Page 1
Patterns of Industrial
Innovation William J. Abernathy and James M. Utterback
How does a company’s innovation-and its response to
innovative ideas-change as the company grows and
matures? Are there circumstances in which a pattern
generally associated with successful innovation is in
fact more likely to be associated with failure? Under
what circumstances will newly available technology,
rather than the market, be the critical stimulus for
change! When is concentration on incremental
innovation and productivity gains likely to be of
maximum value to a firm? In what situations does this
strategy instead cause instability and potential for crisis
in an organization?
Intrigued by questions such as these, we have examined
how the kinds of innovations attempted by productive
units apparently change as these units evolve. Our goal
was a model relating patterns of innovation within a
unit to that unit’s competitive strategy, production
capabilities, and organizational characteristics.
This article summarizes our work and presents the basic
characteristics of the model to which it has led us. We
conclude that a productive unit’s capacity for and
methods of innovation depend critically on its stage of
evolution from a small technology-based enterprise to
a major high-volume producer. Many characteristics of
innovation and the innovative process correlate with
such a historical analysis, and on the basis of our model
we can now attempt answers to questions such as
those above.
A SPECTRUM OF INNOVATORS
Past studies of innovation imply that any innovating
unit cs most of its innovations as new products. But
that observation masks an essential difference: what
constitutes a product innovation by a small,
technology-based unit is often the process equipment
adopted by a large unit to
improve its high-volume production of a standard
product. We argue that these two units-the small,
entrepreneurial organizations and the larger unit
producing standard products in high volume- are at
opposite ends of a spectrum. in a sense forming
boundary conditions in the evolution of a unit and in
the character of its innovation of product and process
technologies.
One distinctive pattern of technological innovation is
evident in the case of established, high-volume
products such as incandescent light bulbs, paper, steel,
standard chemicals, and internal-combustion engines,
for examples.
The markets for such goods are well defined; the
product characteristics arc well understood and often
standardized; unit profit margins are typically low;
production technology is efficient, equipment intensive
and specialized primarily on the basis of price. Change
is costly in such highly integrated systems because an
alteration in any one attribute or process has
ramifications for many others.
In this environment innovation is typically incremental
in nature, and it has a gradual, cumulative effect on
productivity. For example, Samuel Hollander has
shown that more than half of the reduction in the cost
of producing rayon in plants of E. I. du Pont de
Nemours and Company has been the result of gradual
process improvements which could not be identified as
formal projects or changes. A similar study by John
Enos shows that accumulating incremental
developments in petroleum refining processes resulted
in productivity gains which often eclipsed the gain from
the original innovation. Incremental innovations, such
as the use of larger railroad cars and unit trains, have
resulted in dramatic reductions in the cost of moving
large quantities of materials by rail. In all these
examples, major systems innovations have been
followed by countless minor product and systems
improvements, and the latter account for more than
half of the total ultimate economic gain due to their
much greater number. While cost reduction seems to
have been the major incentive for most of these
innovations, major advances in performance have also
resulted from such small engineering and production
adjustments.
Page 2
Such incremental innovation typically results in an
increasingly specialized system in which economies of
scale in production and the development of mass
markets are extremely important. The productive unit
loses its flexibility, becoming increasingly dependent
on high-volume production to cover its fixed costs and
increasingly vulnerable to changed demand and
technical obsolescence.
Major new products do not seem to be consistent with
this pattern of incremental change. New products
which require reorientation of corporate goals or
production facilities tend to originate outside
organizations devoted to a “specific” production
system; or. if originated within, to be rejected by them.
A more fluid pattern of product change is associated
with the identification of an emerging need or a new
way to meet an existing need; it is an entrepreneurial
act.
Many studies suggest that such new product
innovations share common traits. They occur in
disproportionate numbers in companies and units
located in or near affluent markets with strong science-
based universities or
other research institutions and entrepreneurially
oriented financial institutions. Their competitive
advantage over predecessor products is based on
superior functional performance rather than lower
initial cost, and so these radical innovations tend to
offer higher unit profit margins.
When a major product innovation first appears,
performance criteria are typically vague and little
under- stood. Because they have a more intimate
understanding of performance requirements, users
may play a major role in suggesting the ultimate form
of the innovation as well as the need. For example,
Kenneth Knight shows that three-quarters of the
computer models which emerged between 1944 and
1950, usually those produced as one or two of a kind,
were developed by users.
It is reasonable that the diversity and uncertainty of
performance requirements for new products give an
ad- vantage in their innovation to small, adaptable
organizations with flexible technical approaches and
good external communications, and historical evidence
supports that hypothesis
