Mark Damian Morelli.

Evaluating communications in product development organizations online

. (page 1 of 3)
Online LibraryMark Damian MorelliEvaluating communications in product development organizations → online text (page 1 of 3)
Font size
QR-code for this ebook





Evaluating Communications in
Product Development Organizations

Mark D. Morelli
Steven D. Eppinger

WP #3602-93

August 1993





Evaluating Communications in
Product Development Organizations

Mark D. Morelli
Steven D. Eppinger

WP #3602-93 August 1993


TTie authors wish to thank the product development team members at the research
site, especially the project manager and manager of the benchmark facility organization
for their insights into this project. We also thank Tom Allen and Varghese George for
their helpful comments. Some additional details about this research can be found in Mark
Morelli's master's thesis [Morelli 1993]. Financial support for this project was provided
by the MIT Leaders for Manufacturing Program.

Send correspondence to:

Professor Steven D. Eppinger

M.I.T. Sloan School of Management

30 Wadsworth Street, E53-347
Cambridge, Massachusetts 02139


DEC -7 1993


Evaluating Communications in
Product Development Organizations

Mark D. More Hi
Steven D. Eppinger

M.I.T. Sloan School of Management


This paper develops a novel technique for studying the effectiveness of product
development organizations that conduct concurrent engineering. It is commonly argued
that to create better products more quickly, the degree of coordination among team
members must be enhanced. However, barriers to communication exist across
organizational, cultural, and geographical boundaries in dispersed product development
teams. In contrast to previous paradigms that encourage co-location and increased
communication everywhere, this paper develops a methodology which predicts the
important communication linkages and subsequently measures communication flows to
determine where and to what extent the necessary transfer of information takes place.
We have applied this method to the study of a product development project in industry,
enabling us to learn generally about communication patterns in product development and
specifically to characterize communication for the product development team smdied.

This study documents three major findings: 1 ) 8 1 % of all technical
communication linkages were predicted in advance; 2) communication within co-located
teams and across geographic distances was higher than anticipated; 3) two-way
communication exchange is typical, even if one-way information transfer is predicted.
These results have important implications for the management of large-scale product
development projects; namely, that organizational design can be prescribed ahead of time
by anticipating necessary communication linkages. We are able to conclude: 1) that
essential technical interactions can be planned and do not necessarily require co-location;
and 2) that unplanned, informal interactions should be facilitated by co-location of core
team members. This pajjer concludes with a critical analysis of the research method and
suggested improvements. While this work is aimed at establishing a methodology,
implications are imponant for the management of new product development.


The goal of this work is to define and explore a methodology for examining
communication patterns within product development organizations. The methodology is
based on first predicting important patterns of communication and then measuring
communication to see if the anticipated linkages exist. The following study applies this
methodology to a product development project in industry. The results offer funher
insight into communication patterns in product development and provide a basis for
future research.

This work is motivated by the critical imponance of product development in
today's businesses and the general lack of specific understanding of communication
issues in product development organizations. Successful companies as observed by
Wheelwright and Clark [1992], must anticipate and fulfill customer needs and deliver
products to market quicker than their competition; in a fast paced global economy, only
the companies that do this effectively will survive. Our approach to facilitating the
critical product development function is based on studying the information transfer
between members in the product development organization. Communication studies
have been applied to many organizations but little work has been tailored specifically to
the product development process. Work on product development organizations done by
Baiczak and Wilemon [1991], Griffin and Hauser [1992] analyze patterns of
communication and relate them to the success of projects. Our focus is to prescribe
important communication linkages in a product development organization and then to
measure these linkages to see if they exist.


Current trends in business have made product development more challenging by
intensifying geographical barriers to communication within the organization. Pine [1993]
shows that over the past decade, markets have become more fragmented, reacting to
sophisticated and demanding consumers who expect easy access to low cost products that
provide solutions to specific requirements. Companies have responded with
multinational product development which often means designing products in one locality,
manufacturing in another and selling in yet another. As the case of Motorola [ 199 1 ]
exemplifies, it is often no longer possible or desirable to design and manufacture products
in the same building, or on the same continent, since leveraging company assets across
the globe is imponant for companies serving diverse markets. The requirement to act
globally with new and large product development projects often requires cross-functional

teams divided into several subgroups and distributed over a geographical region [Dean
and Susman 1989].

Even though successful product development projects face greater challenges in
coordinating large, dispersed cross-functional teams, historical barriers to communication
continue to persist. Barriers can arise from organizational structures, incentive systems,
geographical location, cultural differences, leadership styles and project management
practices. Operating within these barriers is difficult for modem product development.
Especially since the practice of concurrent or simultaneous engineering, employed to
speed up the product development process, requires increased coordination [Clark and
Fujimoto 1991; Clausing 1989; Krishnan, Eppinger and Whitney 1993]. This is because
as firms continue to shrink development time through concurrent engineering, tasks must
be overlapped more aggressively, requiring much coordination between tasks.

Communication Patterns in Product Development

Successful product development requires smoothing the barriers to
communication and enhancing communication when and where it is required. It is not
that communication must be enhanced everywhere, rather that specific patterns of
communication have been shown to be related to successful organizations. Previous
studies of large-scale product development such as the work conducted by Clark and
Fujimoto [1991], reveal that successful development relies uf)on intensive
communication between upstream and downstream team members. While Dougherty
[1987] concludes that higher levels of inter-functional communication occur more often
in successful product development projects and diminished communications exist in
failed projects. Allen, Lee and Tushman [1980] have shown that for R&D organizations,
increased internal communication within groups does not correlate to increased project
performance, while organizations strongly benefit from coinmunication with other pans
of the firm. This may be due to diminishing returns of internal communication where the
benefits of communicating more to internal group members does not provide much
additional benefit, while even small amounts of additional communication to others
outside the group provides greater benefit.

Though research has verified the importance of communication, we believe that it
is neither practical nor beneficial to increase communication everywhere. Enhanced
communication when and where it is supposed to take place affects the success of product
development projects [Wheelwright and Clark 1992]. Certain patterns of communication
such as that between project manager and specific subgroups are dependent on the type of
work conducted by the team. Barczak and Wiiemon [1991] support this by showing that

patterns of communication affect the success of product introductions and are dependent
on whether the process focused on developing an entirely new product or improving an
existing one

In order to study patterns of communication, Allen [1986] has identified three
basic typ)es of communication: coordination, information and inspiration. This
classification helps to analyze the work content of communication linkages in product
development organizations. Coordination-type communication entails information
transfer for the purpose of executing tasks and conducting work. Information-type
communication includes the exchange of information where learning takes place or where
new knowledge is gained, such as receiving instruction on new computer software.
Inspiration-type is where new ideas are created but with the same previous knowledge, as
in a brainstorming session to solve a problem. While all three types of communication
can exist, Allen [1986] shows that some organizational linkages within the project may
rely strongly on one or more of these types.

Smoothing Communication Barriers Through Co-location

One practice highly recommended for smoothing communication barriers and
allowing team members to exchange information more easily is the co-location of team
members. Co-location is the placement of (cross-functional) team members at the same
facility and in close proximity to one another in order to increase the performance of the
team by breaking down geographic barriers to communication. Co-location of product
development teams is widely recommended by authors such as Smith and Reinertsen
[1991], Dean and Susman [1989], largely based on Allen's [1977, p.239] research of
R&D organizations. Allen's work shows a higher probability of communication between
workers in close proximity, while those farther away have a markedly decreased
probability of communicating [Allen 1977]. Figure 1 illustrates Allen's curve relating
probability of communication as a function of distance between communicating pairs.
This "proximity barrier" curve shows that for separation distances of more than 30
meters, the curve approaches its asymptote, so that dyadic relationships 35 meters away
have nearly the same probability of communication as individuals separated by 255

10 20 30 40 50 60 70 80
Distance Between Communicators (meters)


Figure 1. The probability that two people will communicate as a function of the distance

that separates them. Taken from Allen [1977, p. 239]. The line represents a

regression from data of seven R&D organizations 0-255 kilometers.

Recommending co-location of product development team members based on this
information raises several concerns. First, Allen's data are compiled from R&D
organizations in which respondents were asked with whom they communicate on a
regular basis to accomplish work [Allen 1977; Allen, Lee and Tushman 1980]. The
answers are highly dependent on the nature of the work within the organization, and there
are distinct differences between basic or applied research and product development
project work. \pVhile Allen's study is for R&D organizations, the interpretation of R&D
included both research and product development, so there is no distinction between the
two in data collection. However, R&D organizations are engaged in developing or
enhancing technology for future products, while new product development is focused on
creating or revising products to be manufactured immediately. We believe these
differences in the nature of the work are significant enough to bring into question the
potential differences in communication patterns that exist. Recommendations based upon
studies of R&D organizational networks involving thousands of respondents across a
multitude of projects may not be entirely relevant to a product development team in
which 50 to 100 individuals of cross-functional backgrounds are required to work
together on a single project.

Fiuthermore, the proximity barrier has confounding effects which may
legitimately explain the reason probability of communication decreases with distance.
For example, people in the same geographic location may be organized by work
functions so that respondents are surrounded by people they are required to communicate
with to accomplish work. It is also possible that people in close proximity communicate

frequently with close neighbors, changing the nature of work to include nearby
communicators. The proximity barrier fails to differentiate whether nearby
communications are required to accomplish work or whether communications across
distances are established based on the needs of the project.

Even if co-location proves to be the organizational design of choice, there are
several costs associated with co-location that many companies find reduce its
attractiveness: 1 ) The expense of moving a large organization under a single roof may be
prohibitive since it can be difficult to obtain a suitable facility and relocate the necessary
employees. 2) Increased alliances and outsourcing limit the range of viable co-location
options. 3) Some team members may be required to work on multiple projects.

Research Focus

This communication study differs from past research in that we compare the
predicted and actual communication linkages in a product development organization.
Past approaches measured communication patterns and correlate observed patterns with
measures of success. In this study, we first predict what patterns of communication are
essential for executing a product development project. We then measure what linkages
are actually established. Finally, by comparing the predicted and actual communications,
we are able to comment uf)on the origin and utility of each pattern. Only in this way can
we study existing communication and determine if anticipated important linkages take

In taking this approach, we are able to explore the following questions:

• Can strong organizational links be predicted accurately in advance of the
development project and do they take place as prescribed?

• Do certain barriers to communication effectively restrict imponant communication
that is supposed to take place, and if not, what are the mechanisms that facilitate
cross-barrier communication?

• Does co-location facilitate necessary communication, and is this practice required for
necessary communication to take place?

• What organizational forms are required to facilitate communication in product
development, and can these solutions be successfully prescribed for a project?


This section develops a novel research methodology to analyze the predicted and
actual information transfers in product development. Our approach involves five steps.

1) We define the critical technical information flows which are of the coordination-type
communication through interviews with project team members.

2) We use the interview data to plan for future coordinations and represent these
interactions in the DSM.

3) We measure what communication takes place within the organization using weekly

4) We use the communication netgraph to represent measured communication linkages
indicated from weekly data.

5) Finally, we develop frameworks to compare predicted and actual communication
linkages to explore the questions posed above.

Predicting Communication Linkages with the Design Structure Matrix

We represent predicted linkages using the design structure matrix (DSM). The
DSM is an analytical tool developed by Eppinger [1991] and Steward [1980], that has
been used to organize and sequence tasks in complex product development projects
[Eppinger, Whitney, Smith and Gebala 1990; Steward 1981]. The philosophy behind the
DSM method described by Eppinger [1991], is that the project can be represented as a set
of individual tasks, and the relationships among these tasks can be analyzed to determine
the underlying structure of the project. DSM research has been driven by the increased
complexity and importance of product development due to the adoption of simultaneous
engineering and design for manufacturing. In an attempt to speed time to market,
managers aggressively overlap design tasks, increasing the coordination requirements
between overlapping functions. Figure 2 shows the increasing overlapping of tasks in
concurrent product development and the need for enhanced information flows. The DSM

Series Development

Concurrent Development








Figure 2. Overlapping of design tasks in concurrent or simultaneous engineering requires
increased levels of information exchange as shown by Krishnan et al. [1993].


has been used by Krishnan et al. [1991], Sequeira [1991], Smith [1992] to study complex
concurrent engineering projects, by McCord [1993] to facilitate effective linkages
between project tasks, and by Osborne [1993] to model development iterations.

In Figure 3, the design tasks for a hypothetical project are shown in the DSM
format. Each task in the project is described by a row and a column of the matrix. For
each task's row, we place a mark in every column from which the task needs input. Then
scanning across each row indicates the informational input required for task completion.
Reading down any column then indicates where the task's informational output must flow
to downstream processes. For example, row D shows that completion of task D requires
information from tasks B, C and E, and reading down column B indicates that
information from this task is required by tasks C, D, and G. The objective of the DSM is
to plan for the most efficient structure for the development project and to map the
information interdependencies between tasks which can facilitate the passing of
information between tasks. We depict a set of tasks to be executed concurrently by
drawing a block to group these tasks. For more information on the DSM see [Eppinger
1991; Steward 1981; Morelli 1993].


A Determine specifications

B Design concept

C Design component 1

D Design component 2

E Design tooling

F Integrate design & tooling

G Prototype manufacture

H Product test




• X X
X • X X
X X • X
X X •


X •
X •

Figure 3. Example of a DSM for a hypothetical project. Each mark represents a need for

information transfer. The task sequence is depicted by the ordered listing, and

concurrent activities are identified by the block grouping tasks in the matrix.

Representing Measured Relationships with the Communication Netgraph

Once the coordinating linkages are predicted, the communicadons are measured
to determine to what extent the linkages actually took place. Relationships between
individuals within organizations have been studied for some time in the field of network
analysis [Moreno 1978; George and Allen 1989]. The resulting network structure of
dyadic communication relationships is called a communication network. A typical

communication network is shown in Figure 4. Each numbered node represents a person
and the lines connecting the nodes represent significant relationships.

Figure 4. Typical communication network for a small organization.

The resulting complexity of generating communication networks in this manner
prompted Allen and George [1989, 1993] to develop another framework in which to
conduct network analysis. One result of their work is a computer tool known as A
Graphic Network Interpreter (AGNI) which was used in this study to represent and
analyze networks. The AGNI format enables researchers to conduct network analysis
using a graphical matrix representation of networks called a netgraph. The netgraph is a
pictorial representation of networks in which the nodes can be used to represent
individuals and the relationships between individuals can be depicted by the presence of a
symbol in the matrix linking two individuals. The relational data between individuals can
be used to create an array of elements that represent contacts between individuals within
the organization. While positions along the columns, rows and the matrix diagonal
represent individuals in the organization. The netgraph would be completely symmetrical
if for each communicating pair, both members indicated that a significant communication
took place.

AGNI is used to analyze the data, represent the networks, and to rearrange the
rows and columns so that certain relationships can be highlighted such as the frequency
of communication or organizational boundaries within the organization. The use of color
or shaded symbols can be used to delineate organizational subsets such as individuals
with the same job function or an attribute shared between individuals. Figure 5 shows a
netgraph where the rows and columns have been arranged by co-located groups.
Individuals in the same immediate workplace are grouped by solid lines. Symbols
outside the blocks represent communication linkages bridging individuals in separate
groups. The two different tile shades in this case represent either high or low frequency


I ■ ■ H


■ ■

1 ■

Figure 5. Example of a netgraph grouped by geographical boundaries,
with shades denoting frequency of information exchange.

Comparing Predicted and Actual Communication

We use the DSM to prescribe the important flows of information that should take
place in product development. The communication netgraph shows the existing patterns
of communication among product development team members. Applying these tools to
the same product development process allows us to explore the effectiveness of
information transfer within the organization.

Although the DSM and communication netgraph each represent information
transfer, they are inherently different firameworks, and a transformation mapping between
the two is required to conduct comparisons. Figure 6 shows the steps taken in order to
compare predicted and actual communication information. The resulting "comparison
netgraph" is a summary of the predicted communication from the DSM, and the actual
communications represented by the actual-communication netgraph.

The DSM is first translated into the predicted/potential-communication netgraph.
Conversion of the DSM into such a netgraph requires each task on the DSM to be
assigned a corresponding person or group of people. In order for a communication
linkage to be established in the netgraph, the two individuals or groups of individuals
must be required to work on a given task together indicated by a mark on the DSM (in
this case an 'X'). Confounding effects occur when more than one person is assigned to a
single task on the DSM, because it is not known if all or a fraction of the members will
actually communicate to accomplish the task. Therefore, the predicted/potential


communication netgraph indicated in Figure 6, contains organizational linkages as
predicted by the DSM, but not all links need to be fulfilled to accomplish required tasks.

Design Predicted/Potential

Structure Matrix Communication





tf • 1


. .••,1 •

• • —9 •••■

ft •

1— •»!




U 1



Comparison Netgraph

Figure 6. Creating a comparison between predicted and actual communications.

Next, survey data are used to create the actual-communication netgraph. The
actual-coinmunication netgraph is obtained by aggregating weekly survey data. No lower
threshold limit is set to establish a communication linkage and only one panner was
required to indicate that communication took place. Since questionnaires are issued
randomly once a week, only a fraction of the total communication is represented in the

1 3

Online LibraryMark Damian MorelliEvaluating communications in product development organizations → online text (page 1 of 3)