistent with
classic work on contingency theory, which has found that organizational structures
and processes are significantly shaped by organizational technologies, not the other
way around.

DISCUSSION
Glaser and Strauss (1967) noted that the development of abstract and generaliz-
able theoretical insight (formal theory) often emerges through the development
of context-dependent substantive theory. Glaser and Strauss further argued that
Tenhiälä and Ketokivi 197

development of substantive theory is crucial for a scientific discipline to be relevant


to practice: substantive theories are theories about practice in a specific context. In
this article, we have embraced this premise and accordingly, developed a substan-
tive theory of order management in the context of MTO manufacturing of complex
products. Our inquiry started with the conventional wisdom on order management
and then progressed toward a more context-dependent contingency theory. The
resulting combination of hypothetico-deductive and inductive analysis sheds light
on the specific problems of and solutions to the customization-responsiveness
challenge (McCutcheon et al., 1994). In particular, in our qualitative part, we
found that an exclusively hypothetico-deductive mode may lead to an oversight
of important insights. At the same time, conventional inductive research designs,
epitomized by Eisenhardt’s (1989) seminal paper, may not be as effective because
the phenomenon of interest in our study is not so much novel or under-researched
as it is established but idiosyncratic; there is a lot of a priori theory that can be
exploited. The mixed-methods approach used in this paper gave us the best of
both worlds: an abundance of extant theoretical insight, which was useful in the
hypothetico-deductive phase, and a rich qualitative database from which we were
able to uncover inductively insights that would have been impossible to uncover
through deductions from extant theory.
Our inquiry reveals the multilevel nature of customization and proposes
how different approaches to customization influence the effectiveness of order
management practices in MTO production. Our general finding is consistent with
established theories of organization design, which generally link requisite infor-
mation processing capacity to task uncertainty (Galbraith, 1973). Tushman and
Nadler (1978) further extended this general association by suggesting other fac-
tors that affect the organization’s requisite capacity to process information. We
now know that factors such as task characteristics (Haas, 2010), task environ-
ment (Flynn & Flynn, 1999), and interunit interdependence (Gattiker & Goodhue,
2005) all link to requisite information-processing capacity. Our results extend this
research by showing that product customization is yet another important factor that
links to requisite information processing capacity. Specifically in the context of
MTO manufacturing, we have found that non-customizers have the lowest requisite
information-processing capacity and can rely solely on ATP verifications. Custom
assemblers, in turn, have higher requisite information processing capacity, which
can be met by also employing PC tools and CM practices. Finally, custom pro-
ducers have the highest requisite information processing capacity and need all the
solutions and tools available. Indeed, their information-processing needs may not
be fully satisfied with existing solutions. Consequently, they must seek novel so-
lutions to their intensive and simultaneously idiosyncratic information-processing
challenges.
We observed in particular that custom producers had problems with the ap-
plication of ATP verifications. Turning to the organization design literature here
may lead to insights. Specifically, Galbraith (1973) proposed that information-
processing capacity can be enhanced not only by investing in information systems
(e.g., ATP verification tools) but also by improving lateral cross-functional re-
lationships. One solution to tackle the custom producer’s problem would be to
assign liaisons to operate between sales, engineering, and production functions.
198 Customization-Responsiveness Squeeze

The task of these liaisons would be to collect information on the lead times of
the unique components, and based on this information, determine feasible delivery
dates. Alternatively, use of cross-functional teams could provide a similar liaison
or integration effect.

Implications to Practice
The qualitative data in particular led us to the conclusion that the question we
asked in the beginning of the article is important to the practitioners and that
academic research can further inform practitioners on potential solutions. The
interviews also provided us with a reality check of sorts by revealing that extant
academic research on order management may have been ignorant of some of the
central contingencies and thus, resulted in simplistic prescriptions. We hope that
the qualitative insights of this study demonstrate that discussing and debating the
results of academic research with practitioners is one effective remedy to simplistic
prescriptions (Mohrman, Gibson, & Mohrman, 2001).
The MTO manufacturers of our sample constantly face three critical ques-
tions in their daily operations: (i) What kind of a product will be offered to meet
the customer’s idiosyncratic needs?; (ii) What is a feasible promised delivery
date?; (iii) How will change requests after order entry be managed? PC tools, ATP
verifications, and CM practices all help managers as they seek answers to these
questions. At the same time, the applicability of each tool depends fundamentally
on how exactly the products are customized.
To be sure, some of the empirical findings will appear trivial to the prac-
titioner. For instance, it is hardly surprising to observe that non-customizers do
not benefit from using PC tools. But other observations do challenge conventional
managerial wisdom. The fact that custom producers (in addition to custom assem-
blers) benefit from using PC tools is neither discussed in the extant literature nor
immediately obvious. Other intriguing observations include the notion that custom
assemblers may have been too hasty to abandon ATP verifications and that they
can also benefit from CM practices.
In addition to specific technical observations listed above and in the results
section, we wish to highlight a more general message to the practitioner: MTO
environments differ in terms of how products are customized, and these differ-
ences influence the day-to-day management of MTO production processes. This
is particularly crucial in benchmarking efforts. Managers must understand which
customization gestalt their processes belong to in order to choose appropriate
benchmarking partners. For example, a custom assembler benchmarking a non-
customizer might be seriously misled into thinking that they can get by without
implementing CM practices.
Another more general message from the results is a word of caution about
“best practices.” Even in our rather narrow segment of MTO manufacturing, one
cannot conclude that certain solutions would “work for most companies.” This
is because different customization gestalts either necessitate completely different
order management practices or the practices must be applied differently in different
gestalts. Managers who choose their practices based on a careful analysis of their
own idiosyncratic environments will not be misled. But those managers who rely on
Tenhiälä and Ketokivi 199

general prescription from, say, ERP system consultants, can be misled, especially
if the proposed solutions have a one-size-fits-all flavor to them.

Limitations and Further Research


There is only so much that can be done in a single study, and certain opportunities
for further contributions arise from the limitations of this study. First, additional
work could be done in the domain of measuring the order management variables.
The operationalizations of the present study are relatively rough and do not dis-
tinguish between different ways of doing the ATP verifications, for instance. ATP
methods vary from rudimentary checks of available slots in master schedules to
sophisticated simulation algorithms that reschedule production plans to accom-
modate and expedite new orders. It could be that different methods have different
effects, or they may work differently in different operating environments. The
same may also apply to the other two variables. For example, PCs embedded
in ERP systems may be better or worse than standalone PC tools. Alternatively,
the effectiveness of both solutions could depend on the context of their applica-
tion. Similarly, the CM practices can be managed either manually or within an
information system, and this choice may well influence their effectiveness.
Another possible avenue of further research is in the variables’ causal link-
ages, which cannot be truly tested in a cross-sectional study. For example, when we
see the positive relationship between ATP verifications and delivery performance,
one interpretation is that ATP verifications are beneficial. However, one could just
as well argue that good delivery performance enables the use of ATP verifications.
The true direction of the relationship can only be tested in longitudinal studies of
implementations of these techniques.
Third, this study was conducted in a very specific context of MTO manufac-
turing of complex machinery. It would be interesting to test how context-dependent
our findings are. In further studies, one might look into other industrial products
but also the production processes for services. Specifically, recent research has im-
plied that the multilevel nature of customization applies to services as well (Voss
& Hsuan, 2009). Further, the order management tools examined in this article are
by no means limited to manufacturing contexts, but are indeed applicable in ser-
vice operations as well. The CM practices actually originated in project business
where the service content of deliverables is normally high (Lyon, 2004). Similarly,
sophisticated PC tools are becoming available to service contexts too (Ardissono
et al., 2003). The same may not yet apply to systematic ATP verification tools, but
accurate delivery date promising would certainly be an advantage in many service
processes. Consequently, it could be studied how our findings apply in service
processes.
In extending these research questions to other contexts, one must recog-
nize the context-dependence of measurement. Our measures for component- and
configuration-level customization, for instance, are not meant here as general
measures of customization. Rather, they are measures that were formulated to ad-
dress customization in the context of MTO manufacturing of relatively complex
products. The use of buffer inventories, for instance, is not in any general sense a
measure of component-level customization. In the context of MTO manufacturing,
200 Customization-Responsiveness Squeeze

however, it works well as an indicator of component-level customization: use of


buffer inventories for components is—quite understandably—a strong indication
of a low level of component-level customization.

CONCLUSION
A detailed empirical inquiry often reveals determinants of performance to be much
more complex and context-specific than we assume (March & Sutton, 1997). In this
article, we have examined the relationship between product customization, order
management, and operational performance in MTO manufacturing. We have found
that from the point of view of trying to understand performance determinants, we
must first understand that MTO manufacturing subsumes very heterogeneous op-
erations where customization has different meanings. Important differences arise
from the fact that MTO products can be customized at two different levels: com-
ponent and configuration. This gives rise to three product customization gestalts
with significantly different information-processing needs. Specifically, we have
found the order management challenges of custom producers, custom assemblers,
and non-customizers to be quite diverse. In our empirical analysis of the MTO
manufacturing of complex industrial products, we have further found that these
contingencies have fundamental implications to the effectiveness of various order
management tools and practices. Our study shows that a detailed understanding
of the nature of product customization is required if we wish to understand how
orders should be managed in MTO manufacturing and how the performance of
MTO production processes is ultimately determined.

REFERENCES
Ardissono, L., Felfernig, A., Friedrich, G., Goy, A., Jannach, D., Petrone, G., et
al. (2003). A framework for the development of personalized, distributed
web-based configuration systems. AI Magazine, 24(3), 93–108.
Baldwin, C. Y., & Clark, K. B. (2000). Design rules: The power of modularity.
Cambridge, MA: The MIT Press.
Barut, M., & Sridharan, V. (2005). Revenue management in order-driven produc-
tion systems. Decision Sciences, 36(2), 287–316.
Bertrand, J. W. M., Zuijderwijk, M., & Hegge, H. M. H. (2000). Using hierarchical
pseudo bills of material for customer order acceptance and optimal material
replenishment in assemble to order manufacturing of non-modular products.
International Journal of Production Economics, 66(2), 171–184.
Bixby, A., Downs, B., & Self, M. (2006). A scheduling and capable-to-promise
application for Swift & Company. Interfaces, 36(1), 69–86.
Bollen, K. A., & Lennox, R. (1991). Conventional wisdom on measurement: A
structural equation perspective. Psychological Bulletin, 110(2), 305–314.
Bourgeois, L. J., III. (1979). Toward a method of middle-range theorizing. Academy
of Management Review, 4(3), 443–447.
Tenhiälä and Ketokivi 201

Configure One’s. (2011). Configure one’s product configurator, ac-


cessed April 13, 2011, available at http://www.configureone.com/
Products_configurator.aspx.
Curtis, R. F., & Jackson, J. (1962). Multiple indicators in survey research. American
Journal of Sociology, 68(2), 195–204.
Daft, R. L. (2004). Organization theory and design (8th ed.). Mason, OH:
Thomson/South-Western.
Danese, P., & Romano, P. (2004). Improving inter-functional coordination to face
high product variety and frequent modifications. International Journal of
Operations & Production Management, 24(9), 863–885.
Duray, R., Ward, P. T., Milligan, G. W., & Berry, W. L. (2000). Approaches
to mass customization: Configurations and empirical validation. Journal of
Operations Management, 18(6), 605–625.
Eisenhardt, K. M. (1989). Building theories from case study research. Academy of
Management Review, 14(4), 532–550.
Feitzinger, E., & Lee, H. L. (1997). Mass customization at Hewlett-Packard:
The power of postponement. Harvard Business Review, 75(1), 116–
121.
Fisher, M. L., & Ittner, C. D. (1999). The impact of product variety on auto-
mobile assembly operations: Empirical evidence and simulation analysis.
Management Science, 45(6), 771–786.
Flynn, B. B., & Flynn, E. J. (1999). Information-processing alternatives for cop-
ing with manufacturing environment complexity. Decision Sciences, 30(4),
1021–1052.
Fornell, C., & Larcker, D. F. (1981). Structural equation models with unobservable
variables and measurement error: Algebra and statistics. Journal of Market-
ing Research, 18(3), 382–388.
Forza, C., & Salvador, F. (2002a). Product configuration and inter-firm co-
ordination: An innovative solution from a small manufacturing enterprise.
Computers in Industry, 49(1), 37–46.
Forza, C., & Salvador, F. (2002b). Managing for variety in the order acquisition
and fulfilment process: The contribution of product configuration systems.
International Journal of Production Economics, 76(1), 87–98.
Forza, C., & Salvador, F. (2006). Product information management for mass cus-
tomization: Connecting customer, front-office and back-office for fast and
efficient customization. New York, NY: Palgrave Macmillan.
Framinan, J. M., & Leisten, R. (2010). Available-to-promise (ATP) systems: A
classification and framework for analysis. International Journal of Produc-
tion Research, 48(11), 3079–3103.
Galbraith, J. R. (1973). Designing complex organizations. Reading, MA: Addison-
Wesley.
Galbraith, J. R. (2005). Designing the customer-centric organization: A guide to
strategy, structure, and process. San Francisco, CA: Jossey-Bass.
202 Customization-Responsiveness Squeeze

Gattiker, T. F., & Goodhue, D. L. (2005). What happens after ERP implementation:
Understanding the impact of interdependence and differentiation on plant-
level outcomes. MIS Quarterly, 29(3), 559–585.
Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies
for qualitative research. London, UK: Wiedenfeld and Nicolson.
Guess, V. C. (2002). CMII for business process infrastructure. Scottsdale, AZ:
Holly Publishing.
Haas, M. R. (2010). The double-edged swords of autonomy and external knowl-
edge: Analyzing team effectiveness in a multinational organization. Academy
of Management Journal, 53(5), 989–1008.
Hanna, A. S., Camlic, R., Peterson, P. A., & Lee, M.-J. (2004). Cumulative effect
of project changes for electrical and mechanical construction. Journal of
Construction Engineering and Management, 130(6), 762–771.
Harter, D. E., Krishnan, M. S., & Slaughter, S. A. (2000). Effects of process
maturity on quality, cycle time, and effort in software product development.
Management Science, 46(4), 451–466.
Harville, D. A. (1977). Maximum likelihood approaches to variance component
estimation and to related problems. Journal of the American Statistical As-
sociation, 72(358), 320–338.
Hayes, R. H., & Wheelwright, S. C. (1979). The dynamics of process-product life
cycles. Harvard Business Review, 57(2), 127–136.
Hegde, V. G., Kekre, S., Rajiv, S., & Tadikamalla, P. R. (2005). Customization:
Impact on product and process performance. Production and Operations
Management, 14(4), 388–399.
Hickson, D. J., Pugh, D. S., & Pheysey, D. C. (1969). Operations technology
and organization structure: An empirical reappraisal. Administrative Science
Quarterly, 14(3), 378–397.
Huffman, C., & Kahn, B. E. (1998). Variety for sale: Mass customization or mass
confusion? Journal of Retailing, 74(4), 491–513.
James, L. R., Demaree, R. G., & Wolf, G. (1984). Estimating within-group inter-
rater reliability with and without response bias. Journal of Applied Psychol-
ogy, 69(1), 85–98.
Kilger, C., & Schneeweiss, L. (2005). Demand fulfilment and ATP. In H. Stadtler
& C. Kilger (Eds.), Supply chain management and advanced planning: con-
cepts, models, software and case studies (3rd ed.). Berlin, Germany: Springer,
179–195.
Klein, R. (2007). Customization and real time information access in integrated
eBusiness supply chain relationships. Journal of Operations Management,
25(6), 1366–1381.
Koh, S. C. L., Gunasekaran, A., & Saad, S. M. (2005). A business model for
uncertainty management. Benchmarking: An International Journal, 12(4),
383–400.
Tenhiälä and Ketokivi 203

Koh, S. C. L., & Simpson, M. (2005). Change and uncertainty in SME manu-
facturing environments using ERP. Journal of Manufacturing Technology
Management, 16(6), 629–653.
Krajewski, L., Wei, J. C., & Tang, L. L. (2005). Responding schedule changes
in build-to-order supply chains. Journal of Operations Management, 23(5),
452–469.
Krishnan, V., & Ulrich, K. T. (2001). Product development decisions: A review of
the literature. Management Science, 47(1), 1–21.
Lawrence, P. R., & Lorsch, J. W. (1967). Differentiation and integration in complex
organizations. Administrative Science Quarterly, 12(1), 1–47.
Lyon, D. D. (2004). Practical CM III: Best configuration management practices
for the 21st century. Pittsfield, MA: Raven Publishing Company.
March, J. G., & Sutton, R. I. (1997). Organizational performance as a dependent
variable. Organization Science, 8(6), 698–706.
McCutcheon, D. M., Raturi, A. S., & Meredith, J. R. (1994). The customization-
responsiveness squeeze. Sloan Management Review, 35(2), 89–99.
Miller, D. (1981). Toward a new contingency approach: The search for organiza-
tional gestalts. Journal of Management Studies, 18(1), 1–26.
Mohrman, S. A., Gibson, C. B., & Mohrman, A. M., Jr. (2001). Doing research
that is useful to practice: A model and empirical exploration. Academy of
Management Journal, 44(2), 357–375.
Paternoster, R., Brame, R., Mazerolle, P., & Piquero, A. (1998). Using the correct
statistical test for the equality of regression coefficients. Criminology, 36(4),
859–866.
Patterson, H. D., & Thompson, R. (1971). Recovery of inter-block information
when block sizes are unequal. Biometrika, 58(3), 545–554.
Pibernik, R. (2005). Advanced available-to-promise: Classification, selected meth-
ods and requirements for operations and inventory management. Interna-
tional Journal of Production Economics, 93, 239–252.
Pine, B. J., II. (1993). Mass customization: The new frontier in business competi-
tion. Boston, MA: Harvard Business School Press.
PMI. (2006). Practice standard on project configuration management (final expo-
sure draft v.1.0). Newtown Square, PA: Project Management Institute.
Podsakoff, P. M., & Organ, D. W. (1986). Self-reports in organizational research:
Problems and prospects. Journal of Management, 12(4), 531–544.
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common
method biases in behavioral research: A critical review of the literature and
recommended remedies. Journal of Applied Psychology, 88(5), 879–903.
Proud, J. F. (2007). Master scheduling: A practical guide to competitive manufac-
turing (3rd ed.). Hoboken, NJ: John Wiley & Sons.
Ramdas, K. (2003). Managing product variety: An integrative review and research
directions. Production and Operations Management, 12(1), 79–101.
204 Customization-Responsiveness Squeeze

Randall, T., & Ulrich, K. (2001). Product variety, supply chain structure, and firm
performance: Analysis of the U.S. bicycle industry. Management Science,
47(12), 1588–1604.
Riley, D. R., Diller, B. E., & Kerr, D. (2005). Effects of delivery systems on change
order size and frequency in mechanical construction. Journal of Construction
Engineering and Management, 131(9), 953–962.
Robertson, D., & Ulrich, K. (1998). Planning for product platforms. Sloan Man-
agement Review, 39(4), 19–31.
Safizadeh, M. H., Ritzman, L. P., & Mallick, D. (2000). Revisiting alternative
theoretical paradigms in manufacturing strategy. Production and Operations
Management, 9(2), 111–127.
Salvador, F., Forza, C., & Rungtusanatham, M. (2002). Modularity, product variety,
production volume, and component sourcing: Theorizing beyond generic
prescriptions. Journal of Operations Management, 20(5), 549–575.
Salvador, F., & Forza, C. (2004). Configuring products to address the
customization-responsiveness squeeze: A survey of management issues and
opportunities. International Journal of Production Economics, 91(3), 273–
291.
SAP. (2011a). SAP Library: mySAP ERP 6.0: Variant Configuration. Re-
trieved April 13, 2011, from http://help.sap.com/saphelp_erp2004/helpdata/
en/92/58d455417011d189ec0000e81ddfac/frameset.htm.
SAP. (2011b). SAP Library: mySAP SCM 5.0: Global Available-to-Promise.
Retrieved April 13, 2011, from http://help.sap.com/saphelp_scm50/
helpdata/en/26/c2d63b18bc7e7fe10000000a114084/frameset.htm.
Shenhar, A. J. (2001). One size does not fit all projects: Exploring classical con-
tingency domains. Management Science, 47(3), 394–414.
Sheu, C., & Wacker, J. G. (1997). The effects of purchased parts commonality on
manufacturing lead time. International Journal of Operations & Production
Management, 17(8), 725–745.
Siemsen, E., Roth, A., & Oliveira, P. (2010). Common method bias in regres-
sion models with linear, quadratic, and interaction effects. Organizational
Research Methods, 13(3), 456–476.
Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to
basic and advanced multilevel modeling. London, UK: Sage Publications.
Sousa, R., & Voss, C. A. (2001). Quality management: Universal or context de-
pendent? Production and Operations Management, 10(4), 383–404.
Sousa, R. (2003). Linking quality management to manufacturing strategy: An
empirical investigation of customer focus practices. Journal of Operations
Management, 21(1), 1–18.
Sousa, R., & Voss, C. A. (2008). Contingency research in operations management
practices. Journal of Operations Management, 26(6), 697–713.
Stadtler, H. (2005). Supply chain management and advanced planning – basics,
overview and challenges. European Journal of Operational Research, 163(3),
575–588.
Tenhiälä and Ketokivi 205

Starr, M. K. (1965). Modular production – a new concept. Harvard Business


Review, 43(6), 131–142.
Su, J. C. P., Chang, Y.-L., & Ferguson, M. (2005). Evaluation of postponement
structures to accommodate mass customization. Journal of Operations Man-
agement, 23(3–4), 305–318.
Swafford, P. M., Ghosh, S., & Murthy, N. (2006). The antecedents of supply chain
agility of a firm: Scale development and model testing. Journal of Operations
Management, 24(2), 170–188.
Szwejczewski, M., Mapes, J., & New, C. (1997). Delivery and trade-offs. Interna-
tional Journal of Production Economics, 53(3), 323–330.
Thirumalai, S., & Sinha, K. K. (2009). Customization strategies in electronic
retailing: Implications of customer purchase behavior. Decision Sciences,
40(1), 5–36.
Tu, Q., Vonderembse, M. A., Ragu-Nathan, T. S., & Ragu-Nathan, B. (2004). Mea-
suring modularity-based manufacturing practices and their impact on mass
customization capability: A customer-driven perspective. Decision Sciences,
35(2), 147–168.
Tushman, M. L., & Nadler, D. A. (1978). Information processing as an integrating
concept in organizational design. Academy of Management Review, 3(3),
613–624.
Van Wezel, W., Van Donk, D. P., & Gaalman, G. (2006). The planning flexibility
bottleneck in food processing industries. Journal of Operations Management,
24(3), 287–300.
Venkatadri, U., Srinivasan, A., Montreuil, B., & Saraswat, A. (2006). Optimization-
based decision support for order promising in supply chain networks. Inter-
national Journal of Production Economics, 103(1), 117–130.
Venkatraman, N. (1989). The concept of fit in strategy research: Toward verbal and
statistical correspondence. Academy of Management Review, 14(3), 423–444.
Vollmann, T. E., Berry, W. L., Whybark, D. C., & Jacobs, F. R. (2005). Manu-
facturing planning and control for supply chain management (5th ed.). New
York, NY: McGraw-Hill Irwin.
Voss, C. A., & Hsuan, J. (2009). Service architecture and modularity. Decision
Sciences, 40(3), 541–569.
Weng, Z. K., & Parlar, M. (2005). Managing build-to-order short life-cycle prod-
ucts: Benefits of pre-season price incentives with standardization. Journal of
Operations Management, 23(5), 482–495.
Woodward, J. (1965). Industrial organization: Theory and practice. London, UK:
Oxford University Press.
Yeo, K. T., & Ning, J. H. (2002). Integrating supply chain and critical chain
concepts in engineer-procure-construct (EPC) projects. International Journal
of Project Management, 20(4), 253–262.
Yuan, K.-H., & Bentler, P. M. (2000). Three likelihood-based methods for mean
and covariance structure analysis with nonnormal missing data. Sociological
Methodology, 30, 165–200.
206 Customization-Responsiveness Squeeze

Zorzini, M., Corti, D., & Pozzetti, A. (2008). Due date (DD) quotation and capacity
planning in make-to-order companies: Results from an empirical analysis.
International Journal of Production Economics, 112(2), 919–933.

Antti Tenhiälä is an assistant professor of operations management at IE Business


School in Madrid, Spain. He holds MSc, LSc, and DSc degrees in industrial en-
gineering and management from Helsinki University of Technology. His current
research is focused on day-to-day operations management in complex manufac-
turing industries and it explores the contingency factors that explain the variance
in the effectiveness of different planning methods, decision support systems, and
communication practices.

Mikko Ketokivi is a professor of operations management at IE Business School


in Madrid, Spain. He holds BS and MSc degrees in business administration from
Iowa State University and a PhD in operations and management science from
University of Minnesota. He teaches organization theory, organization design,
research design, and statistical research methods. In his research, he focuses on
organization design, methodology, scientific reasoning, and rhetoric. His research
has been published in various operations management and general management
journals, including Administrative Science Quarterly, Academy of Management
Review, Strategic Management Journal, Journal of Operations Management, and
Production and Operations Management.

You might also like