Intepretive Barriers to Successful Product Innovation in Large Firms

Abstract

The development of commercially viable new products requires that technological and market possibilities are linked effectively in the product's design. Innovators in large firms have persistent problems with such linking, however. This research examines these problems by focusing on the shared interpretive schemes people use to make sense of product innovation. Two interpretive schemes are found to inhibit development of technology-market knowledge: departmental thought worlds and organizational product routines. The paper describes in some depth differences among the thought worlds which keep innovators from synthesizing their expertise. The paper also details how organizational routines exacerbate problems with learning, and how successful innovators overcome both interpretive barriers. The main implication of the study is that to improve innovation in large firms it is necessary to deal explicitly with the interpretive barriers described here. Suggestions for practice and research are offered.

Full text

Interpretive Barriers to Successful Product Innovation in Large Firms
Author(s): Deborah Dougherty
Source:
Organization Science,
Vol. 3, No. 2 (May, 1992), pp. 179-202
Published by: INFORMS
Stable URL: http://www.jstor.org/stable/2635258
Accessed: 30/06/2010 11:10
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=informs.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Organization Science.
http://www.jstor.org

ORGANIZATION SCIENCE
Vol. 3, No. 2, May 1992
Printed in U.S.A.
INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT
INNOVATION IN LARGE FIRMS*
DEBORAH DOUGHERTY
Management Department, The Wharton
School, University of Pennsylvania,
2000 Steinberg-Dietrich, Philadelphia, Pennsylvania 19104-6370
The development of commercially viable new products requires that technological and
market possibilities are linked effectively in the product's design. Innovators in large firms
have persistent problems with such linking, however. This research examines these problems
by focusing on the shared interpretive schemes people use to make sense of product
innovation. Two interpretive schemes are found to inhibit development of technology-market
knowledge: departmental thought worlds and organizational product routines. The paper
describes in some depth differences among the thought worlds which keep innovators from
synthesizing their expertise. The paper also details how organizational routines exacerbate
problems with learning, and how successful innovators overcome both interpretive barriers.
The main implication of the study is that to improve innovation in large firms it is necessary
to deal explicitly with the interpretive barriers described here. Suggestions for practice and
research are offered.
(INNOVATION; NEW PRODUCTS; INTERDEPARTMENTAL COLLABORATION;
INTERPRETATION)
Successful product innovation is vital to many firms. This paper builds on three
findings in the literature to explore the product innovation process in large firms. The
first finding is that the commercial success of a new product depends on how well the
product's design meets customers' needs (Rothwell et al. 1974; Lilien and Yoon
1988). An effective design requires that technological possibilities for a product are
linked with market possibilities, e.g., who are the users? what will they use the
product for? The second finding is that collaboration among the technical, marketing,
manufacturing, and sales departments contributes to a new product's success (Bonnet
1986; Dean and Susman 1989). The third finding is that product innovators often do
not link technological and market issues, and often do not collaborate across
departments (Cooper and Kleinschmidt 1986; Souder 1987).
This research seeks to explain why innovators fail to develop a comprehensive
appreciation of their product in its market. Three implications are developed from
the findings. First, collaboration is necessary to technology-market linking, which
means that collaboration enhances the product's design along with improving the
execution of the development process. Second, the styles in which people organize
their thinking and action about innovation-their "interpretive schemes"-are major
barriers to linking and collaboration. Like "culture," such schemes provide shared
assumptions about reality, identify relevant issues, and help people make sense of
those issues (Daft and Weick 1984; Bartunek 1984). In the case of product innova-
tion, two interpretive schemes become interpretive barriers: (1) departments are like
different "thought worlds," each focusing on different aspects of technology-market
knowledge, and making different sense of the total; and (2) organizational routines
separate rather than coordinate the thought worlds, further constraining joint learn-
ing. The third implication is that correcting the innovation problems caused by these
interpretive barriers requires cultural solutions, not only structural ones.
*Accepted by Richard L. Daft; received August 1989. This paper has been with the author for two
revisions.
179
1047-7039/92/0302/0179/$01.25
Copyright ?3 1992, The Institute of Management Sciences

180 DEBORAH DOUGHERTY
OPCO-An Example
Before defining terms and then explaining how these implications arise, excerpts
from a case in the research are presented to illustrate thought worlds and routines,
and to introduce the research. The data management division of OPCO, a communi-
cations company, decided to develop and sell a software accounting system that
would process credit card transactions over their data network. They purchased a
start-up firm with a system under development to enter the business quickly, and,
after several months additional work at OPCO, the service was introduced to the
market. The system did not operate properly, however, and was cancelled. As
suggested by the quotes below, the innovation encountered many problems. The
quotes also suggest that the participants interpret issues with unique departmental
perspectives. The business manager sees no need to talk with customers since he
considers the market opportunity to be obvious. He feels, however, that they failed to
position the product properly against competitors:
[Interviewer: Did you talk to customers?] No, because we knew that this was needed. We could
see some competitors getting into the business. ... We could see that this was logical for us to do,
so we decided to go.... Had we done some market research and defined needs more carefully, and
figured out the dozens of pieces we would need for a full system, we'd be positioned with a much
better strategy.
The technical director sees no "market" problems at all with the effort. He
describes design problems, however, that perhaps could have been resolved had a
thorough analysis of users and how they operate been carried out:
There were no market problems with this product. ... Our mistake was we didn't understand the
application in total. ... We had a difficult time trying to figure out the relationships between us
here at the operating level and the retail establishments, and the relationships between them and
their banks and credit card clearing houses. ... There were a lot of players involved, which is
different from (our regular product), where we interface with one customer at a time. It looks very
nice theoretically, but the more relationships there are, the more complex the recovery.
The sales support person downplays general positioning and technological design
to blame instead their failure to specify which users in the market could best use the
product:
I have never seen a definition for this service. There are no criteria on what makes a good or bad
customer for this product. ... One person here had a pretty good understanding of what kind of
customer would benefit from a system like this. More people here should have known. We needed a
brain transplant.
The innovators also followed established routines at OPCO Data to develop and
launch this new service. These routines included project teams and matrices, struc-
tures that are recommended for innovation (e.g., Tushman and Nadler 1986), but all
participants note how they did not work. The routines did not synthesize components
of the innovation itself nor relevant knowledge, and they squashed interaction:
We were not successful in fully integrating the new business within the organization. A new
product is unique-it has different distribution, different billing, a myriad of things have to work out
well. It is difficult for a small organization [referring to the business unit] to handle all these issues,
so things fall apart. We didn't see the pitfalls (business manager).
At OPCO we tend to categorize people into roles, and give people only what they need to
know. ... There are little shadings of meaning that get lost in the requirements statement from
marketing (technical director).

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 181
They [from the small company] were a very tight group, and they all talked to one another all the
time. But when we brought them here they were dispersed into our matrix... (sales support).
In hindsight at least, these innovators knew that they should link technological and
market issues, but they did not. This case along with 17 others is analyzed to explain
why.
Conceptual Background
Components of Technology-Market Linking
To understand problems with technology-market linking, it is first necessary to
understand what that linking comprises. Research suggests that technology-market
linking has a process and a content component. On the process side, linking involves
the construction of new knowledge about the product and the market. Henderson
and Clark (1990) suggest that nonroutine innovations require new "architectures," in
which innovators break out of existing procedures and know-how and reconfigure
components of design and procedure into a new framework. Freeman (1982) de-
scribes product innovation as a "complex coupling" between market needs and
technologies over time. Linking technological and market possibilities is challenging,
because choices must be made among multiple design options, each with different
outcomes. At the same time the market may be new so it is difficult to determine who
the most likely customers are and what they actually need (Clark 1985). Developing
innovative products is thus a process of double loop learning (Argyris and Schon
1978), in which new insights are incorporated and the premises themselves are
reconsidered.
On the content side, linking means that a complex array of specific insights must be
gathered and brought together. Dougherty (1990) finds that successful new product
developers had more insight into users' applications, technological trends, distribu-
tion systems, and market segments. Urban and von Hippel (1988) suggest that
developers determine key trends in both the technology and market areas, and then
search for "lead users" who can identify viable design specifications. According to
Bonnet (1986), integration of R & D and marketing facilitates both the assessment of
commercial viability and the optimization of design characteristics. Requisite knowl-
edge for new products is thus multi-faceted, multi-leveled, and detailed.
Organizational
Barriers to Technology-Market Linking
To understand why people do not link technology and market issues effectively, it
helps to understand what prevents them from frame-breaking learning, and from
gathering and connecting diverse insights. Organization research contains numerous
references to interpretive schemes at the department and organization levels which
may intervene in these necessary innovation processes. Departments can develop
different perspectives through which they might separate rather than combine infor-
mation, including cognitive orientations such as goals, time frames, and formality
(Lawrence and Lorsch 1967), languages (Tushman 1978), perceptions (Dearborn and
Simon 1958), occupational cultures (Van Maanen and Barley 1984), or power (Riley
1983).
At the organizational level, firms create "programs" (March and Simon 1958) or
"routines" (Nelson and Winter 1982) which can inhibit new product development.
According to Nelson and Winter, routines are regular and predictable behavior
patterns, which, first, comprise the organizational memory. When the firm is in a state
of routine operation, each person knows his or her job, and there is no need to know
others' jobs. Second, routines represent a truce for intra-organizational conflict,

182 DEBORAH DOUGHERTY
which, as Perrow (1986) notes, binds the firm in a network of practices that are
difficult to alter. Third, routines are standards which managers try to keep from
changing.
Few organization studies have connected departmental differences and organiza-
tional routines directly to new product development and technology-market linking,
however. Lawrence and Lorsch (1967)'s often cited insights concern ongoing busi-
nesses, not discrete new products. Several studies suggest that their cognitive differ-
ences do not differentiate departments for new product development (Harrison 1980;
Gupta, Raj and Wilemon 1986). And, while organization theorists may refer to
knowledge development (Galbraith 1982; Kanter 1983), most concentrate on struc-
tures and cultures for innovativeness in general, not for product innovation in
particular.
Toward a More Complete Model
Fleck (1979) was the first to apply differences in interpretive schemes to innovation
(or at least to scientific discovery-Douglas 1987). His views suggest how to integrate
the insights outlined above into a more complete understanding of the organizational
and interpretive processes underlying product innovation and technology-market
linking. Fleck emphasizes the social basis of cognition, and adds that innovations
often are epistemologically unsolvable by any one person. They require insights from
a variety of specialties, called "thought collectives" or "thought worlds" as Douglas
(1987) proposes to retranslate the term. A thought world is a community of persons
engaged in a certain domain of activity who have a shared understanding about that
activity. Microbiologists, plumbers, opera buffs, and organizational departments all
can be viewed as thought worlds.
Departmental Thought Worlds. Two aspects of thought worlds are relevant to
product innovation: their "fund of knowledge"-what they know, and their "systems
of meaning"-or how they know. According to Fleck what is already known influ-
ences the method and content of cognition. Thought worlds with different funds of
knowledge cannot easily share ideas, and may view one another's central issues as
esoteric, if not meaningless. A thought world also evolves an internally shared system
of meaning which provides a "readiness for directed perception" based on common
procedures, judgements, and methods. These systems of meaning produce an "intrin-
sic harmony" for the thought world, so ideas that do not fit may be reconfigured or
rejected outright.
For new product development, one could infer that departmental thought worlds
would selectively filter information and insights. Because of specialization, a certain
thought world is likely to best understand certain issues, but also to ignore informa-
tion that may be equally essential to the total task. Their intrinsic harmony would also
reduce the possibility for creative joint learning, since members of a department may
think that they already know everything.
Organizational
Routines. Fleck does not deal with organizations per se, but argues
that the collective action necessary to innovation is motivated by pressure from the
social context. "Collective work proper" (1979, p. 99) is different from additive work,
as when people come together to lift a large rock. Rather, it refers to "the coming
into existence of a special form, like a soccer match, conversation, or orchestra." This
new social form alters the thought worlds' existing readiness for directed perception
to allow new possibilities for discovery and new facts. Fleck studied the development
of the Wassermann test for syphilis, and argues that the disease was undefined for
400 years in part because there were no means for collective action. Different groups
such as astrologers, priests, pharmacists, and physicians operated with their own

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 183
theories. If it were not for the "insistent clamor of public opinion for a blood test"
(1979, p. 77), Wassermann would never have gathered the collective experience
necessary to develop a test. Wassermann began with the incorrect immunology
perspective that syphilis is caused by a virus. But because of external pressure, his
persistence coupled with developments from chemistry, medicine, and laboratory
thought worlds to uncover the actual cause of the disease (a spirochete). A new social
form among these diverse groups emerged, and they collectively produced a test for
the disease.
Fleck's dynamics can be combined with Nelson and Winter's (1982) routines to
suggest that the organizational context affects the thought world's capacity to collabo-
rate, inhibits the development of new knowledge, and keeps innovators from creating
a new social form.
Research Questions. This study will explore these dynamics by addressing the
following questions: (1) What are the different funds of knowledge and systems of
meaning for new products in the departmental thought worlds, and how do they
affect product innovation? (2) What are the routines that inhibit product innovation,
and how do they affect collective action among the thought worlds?
Methods
Data regarding 18 new product efforts in five firms were collected by interviewing
80 people from different departments who worked on these products. Schall's (1983)
multi-method analysis suggests that such interviews are a valid means to assess
departmental interpretive differences. An embedded, multiple case design was used
(Yin 1989) in order to consider the effects of organization and product success and
failure on the findings (Bailyn 1977).
Organizations
and Products
Two of the firms are' in the computer/communications industries (OPCO and
SALECO) and three are in the chemical materials industries (TECHCO, COMPCO,
and PRODCO). All generate over $1 billion annually, employ over 20,000 people,
and are over 35 years old. The criteria used to select products for the study are:
(1) they incorporated new or unfamiliar technology for the firm and/or were
marketed to new or unfamiliar users; (2) the products were almost or already
introduced to the market, to eliminate variance due to development stage; and (3)
some products were commercially successful, and some were failures. "Success" was
defined as generating at least as much profit as planned and "failure" as cancellation
after introduction.
Data were collected in two stages. First, eleven cases were studied at TECHCO
and OPCO. At TECHCO, a manager in each of the six product groups in its
industrial division was asked to suggest a product from his group that met the criteria.
One said he was too busy, so five products were selected. At OPCO, managers
directed me to their voice venture (one product), and their data division, where five
more cases were studied. Only one of these eleven cases was successful as defined
above, however, and even though six were in the market their profitability was still
uncertain. A second stage of data collection was undertaken to add more successful
cases. SALECO, COMPCO, and PRODCO were asked to identify one success
already introduced, and one failure that had been introduced but cancelled. At
COMPCO, people also described an uncertain product, and this was included.
Seventeen products were in the market, so all but one were in the post-introduc-
tory stage when people were interviewed. The products are described in Appendix A.
In addition to success and failure, a new category labeled "uncertain" was made for

184 DEBORAH DOUGHERTY
TABLE 1
Comparing
Product Innovation Cases by Success Status
Successful Uncertain Failed**
(4 cases) (7 cases) (6 cases)
Comparisons
Average Time in
Development Before 31 26 23
Introduction in Months (1)
Average Time After
Introduction in Months 42 19 18
Unfamiliarity
to Firm (% of cases): (2)
% new technology 50% 63% 50%
% new manufacture 75 88 83
% new market segments 50 75 50
% new applications 100 88 67
% new distribution 100 100 67
**Successful = already introduced and meeting/exceeding expectations;
Uncertain = not meeting expectations but not cancelled (7 cases already introduced, 1 still in develop-
ment);
Failed = already introduced and subsequently cancelled.
(1) The starting time is when the particular product began to be developed, not when underlying
technologies were invented.
(2) If 2 of the participants said that the product was unfamiliar to the firm in this area, it was coded as
unfamiliar.
the seven cases that were in the market but not generating profits as expected. Table
1 compares the successful, uncertain, and failed cases. The successful cases were in
development longer on the average, but this difference is because of one case which
took over five years. The uncertain cases, however, have been out on the market for
less time than the successes. Post-introductory time may enhance success, since some
of the uncertain cases will eventually become successful, while some will be cancelled.
The success or failure might depend on the case's unfamiliarity to the firm. To
compare cases on unfamiliarity, if two people said that the product was unfamiliar in
any of the five areas noted, it was coded as such. As can be seen in Table 1, there is
no difference in unfamiliarity by success status.
The People and the Interviews
At least two people who worked on each product were interviewed, and most were
still working on the product. All were operational and middle level managers, and
only six had less than four years experience with the firm. All interviews followed the
same protocol: describe the product; outline your role; tell the story of the product;
and describe customers, technology, and working relationships with others. The
interviews were structured around these general questions, but unstructured regard-
ing what the person chose to emphasize. The interviews lasted from about one hour
to over two hours. Notes were taken and filled in as soon as possible afterwards.
Table 2 summarizes the people interviewed by product, department, and firm.
Retrospective interviews have two important sources of potential distortion: mem-
ory failure and attribution bias. To guard against memory failure, people were asked
regularly for dates and names to keep them grounded in particulars, and one person
was reinterviewed briefly to clarify any conflicts in reported events. In addition,
archival data were reviewed for three of the products, and no conflicts with people's
stories were found. To check for attribution bias, the uncertain cases were compared

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 185
TABLE 2
Number of People Interviewed
by Department, Company, and Product Case
Company and Case: Technical Field Manufacturing Planners Total
TECHCO (chemicals)
*F Battery 1 1 2
U CRTDevice 3 1 1 5
U Video Device 2 1 3
U Medical System 2 1 1 4
S Film Cover 1 2 3
Others 1 2 3
OPCO (communications)
F Accounting System 2 1 1 4
F Document System 2 2 4
U Software System- 1 2 3
U Text System 4 1 1 2 8
UVoice System 1 5 1 1 8
U Transmit System 1 2 3
Others 3 3
SALECO (computers)
FSystem II 2 3 2 7
S System I 1 1 2 4
Others 2 2
COMPCO (chemicals)
F Hardpoly 2 2 1 5
U Stretchpoly 1 1 2
S Hotpoly 1 1 2 4
Others 2 1 3
PRODCO (chemicals)
F Pit Liner 1 1 2
S Roof Liner 1 1 2
Others 1 1
TOTAL 28 19 8 30 85**
*F = failure, U = Uncertain, S = Successful.
**Several people were interviewed twice, so figures add to more than 80.
to the successes and failures. Many of these people presented their product as a
success, yet the measures of their knowledge matched the failed people's more
closely, suggesting that attribution bias does not seriously affect the measures (see
Dougherty 1987, 1990, for details). Failed developers may have underreported the
knowledge they had due to attribution bias. I conclude that memory failure and
attribution bias do not dominate the data, but they may distort the findings some-
what.
Since labels vary by firm, people were categorized into departments (used in the
same sense as "function") based on their responsibilities when they worked on the
product. Four labels are used: (1) technical: those who worked in research or
engineering; (2) field: those who worked in sales or customer relations; (3) manufac-
turing: plant and purchasing people, or manufacturing engineers and (4) planners:
those who handled market research or business analyses but were not in regular
contact with customers.
Data Analysis and Findings
The findings and methods of analysis are interdependent, and are described
together for each issue being studied.

186 DEBORAH DOUGHERTY
TABLE 3
Technology-Market
Content
Analysis of Interviews and
Cognitive Orientations by Department
Issues and
Orientations Technology Field Manufacturing Planners Statistic
% of Interview:
Business Issues 23% 24 17 42 F = 16.1
p = 0.000
Customer Issues 17 28 10 18 F = 12.9
p = 0.000
Selling Issues 4 22 3 10 **Chi 2:
p = 0.000
Technical Issues 49 21 27 23 F = 27
p = 0.000
Manufacturing 8 6 39 6 F = 71
Issues p = 0.000
Orientations
mean ranks:**
Long-Short time 34 33 25 44 **Chi 2:
p = 0.11
Task-People 39 20 43 42 **Chi 2:
p = 0.006
**The four departments do not have homogeneous variances on these measures, per a Bartlett test. The
original measures were ranked by case from 1 to 71 (71 of the original 85 interviews are used in this
analysis because "others" and people who did not work on the product prior to its launch were deleted). A
Kruskall-Wallis test that compares mean ranks was run. The mean percentage is reported for the Selling
Issues to be consistent with the other content measures, but the statistic is a Chi Square used for the K-W
test on the ranked scores. Two orientation measures are the mean ranks, with the higher rank reflecting
more long-term or task orientation.
Thought Worlds
Technology-Market
Funds of Knowledge. The first research question asks what are
the different funds of knowledge about technology-market issues in the different
departmental thought worlds? It is assumed that the content of the thought worlds
would be evident in the emphases in people's stories. To assess these emphases, a
coding scheme to measure "technology" and "market" was developed with five
categories: (1) product technology and design issues, (2) manufacturing (plant, suppli-
ers, materials), (3) business issues including segments, competition, and size forecasts,
(4) customer issues such as needs and problems, and (5) distribution or selling issues.
The author coded each statement of each interview into one of these five areas. A
research assistant coded a subset of interviews, and we had 80% agreement. Issues
we disagreed over were redefined and all interviews were recoded by the author.
Measures were also developed for two of Lawrence and Lorsch's (1967) "cognitive
orientations." Time orientation was measured by counting the number of statements
that referred to events more than one year prior to the start of the product, or one
year into the future beyond the date of the interview (divided by the number of pages
to control for interview length). This assumes that people with a longer time
orientation would discuss more long-term events. A task versus people orientation
was measured by a ratio of statements that referred to working with people versus
those that referred to a task in the development.
To make comparisons as similar as possible, only those people who worked directly
on the product prior to introduction were included in the analysis. Table 3 summa-
rizes the technology-market contents and cognitive orientations of each thought world

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 187
using this subset of 71 interviews. One-way analyses of variance suggest differences on
six measures, with long-short term orientation statistically significant at only p = 0.11.
Planners dwell on business analyses in their stories and are the most long term, while
technical people emphasize design issues. The field people discuss customer and
distribution issues more often, and are the most people oriented.
If the departments are like thought worlds, however, one would also expect to see
these differences regardless of company and success versus failure. Two-way analyses
of variance between the thought world fund of knowledge measures and orientations
were run by department and by company and success status to check on this
expectation. See Appendix B for the results. The department has a main effect in five
of the seven comparisons. There are also a few differences by company and success
status, as discussed in the Appendix. Since the department main effect is found across
the firms and success outcomes, however, these findings lend support to the idea that
departmental thought worlds differ systematically on technology-market issues re-
gardless of the firm or product status.
Systems of Meaning. The Fleck model suggests that departments not only know
different things, but also know things differently. That is, each would have a different
system of meaning through which its members interpret technology-market issues. To
understand these systems of meaning, dimensions to describe them were developed
following Strauss's (1987) methods for qualitative analysis. First, people's descriptions
of other departments and frustrations with them, how they thought about customers,
and what they considered important to product development were written onto
separate coding sheets, These were then compared across department, case, and firm
to search for underlying patterns or themes which summarized the essence of the
departmental differences. Preliminary themes were discussed with colleagues and
with several of the people interviewed.
Three final themes seem to most distinguish how the four thought worlds interpret
technology-market linking and new products: (1) what people see when they look into
the future, including issues that are most uncertain; (2) what people consider to be
the critical aspects of the product development process; and (3) how people under-
stand the development task itself. By looking at technology-market issues through the
unique combination of these themes, people in each thought world understood the
product in qualitatively different ways. Each thought world had an "intrinsically
harmonious" perspective on the product which did not overlap extensively with
perspectives held by other departments. Table 4 summarizes the unique understand-
ings about product innovation which arise from these different systems of meaning.
These themes are first outlined, and then the different Views they engender are
illustrated with case material. The first theme captures the future orientation inher-
ent in new product development. It also highlights the fact that people understand
the future by in effect sighting along different emerging trends-technological change
versus customer shifts versus market evolution (see Table 4). They make different
sense of the nebulous future by looking at disparate aspects of it. What they see
seems uncertain, while what they do not see does not seem particularly uncertain or
even noteworthy. These contrasts can be seen in the OPCO excerpts. The business
planner worried about positioning against competition while the field person worried
about identifying the right potential customers. They partitioned the product into
separate sources of uncertainty, which may have kept them from developing a more
comprehensive understanding of the market.
The second theme comprises people's understanding of the development process
itself. Each department concentrates on different subsets of the overall process.
People do not ignore the activities they do not deal with directly, and do not merely

188 DEBORAH DOUGHERTY
a
'9'''3-2'S!Xt''t:'E'2-5
~a
[
= W i . _ ~~ ~ ~
~~~~~~~~~~~i
e4- i

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 189
argue over relative priorities. Rather, they gloss over the concerns of others, and tend
not to appreciate their complexities (Van Maanen 1979). Recall that OPCO's techni-
cal person saw no "market" problems for the product effort described. Had they
identified a set of retailers who needed the service but who had simple transaction
needs, however, perhaps they could have designed a workable system. As Table 4
suggests, each department seeks inputs from the others that differ from the others'
primary focus. Technical people, for example, concentrate on solving design prob-
lems, and expect field people to tell them exactly what customers want in the design.
Field people, however, cannot identify these "specs" because in their view users are
uncertain. Rather, field people consider that product innovation is to meet shifts in
customer needs, and expect technical people to produce alternate designs quickly.
The third category concerns how people understand the task itself. As Table 4
shows, all but the planners understand product development in concrete, hands-on
terms, so all these departments have difficulty making sense of planners' reports.
There are also significant differences within these three hands-on groups, however.
The technical people think that the product has a specific reality, while field people
think it is a possibility. These contrasting perspectives can seriously impede a
dialogue over what the product is and how it should be developed. Excerpts from the
interviews illustrate these thought worlds for product development.
The Technical People. The product's design dominates the technical people's un-
derstanding of product development. When they look into the future, they see ever
emerging design possibilities and numerous technological trade-offs. For example, an
engineer with SALECO's System II discussed the decision over the disk drive at some
length:
The diskette started as a single side, but we had technical problems with that. ... Also when we
started there was no question that we'd do half size. That was a new technology so it had to be
single sided. But a guy in the group said in a few months we could fix the problems, so let's take a
risk and forget single sided and do double sided. That helped push the technology... (SALECO)
Consistent with this view of the future, technical people emphasize establishing the
product's performance specifications rather than what the customer does with the
product. This is a complex and often frustrating process, as this engineer's comments
about a communication system illustrate:
There were a lot of specs, but these were only detailed conceptually. They wanted "something
like this." What ended up as a result is that the specs get interpreted more widely. You end up
delivering something they didn't ask for. ... I was working with one or two people (at the customer
organization). Then they show it to fourteen others who say: "Oh My God!! We didn't want that!!"
(OPCO)
Technical people define the market in terms of what the product does, and may
overlook business aspects such as how many people will pay how much for the
product. For example, the market for the battery was defined as "battery users," not
particular sets of industrial customers who might need a more permanent energy
source. Most technical people view the development task as a tactile activity that
results in something objectively real. By concentrating on technological possibilities,
however, they may assume the market is obvious and that customer needs are
straightforward.
The Field People. As field people look into the future, they see constantly emerging
customer applications rather than changing technological possibilities. Such changes
are more immediate, often shifting with the user's model changes or competitor's

190 DEBORAH DOUGHERTY
product or price changes, giving them a short-term, action orientation. Field people
also seem to take a customer-by-customer view of the market. Consider:
You need to listen to what the customers want; what is he ready to buy? what is he looking
for?... You have to be specific, applications oriented. I want the least amount possible in the
shortest time, but [engineering] may take three years. ... The more (engineering and operations)
are buffered from customers, the more they tend not to understand the urgency. (OPCO)
Field people emphasize creating or crafting the sale, not the product, and describe
this activity as vividly as the engineers discuss disc drives or communication networks.
Explains a voice service field person:
It's a blast to let it go. I never go on a one on one meeting. I always try to have a minimum of
three people (from the prospective customer's organization) and I throw out functions until I find a
use. ... It's the most amazing thing in the world. Usually there's at least one guy who's determined
not to like the product. He sits pulled away from you like he won't let you penetrate his shield. But
then after ten minutes he pops up in his chair and starts coming up with ideas ... (OPCO)
To field people, the task is real but not tactile. Developing a new product means
establishing new relationships and buying-selling arrangements. The relational nature
of their work is evident in this quote:
We know what they need. The market is obvious. But the selling process is complex. Who are they
and what do they want is clear. But there are six or eight decision makers. No one says yes but
anyone can say no. ... The production guy wants to know if his yield will be better ... The quality
control guy says 'will I have to change my tests?' The sales manager says 'will my customers like the
finished product as well?' The purchasing guy says 'what will you do for me?'... You need to work
with all these guys and their bosses. (COMPCO)
Note also that the speaker downplays the complexities of doing a market analysis
by focusing on his own tasks. Field people concentrate on what the user wants to do
with the product, but also assume that customer needs are unique or constantly
changing.
The Manufacturing People. Manufacturing people worry about the plant or opera-
tions, and are concerned that the others do not appreciate their special inflexibilities.
A person at COMPCO explained that manufacturing is very concerned that market-
ing will take orders for products they cannot produce. " . . . I don't like (them) taking
risks (they) don't know (they) are taking." It seemed that these inflexibilities push
manufacturing people to live in the ever present now. The director of operations for
the voice service explained his problems with the others as follows:
Sales and marketing live in the future and my needs are today. They are forever saying "why
don't we do this?," or "isn't that easy to do?" But based on limited capacity now I can't do that. It's
the same with networking. Sales and engineering wanted to bring up all the nodes at once! We said
no, let's test it and do it one at a time ... They know, they hear, but they aren't involved (as closely).
And they don't get the 5,000 calls from customers (when the system fails)... There needs to be
more interface between those who design the future and those who live in the real world. (OPCO)
Note that, like the others quoted above, the person is aware of the others' concerns
and issues, but emphasizes his problems over theirs.
Manufacturing people concentrate on reliability and quality, evaluating and defin-
ing the product in these terms. For example, a manufacturing engineer picked up a
keyboard for the failed System II and threw it into a corner of the room to show how
well built it was. "Look at that!" he said. "That's a damn fine keyboard!" (SALECO).
It turns out that the system was a commercial failure (in part) because the keyboard
was difficult to use. How often it could be thrown around was not a factor.

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 191
The Planning People. The future that planners see consists of emerging business
possibilities such as the size of the market and total revenue potential. For planners,
both the technological trends and specific customer applications pale into abstraction.
As a planner at TECHCO put it: "We locate markets and make recommendations if
it's worthwhile to enter them. And that depends on the margins, or the amount of
money you will make." Analyzing "the business" also can be very frustrating and
uncertain, as one explained:
The environmental scan is the most difficult part. There isn't enough information available. We
looked at traditional sources (of information) including market research firms. But the problem is
they are guessing too, they develop scenarios. An awful lot of projecting from just a few numbers
goes on in this business. (OPCO)
The information the planners need for their reports is difficult to gather, so they
resort to modeling. For example, several people at OPCO spent three months
developing a model to predict the size of an electronic data transfer market, and then
they developed their business proposal around the estimate. To do such extensive
analyses with such ambiguous data, planners cannot treat design and applications as
constantly emerging. Instead they abstract these issues into more general scenarios.
The nature of the planners' work is conceptual, not concrete, which means that their
concrete-thinking colleagues will have trouble with their plans. For example, a
technical person said: "It's hard to know what to do with reports we get from
marketing." A field person said: "I'll take luck to market research any day."
Thought Worlds and Product Innovation. Interpretive differences between depart-
mental thought worlds play a strong role in problems with collaboration over
technology-market linking. From the outside looking in, one can see the conventional
stereotypes for each department in the sketches above: technical people never settle
on a design, field people are short term, manufacturing people always say no, and
planning people are conceptual. But from the inside looking out, each thought world
is truly concerned with the successful development of the product. And as can be
seen, each has an important insight into the product or market that is essential to a
new product's development. Each emphasizes different aspects of development,
however, and conceives of the whole in a different way.
When seen from the perspective of thought worlds, the collaboration problem runs
deeper than conflicts over personality types or goals. Indeed, to attempt to resolve the
problem through negotiation over goals may only begin to touch on the divergent
understandings which lay at the heart of the problem. Nor is the problem like the
proverbial set of blind men touching a different part of an elephant. It is more like
the tales of eye witnesses at an accident, or of individuals in a troubled
relationship-each tells a "complete" story, but tells a different one. Despite their
potential benefits, departmental thought worlds separate the market-technology is-
sues, limiting the possibility of a comprehensive understanding. The thought worlds
also focus inward, reducing the possibility of learning. These data suggest that, to
overcome the thought worlds' "inherent tenacity" (Fleck 1979) to focus on their own
perspectives, managers may need to proactively foster collective action.
Organizational Routines
It became clear from the research that the thought worlds did not operate
independently from the organizational context, however. Next, the analysis seeks to
identify the organizational routines that affect product innovation, and how they
affect collective action among the thought worlds. Most people described how
products are usually developed in their firm, so the data provide some insight into

192 DEBORAH DOUGHERTY
routines. Descriptions of usual practice for product development (including evaluat-
ing the product, making decisions, and determining market needs) were transcribed
onto coding sheets. The same kinds of practices used for the new product were also
extracted, and compared with the usual practices by success status.
The analysis suggested three important patterns. First, three routines encourage
thought world separation and inhibit learning. These routines were systematically
violated in the successful cases, followed in the failed cases, and followed partly in
uncertain cases. Second, the successful innovators did create a new social order for
innovation. Third, the routines in these firms were very strong. Breaking out of them
to establish an innovative social order seemed to be an unusual and often temporary
event.
Organizational Routines, New Products, and Thought Worlds. The three routines
found to affect product innovation concerned interdepartmental relations, market
definition, and product standards. The first routine governed thought world relations
by prescribing narrow roles and limited relationships. People would routinely do their
own work and expect the same of others. Even when interactive structures such as
task forces were used routinely, relationships were constrained, so the possibility
of creating a new collective order which produced cross-fertilization and mutual
learning was limited. The second routine imposed a predetermined definition of
technology-market issues on product efforts. This reduced people's search for new
information as well as Lie likelihood of frame-breaking learning. The third routine
imposed standards which did not fit these new products. The standards varied by
firm, and included set pay-back periods, profit margins, quality, and use of inhouse
facilities. Following them forced developers to redefine the new product as an
established business, further reducing new learning.
Successful developers violated all three routines, and created a new social order for
their collaborative efforts. They developed mutually adaptive interactions in which
knowledge of the work was developed as the work unfolded, as Mintzberg (1979)
describes. They also created an alternate definition of the product in its market that
was grounded in actual use, and developed appropriate standards to evaluate their
efforts. The outcome was collective action as Fleck (1979) describes it: each thought
world rechanneled its readiness for directed perception from an inward to an
outward focus so that their knowledge could join to produce new insights and new
facts.
An Example of Routines at SALECO: System I vs. System II. To illustrate the
routines and how successful developers stepped out of them to create a new social
order, SALECO's successful System I is contrasted with its failed System II. Both
products used similar technology and some of the same people, and the failure
followed the success, so the comparison holds constant some of the effects of
individuals, technology, and experience. Appendix A summarizes how the other cases
related to their organization's routines.
SALECO is a computer and communications manufacturer which has dominated
several niches in these markets. The director of market research explained that, until
five or ten years previously, SALECO's products were self-contained "turnkey"
operations. SALECO maintained close relations with their dependent, installed base
of business users through an extensive sales force. These customers' changing needs
were easily monitored with in-depth customer profiles; product development became
a sequential and highly organized process which abstracted customers' needs into
financial indicators. A planner explained that they would look at a price-performance
curve, and come up with a new product that sits further down the curve. "We would
know what the market requirement is, and would invent nuances on the technology to

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 193
do it." Predetermined goals and specifications were used to coordinate complex
interdepartmental relationships, managed in a matrix structure. "The market" was
given, so little energy would routinely go into feretting out user needs. And, befitting
a company with market dominance, product norms included high profit margins, high
quality production, and complete control of the product by handling all aspects
inhouse.
The successful System I violated these carefully orchestrated, big business-oriented,
financials-focused product routines, while the failed System II followed them. One
person explained the overall routine violation as follows:
The unique thing was they cut off (the System I group) from the culture. Basically a few top
executives decided to play Daddy Warbucks. They disconnected (the team leader) from the normal
process of building business cases.
The System I team's interdepartmental relationships encouraged ideas to be heard
and built on and enhanced appreciation of one another's contributions. They inter-
acted extensively, and, according to a field person, all participated in all the aspects of
the product. Thus, rather than separating various concerns into pre-established role
behavior, all problems and ambiguities were addressed from all angles, producing a
more comprehensive design. Rather than coordinate by the usual formal but abstract
plans,. one member said "
.. . we had no formal business plan, but everybody knew
what it was." They did not create a single group mind, however. One explained that
each department considered itself to be the one most important to the project, but
that the leader fostered a sense of appreciation:
. . . (the team leader) always listened to everyone when it came to making a decision, so even if his
decision differed from your recommendation, you knew your information was as important as the
others'.
Along with redefining relationships, they redefined the task, and perhaps stepped
out of the usual political truce that Nelson and Winter (1982) describe. The group
created a vivid view of the product in its market which was more simple yet more
realistic than SALECO's usual abstractions of price-performance ratios, market
penetration estimates, and volume projections. The team used these general data, but
combined them with first-order, direct experience with customers. An engineer
describes their technology-market linking in what is almost a field person's perspec-
tive, as follows:
The first thing was to define what the product was, who would buy it, and what they would use it
for ... You have to get into the hearts and minds of users... If you can't explain the product in 30
seconds, you're dead.
People's descriptions of their development process were full of instances in which
they broke out of usual perspectives. For example, they broke out of SALECO's
notion of customers as business people, and came to appreciate the new users for
themselves. According to a planner:
I remember our first focus group. It was a riot. There was this guy out there in a green T-shirt,
long side burns and a flat top, jeans, a belt with a big silver buckle, and cowboy boots. He happened
to be the president of the local micro-computer club. It's frightening when you realize that on the
other side of the oneway mirror there was a room full of men in [conservative business attire]. We
had to understand that that guy was our new customef. It took a leap of faith.

194 DEBORAH DOUGHERTY
Third, they broke most of SALECO's product norms and created new standards
appropriate to the System I market. Instead of the usual inhouse manufacture which
"made equipment you can drop off ten story buildings," according to one, the
machine was assembled from off-the-shelf parts purchased externally. This more
effectively met the needs of low price in a market where "industrial strength"
durability was not critical. Instead of total control by SALECO, they designed an
open architecture to allow others to write software for sale. The group even published
a book explaining how outsiders could build on the machine with their own products.
Instead of careful quality control, some of the external software had bugs in it-the
number of applications was more important to this market than perfect operation.
Less than a year later, this business unit began work on a follow-on product aimed
at both professionals and home use. A number of new people were involved, but so
were many of the System I group, including the leader and the senior people who
played "Daddy Warbucks." More like SALECO's usual product development, how-
ever, the plan was simply to continue the momentum begun by System I. Interdepart-
mental relationships seemed to have also reverted to routine, and the inwardness of
thought worlds dominated the people's work. When asked to describe the System II
organization, an engineer went to his board and drew a circle to show the System I
team. Then he drew an arrow over to a matrix to represent the System II organiza-
tion, with columns for departments, and rows for various products under develop-
ment-SALECO's routine matrix form. A planner explained that some people from
the first effort were not fully focused on this product, that several key people left, and
that the group had lost what he called its "group think."
For the technology-market definition, System II developers relied on assumed
applications and users, as usual. A market had been "analyzed" with the routine
calculus of price-performance trade-offs and buyer potential, and not questioned
again. The design had unique features but these were not tested with users. As it
turns out, few users liked t-he product. An engineer explained:
We didn't get the system into real scenarios to test out our premises. We were overconfident...
We all thought we were very smart ... We made a lot of decisions daily to change the product based
on what we thought we understood about the market place.
A market researcher noted that perhaps some focus groups would have picked up the
problem they found after introduction with certain features. But he said that the
design and manufacturing process had generated such a momentum, that:
We would have been disappointed with the negative feedback, but we would have gone on with
(the design) anyway.
Third, the System II team followed some of SALECO's usual product standards
that were appropriate to its large systems, but not, perhaps, for System II. In-house
manufacturing rather than assembly was used, and reliability and durability were
emphasized. According to a manufacturing person:
The product was fantastic. But it was high priced. You can't cut price or quality. There is no way
SALECO would sell a lower quality product. But then people would say why should I pay so much
for this computer, instead of paying attention to all the enhancements we put on it... It has
half-size disk drives... Our fundamental error was in the design of [a certain part]. It didn't have
user flexibility. That got us a lot of negative press. But the function was great. It should have sold
like hotcakes.

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 195
By following the usual product development routines, the System II team produced
a high quality, rather expensive but durable machine-just what SALECO's routines
are designed to produce. Unfortunately the team did not begin with an established
market, which the routines also presume, so the system did not fit any real needs.
Sales were not high enough to cover the manufacturing costs, so the company shut
down the plant, laid off the factory workers, and cancelled the product.
Discussion
An extensive literature tells managers how they ought to develop new products,
and how they ought to design their organizations for innovation. This study has
examined product innovation in practice in order to understand why these prescrip-
tions are not often achieved. The research is limited by its exploratory nature, and the
fact that commitment, leadership, politics, and other possible factors have not been
included. However, it describes how two interpretive schemes can become barriers to
effective technology-market linking. Departmental thought worlds partition the infor-
mation and insights. Each also has a distinct system of meaning which colors its
interpretation of the same information, selectively filters technology-market issues,
and produces a qualitatively different understanding of product innovation. Organiza-
tional product routines reinforce thought world separation by providing for only
limited interaction, and further inhibit the kind of collective action that is necessary
to innovation.
Implications for Theory and Practice
This research suggests two important implications for the study and practice of
innovation. The first is that theorists and practitioners need to pay attention to the
effects of thought worlds and organizational routines. Innovation is an interpretive
process, so the management of innovation must involve the management of the
interpretive schemes that shape and frame how people make sense of their work.
Innovation requires collective action, or efforts to create shared understandings from
disparate perspectives. The advocation of rational tools and processes, the infusion of
market research information, and the redesign of structures, while important, are not
enough.
The second implication is that the potential barriers these interpretive schemes
may become need to be dealt with specifically and in depth. This study suggests three
intermediary processes which together can help overcome the barriers. Innovators
must: (1) use and build on the unique insights of each thought world, (2) develop
collaborative mechanisms that deal directly with the interpretive as well as structural
barriers to collective action, and (3) develop an organizational context for collective
action that enables both. The work suggests that, unless all three processes occur
together, the thought world boundaries and routines may dominate. If so, pitfalls will
be overlooked, meanings lost, and communication curtailed, as was the experience of
OPCO's accounting system developers.
Building on the Thought Worlds. Each thought world knows about aspects of
technology-market knowledge that others may gloss over. All must actively contribute
to the product design, and actively challenge each other, or the final design will be
awry. This means that reservations or skunkworks comprised solely of R&D people,
as some theorists suggest (e.g., Galbraith 1982), would not be effective. The manage-
ment of innovation is not merely the management of the R & D group or the
coordination of activities, it comprises the collective creation of the product in its
market.

196 DEBORAH DOUGHERTY
This analysis has gone beyond the general differences in orientations usually cited
to identify specific interpretive differences that must be confronted and overcome for
new products. As argued here, innovators from different departments may not
conflict over general goals or even over such intermediate goals as being "market
oriented." They do, however, have unique interpretations of these goals, along the
themes described in Table 4. Technical and field people focus on customers and have
a concrete view of development. This means that reports from the conceptual market
researchers may not make any sense to them, and may be ignored. Field people's
close view of customers could contribute important insights to design, but it may also
keep them from summarizing specifications that technical people need. Technical
people's search for a general design makes them seem unresponsive to field people's
requests for specific changes. Manufacturing people worry that others do not under-
stand the limits of the plant, and so may dig in their heels unless their concerns are
handled openly. These differences can preclude the development of an optimal
design by producing severe frustrations and perhaps withdrawals into separate
thought worlds. They must be deliberately and directly addressed as they come up.
Developing Collaboration Mechanisms. Collaboration mechanisms need to take into
account the interpretive dynamics that separate the thought worlds. Their "intrinsic
harmony" prompts people to focus inward on their own tasks, and to fill in unknowns
from their thought world. Thought worlds most likely always exist where tasks are
specialized, so efforts to surmount them always have to be made. Participants on
interdisciplinary teams may feel "drawn inward" into their thought world over time,
especially if they have other assignments, as is often the case. The analysis suggests
one important way to overcome the inward dynamic. As in the SALECO System I
case, thought worlds can come to a similar understanding of the product in the user's
hands, and innovators can perhaps build a comprehensive appreciation from this
common view. Interdisciplinary responsibility for focus groups, market research plans,
technology audits, and visits with users should enhance collaboration. Structures
alone such as liaison and boundary spanner roles or project groups do not assure that
these dynamics of separation will be overcome. A realistic customer focus may.
Developing the Context for Collective Action. This research also suggests that
thought world barriers cannot be overcome unless aspects of the organizational
context which foster separation are also overcome. The capacity for collective action
depends in part on organization level interpretive dynamics. It is inferred that an
innovative social order must be designed specifically for new products so that:
(1) interactions between thought worlds are based on appreciation and joint develop-
ment, (2) product definitions are based on collective, first-order customer knowledge,
and (3) product norms are based on the specific market. It is inferred further that
general organizational attributes such as a risk-taking climate, visionary leadership,
and/or an integrative culture may not address these day-to-day practices, although
perhaps they increase innovative activity. Moreover, routines, like thought worlds, are
probably inertial, and thus always have to be overcome. No matter what the general
organizational design, managers must foster ongoing processes of knowledge develop-
ment, joint learning, and customer interactions.
Future Research
Research is necessary to clarify and test all these implications. This study has not
considered the structural aspects of thought worlds that might affect their collective
action. Fleck (1979), for example, theorizes that stability of membership, the strength
of the established beliefs, the clarity of boundaries, and power relationships within
the thought world can all have an effect. For departments, one could speculate that

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 197
people's training, experience, and professionalization would affect thought world
separation or integration. Studies might explore whether and how job rotation, team
experience, and bringing the tacit thought world differences to conscious awareness
improve the propensity for collective action. Research on the R & D-manufacturing
relationship is perhaps the most advanced (Adler 1990, Dean and Susman 1989), and
these insights can be elaborated to include all four departments. Other approaches
for mapping frames of reference (Dunn and Ginsberg 1986) should expand our
understanding of departmental differences.
The implications for organizational routines are the most potentially confounded
by attribution bias and the limited number and types of firms in the study. Additional
research is essential to define the routines that inhibit innovation and articulate more
clearly how they work. Research should also identify the routines, if any, which can
foster innovation. The literature contains numerous discussions of general organiza-
tional attributes and phenomena that presumably engender innovation (e.g., Kanter
1983; Galbraith 1982). An important next step is to understand the relationships
between such general contexts and the specific tasks of innovation, including technol-
ogy-market knowledge development and thought world collective action. Focused
research which examines these dynamics directly, and longitudinal research which
sorts out the causal relationships between interpretive barriers and other factors are
necessary.
In conclusion, this work indicates that improvements in the new product success
rate are possible if innovators combine the prescriptive models with attention to these
interpretive barriers, and the ongoing processes necessary to overcome them. Rather
than test theory, this study has applied theory to the problem of new product
development. The insights are more practical than theoretical, but the findings
support the growing literature that understanding the interpretive dynamics of
innovation and change is crucial (Bartunek 1984; Schein 1985; Barley 1986). The
results show that such dynamics apply to product innovation, and can be studied
systematically. Continued research into these interpretive dynamics, along with con-
tinued efforts to join organization theory with insights from marketing and technol-
ogy, should advance our understanding of the complex problem of product innovation
in large firms.
Acknowledgements
The data on which this paper are based come from a study sponsored by the Marketing Science Institute
and a pre-doctoral fellowship from the American Association of University Women. A grant from the
Reginald Jones Center and a Doris and Michael Goldberg Fellowship from the Snider Center for
Entrepreneurial Studies, both of the Wharton School, supported this analysis. I would like to thank Ed
Schein, John Van Maanen, David Thomas, Larry Hirschhorn, Stew Friedman, and Marsha Wilkof for
insightful critiques. This paper has also benefitted significantly from critiques by editor Richard Daft and
three anonymous reviewers for Organization Science.
Appendix A. Summary of Cases and Their Relationships with Routines
This appendix provides a thumbnail sketch of products in this study. See Dougherty (1987) for a
complete description of each case, and Dougherty (1990) for an analysis of development processes. Since
the relationships with organizational routines are important, so the cases are reported by company rather
than success status. All firms and products are disguised.
TECHCO: Producer of chemical based products. Routines for product development stress technological
innovativeness, and market analyses routinely not done. High quality was also a standard. Product
development usually begins in R&D; procedure was to include others in sequence as the product was
ready for a business analysis and marketing plan. People interviewed in Fall 1985.
1. Failed Battery. A technologically unique no-leak battery; originally invented for a product line, but
because of excess production capacity, TECHCO decided to sell battery to others. Toy manufacturers were

198 DEBORAH DOUGHERTY
selected as market since toys were a big user of batteries; no analyses of needs, problems, etc. Introduced
in late 1981. Product required redesign of toys, and was not widely available so toy companies were not
interested. Cancelled after 3 years. Routines. A technology dominant task force used for development,
sponsored by senior VP; market definition based on what the battery could do, not what users needed;
standards of uniqueness and high quality met.
2. Uncertain CRT Device. Electronically transfers computer images to produce a hardcopy. Began in
R&D. Introduced in Fall 1983. Sales very slow; an interdisciplinary team was set up after introduction, and
they were redesigning the product and changing the distribution system. Routines. Technology driven,
limited interdepartmental interaction; marketing research advised against the product but were ignored.
Market defined as white collar office workers; development delayed until quality standards could be met.
3. Uncertain Video Device. Similar to CRT device, but captures television signals. Also began in R&D,
developed for home use. Commercial division took it, so product redesigned for commercial use. Were
design, manufacturing, and quality problems. Concerns over quality and technical design persist. Routines.
Partially broke routines by doing a market test jointly with R&D and marketing people; interaction
seemed good; hung up on established product standards based on chemicals (this was electronic).
4. Uncertain Medical Hardcopy System. A chemical and mechanical technology that produces hardcopy
from medical diagnostics machines. Requires redesign by diagnostics manufacturers. Delayed for over a
year because of manufacturing problems; in the interim a competitor came out with similar product at
much less cost. Now looking for alternate markets and uses. Routines. Set up an interdisciplinary task
force; good interaction except with manufacturing; market plans good but limited to existing markets,
contrary to original plan. Hung up on usual markets and standards.
5. Successful Film Cover. Adapts a light-proof paper wrap TECHCO makes for other products to
contain professional grade photography film purchased from others. Very "low-tech" but intended for new
markets and applications. Test market sales exceeded projections, so introduction advanced by 6 months.
Routines. Began in business unit; product champion incorporated manufacturing and purchasing people on
informal or "bootleg" team; all visited users together; considerable market data gathered; design adjusted
to meet particular problems. Violated product approval and quality procedures.
OPCO: An operating phone company whose routines stressed low risk ventures, short-term planning,
and quarterly operations reviews of all businesses and senior managers. Corporate requires extensive
analyses of all new business plans. OPCO's Data Division evolved a "fast-out" product routine to avoid the
"corporate gauntlet." They sneaked their first product, the Text System, into the market. Interviews from
February to April 1986.
1. Uncertain Voice System. (Not associated with OPCO Data) Electronically transfers voice to digital-
ized packets transmitted over network. One of the first uses of this technology, introduced in 1983; still
losing money. Routines. Began by a planner; general plans with no applications analyses (files reviewed by
author). No interdisciplinary work. During test market product redesigned for discovered market of
interoffice uses; multiple market definitions still operate. A venture group formed 2 years after test market.
2. Uncertain Text System. Similar technology to voice but transfers text. Introduced in the early 1980's;
sales growth high but competition is strong and earnings uncertain. Routines. Developed by a technical
skunkworks, and introduced as a technology. Market plan based on the number of salespeople, not users or
needs. Violated OPCO corporate's analysis, but followed technology driven, fast out routines in division.
No interdisciplinary group or comprehensive market definition; these now being developed.
3. Failed Document System. Allows text system users to produce a hardcopy delivered overnight to
people not on the network. System failed since some mail took 2 weeks; removed from the market.
Routines. Business planning and technical development done separately. Technical and operational plan
poor but chosen because it could be done by the end of the quarter. Original market defined in vague
terms.
4. Failed Accounting System. Transacts credit card purchases for retailers over a network. System
double-debited and double-credited, and then crashed. Removed from market. Routines. Followed usual
planning process with little connection between technology and marketing; business plan was complete, but
particular applications not examined for design. Followed fast-out standard.
5. Uncertain Transmit
System. Product still in development. It will transmit data between manufacturers
and suppliers. Plan created by an interdisciplinary team. The plan violates OPCO's standard for a 2-year
payback, so developers were unsure if it would be approved.
6. Uncertain Software System. Product transfers software over the network, from terminal to terminal or
downloading from a central file. Just introduced, and is being given 90 days to "make" it. Market very
uncertain. Routines. Developed by a planner, with some interaction from technical people. Targeted
market has changed because software firms initially agreed to cooperate but then refused. Market
definition very general-now the "Fortune 1000." Followed standards of general business analyses.
SALECO: see text. Interviews Summer 1986.
COMPCO: Chemical materials producer. Has a dominant market position with a certain grade of
plastic. Product development usually begins with requests from customers processed through field, and

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 199
these requests are responded to quickly. Market need is always known, so little analysis is routinely done.
COMPCO has 102 variations of basic plastic. Interviews Fall 1986.
1. Failed Hardpoly. Combines a harder polymer with its basic one, giving the plastic more strength yet a
lower cost than the stronger polymer alone. Previous efforts to combine these polymers could not overcome
certain problems. The material was introduced into an application but failed the heat test and was
withdrawn. Routines. The material began in technical, not field, but was then treated as a typical user
request. An interdisciplinary task force was appointed, as is the routine, but there was little interaction
between departments, and field assumed that it would perform as they expected.
2. Uncertain Stretchpoly. Combines a flexible polymer with COMPCO's basic material, for added
flexibility and water resistance. The material was introduced 2 years ago into automotive, but does not
meet those needs, and new niches are being explored. Routines. Like hardpoly, the material began in
research, contrary to routines, so it had no clear market demand. Some market research was not used, and
they assumed it would fit into their major market, the automotive business. An interdisciplinary project
team has since been set up, and they are actively monitoring the product.
3. Successful Hotpoly. This is a new kind of polymer for COMPCO, with much greater heat resistance,
and it opens up new market areas for the firm. The material was introduced in later 1982; only 1 other firm
made it. Initial sales slow; field was not selling the product, and they overestimated users' willingness to
TABLE B-1
Thought World Measures by Company
Weighted Mean Score ANOVA results
Thought World Dept effect
Measures SALECO OPCO COMPCO TECHCO Company effect
Interaction effect
(prodco, manuf not
% of Interview: included)
Business Issues 32 27 29 33 D F= 12.7*
C 0.9
I 1.96**
Customer Issues 16 23 23 19 D F= 6.6*
C 2.5**
I 1.2
Selling Issues 12 12 17 9 D F= 12.5*
C 1.6
I 1.6
Technical Issues 32 32 25 29 D F= 31.3*
C 1.5
I 0.4
Manufacturing 7 5 6 8 D F= 0.5
C 1.8
I 0.3
Orientations
mean ranks:
Long-Short time 37 32 37 38 D F = 0.3
C 0.7
I 1.8
Task-People 38 29 32 38 33 D F = 2.6**
C 1.0
I 1.4
*p < 0.05.
**p < 0.1; Department df =2,48; Company df = 3,48: interaction df =6,48.

200 DEBORAH DOUGHERTY
adopt a new product. With changes in sales-rewards and market analyses hotpoly has become successful,
and COMPCO is building a multi-million dollar plant. Routines. A new task force was set up from the
beginning with everyone but field involved; senior management gave strong support; new business and
market plans were made.
PRODCO: A chemical materials producer. The five interviews were not adequate to develop a general
appreciation of PRODCO routines. Interviews Fall 1986.
1. Failed Pit Liner. The product is a membrane-like material that lines industrial waste pits, creating
large containment areas. Previous materials had to be buried a foot underground to avoid exposure to the
sun. A supplier created a new material that resisted the sun, or so they thought. PRODCO entered the pit
business with the new material in the early 1970's. After 3 years, material decayed. All pits had to be
replaced. It seems that usual routines were followed, but in this case failure was an "act of God."
2. Successful Roof Liner. The product is a membrane-like material (but does not decay in the sun) used
for industrial roofs. It was introduced in 1981 and has become a multi-million dollar business. Routines. A
separate business unit was created with people from all departments in it. They worked together
extensively, and the product was designed based on first-order data from users. They violated PRODCO's
usual distribution process, planning process, and reporting structure. The team continues to run the
business.
Appendix B. Thought Worlds Controlling for Company and Success or
Failure of Product
The market-technology knowledge and cognitive orientations were analyzed with a two-way ANOVA by
department and company to see if the thought world differences held across company. Manufacturing
people and the 4 PRODCO people were deleted from the analysis since there are so few of them, so the
TABLE B-2
Thought World
Measures
by Success Status
Weighted Mean Score ANOVA results:
Thought World Dept effect
Measures Successful Uncertain Failed Success effect
Interaction effect
% of Interview:
Business Issues 32 25 24 D F = 13.2*
S 2.0
I 1.38
Customer Issues 20 21 15 D F = 4.12*
S 8.1*
I 0.99
Selling Issues 13 10 8 D F = 9.7*
S 1.3
I 1.3
Technical Issues 23 29 30 D F = 20.7*
S 4.8*
I 1.3
Manufacturing 14 14 17 D F = 71.3*
S 2.5
I 2.3
Orientations
mean ranks:
Long-Short time 46 32 30 D F = 3.08*
S 2.5**
I 0.72
Task-People 25 40 44 D F = 3.6*
S 3.4*
I 0.60
*p < 0.05.
**p < 0.1; Department df = 3,59; Success df = 2,59; Interaction df = 6,59.

INTERPRETIVE BARRIERS TO SUCCESSFUL PRODUCT INNOVATION 201
results are for 4 firms and 3 departments. The figures in Table B-1 are the means by company, weighted by
thought world-the mean for planners, technical and field was added and divided by 3 to adjust for the
different proportions of thought worlds in each firm. There is one interaction effect, and subsequent
analysis shows that OPCO's field people are below average on business issues. There is also one company
effect; analysis indicates that SALECO is below average on customer issues. The results, overall, however,
suggest that the differences by department reported in Table 3 hold regardless of company. These analyses
are exploratory only since there are problems with the variances.
Table B-2 summarizes the means by success status (again, weighted so each thought world has an equal
effect) on the 7 thought world measures. Analyses were also run without the manufacturing people to
check for problems with the small cell sizes, and the results are similar. The department effect is found for
all 7 measures. Successful developers also emphasize customer issues more and technology and manufac-
turing issues less. In addition, the successful developers are long term and people oriented, consistent with
the qualitative result that they have a more comprehensive technology-market knowledge and more
collaborative relationships. Since the successful cases have been out longer, the results may be biased by
time. To check for this, uncertain and failed cases which have been out for less than 2 years were deleted
and the analysis rerun. The results were similar, which suggests no time bias.
References
Adler, P. (1990), "Shared Learning," Management Science, 36, 938-957.
Argyris, C. and D. Schon (1978), Organizational Learning: A Theory of Action Approach, Reading, MA:
Addison Wesley.
Bailyn, L. (1977), "Research as a Cognitive Process: Implications for Data Analysis," Quality and Quantity,
11, 97-117.
Barley, S. R. (1986), "Technology as an Occasion for Structuring: Evidence from Observations of CT
Scanners and the Social Order of Radiology Departments," Administrative Science Quarterly, 31,
78-108.
Bartunek, J. M. (1984), "Changing Interpretive Schemes and Organizational Restructuring: The Example
of a Religious Order," Administrative Science Quarterly, 29, 355-372.
Bonnet, D. (1986), "Nature of the R&D/Marketing Co-operation in the Design of Technologically
Advanced New Industrial Products," R&D Management, 16, 117-126.
Clark, K. (1985), "The Interaction of Design Hierarchies and Market Concepts in Technological Evolution,"
Research Policy, 14, 235-251.
Cooper, R. (1983), "A Process Model for Industrial New Product Development," IEEE Transactions on
Engineering Management, 30, 2-11.
_ and E. Kleinschmidt (1986), "An Investigation into the New Product Process: Steps, Deficien-
cies, and Impact," Journal of Product Innovation Management, 3, 71-85.
Daft, R. and K. Weick (1984), "Toward a Model of Organizations as Interpretive Systems," Academy of
Management Review, 9, 43-66.
Dean, J. and G. Susman (1989), "Organizing for Manufacturable Design," Harvard Business Review,
(January-February), 28-37.
Dearborn, D. and H. Simon (1958), "Selective Perception: A Note on the Departmental Identification of
Executives," Sociometry, 140-144.
Dougherty, D. (1987), "New Products in Old Organizations: The Myth of the Better Mousetrap in Search
of the Beaten Path," unpublished Doctoral Dissertation, MIT, Sloan School of Management.
(1990), "Understanding New Markets for New Products," Strategic Management Journal, 11,
59-78.
Douglas, M. (1987), How Institutions Think, London: Routledge and Kegan Paul.
Dunn, W. and A. Ginsberg (1986), "A Sociocognitive Network Approach to Organization Analysis,"
Human Relations, 40, 955-976.
Fleck, L. (1979), Genesis and Development of a Scientific Fact, T. Trenn and R. K. Merton (Eds.), translated
by F. Bradley and T. Trenn; originally published 1935; Chicago: University of Chicago Press.
Freeman, C. (1982), The Economics of Industrial Innovation, Cambridge, MA: MIT Press.
Galbraith, J. (1982), "Designing the Innovative Organization," Organization Dynamics, (Winter), 5-25.
Glaser, B. and A. Strauss (1967), The Discovery of Grounded Theory, Chicago: Aldine.
Gupta, A., S. Raj and D. Wilemon (1986), "R&D and Marketing Managers in High-Tech Companies: Are
They Different?", IEEE Transactions on Engineering Management, 33, 25-32.
Harrison, F. (1980), "Goal Orientations of Managers and Scientists: An Illusory Dichotomy," IEEE
Transactions on Engineering Management, 27, 74-78.
Henderson, R. and K. Clark, (1990), "Architectures for Innovation: The Reconfiguration of Existing
Product Technology and the Failure of Existing Firms," Administrative Science Quarterly, (March),
(forthcoming).

202 DEBORAH DOUGHERTY
Kanter, R. M. (1983), The Change Masters, New York: Simon and Schuster, Inc.
Lawrence, P. and J. Lorsch (1967), Organization and Environment, Boston: Harvard School of Business
Administration Press.
Lilien, G. and E. Yoon (1988), "Determinants of New Industrial Product Performance: A Strategic
Re-examination of Empirical Literature," IEEE Transactions
on Engineering Management, 36, 3-10.
March, J. and H. Simon (1958), Organizations, New York: Wiley and Sons.
Mintzberg, H. (1979), The Structuring of Organizations, Englewood Cliffs, NJ: Prentice Hall.
Nelson, R. R. and S. G. Winter (1982), An Evolutionary Theory of Economic Change, Boston: Belknap
Press.
Perrow, C. (1986), Complex Organizations: A Critical Essay, (3rd ed.), New York: Random House.
Quinn, J. B. (1985), "Managing Innovation: Controlled Chaos," Harvard Business Review, (May-June),
73-84.
Riley, P. (1983), "A Structurationist Account of Political Culture," Administrative Science Quarterly, 28,
414-437.
Rothwell, R., C. Freeman, A. Horsley, V. Jervis, A. Robertson and J. Crawford (1974), "SAPPHO
Updated. Project SAPPHO Phase II," Research Policy, 32, 58-291.
Schall, M. (1983), "A Communication Rules Approach to Organizational Culture," Administrative Science
Quarterly, 28, 557-581.
Schein, E. (1985), Organizational Culture and Leadership, San Francisco: Jossey-Bass.
Souder, W. (1987), Managing New Product Innovations, Lexington, MA: Lexington Books.
Strauss, A. (1987), Qualitative Analysis For Social Scientists, Cambridge: Cambridge University Press.
Tushman, M. (1978), "Technical Communication in R&D Laboratories: The Impact of Project Work
Characteristics," Academy of Management Journal, 21, 624-645.
and D. Nadler (1986), "Organizing for Innovation," California Management Review, 28, 74-92.
Urban, G. and E. von Hippel (1988), "Lead User Analyses for the Development of New Industrial
Products," Management Science, 34, 569-582.
Van Maanen, J. (1979), "On the Understanding of Interpersonal Relations," In W. Bennis, J. Van
Maanen, E. Schein, and F. I. Steele (Eds.), Essays in Interpersonal Communication, Homewood, IL:
Dorsey Press, 13-42.
and S. R. Barley (1984), "Occupational Communities: Culture and Control in Organizations,"
Research in Organizational
Behavior, 6, Greenwich, CT: JAI Press, 287-365.
Yin, R. (1989), Case Study Research: Design and Methods, Newbury Park, CA: Sage.