C H A P T E R TWO
Value Networks and the
Impetus to Innovate
From the earliest studies of the problems of innovation, scholars,
consultants, and managers have tried to explain why leading
firms frequently stumble when confronting technology change.
Most explanations either zero in on managerial, organizational, and cultural
responses to technological change or focus on the ability of established
firms to deal with radically new technology; doing the latter requires
a very different set of skills from those that an established firm historically
has developed. Both approaches, useful in explaining why some companies
stumble in the face of technological change, are summarized below. The
primary purpose of this chapter, however, is to propose a third theory of
why good companies can fail, based upon the concept of a value network.
The value network concept seems to have much greater power than the
other two theories in explaining what we observed in the disk drive
industry.
ORGANIZATIONAL AND MANAGERIAL EXPLANATIONS
OF FAILURE
One explanation for why good companies fail points to organizational
impediments as the source of the problem. While many analyses of this
type stop with such simple rationales as bureaucracy, complacency, or
30 WHY GREAT COMPANIES CAN FAIL
‘‘risk-averse’’ culture, some remarkably insightful studies exist in this
tradition. Henderson and Clark,1 for example, conclude that companies’
organizational structures typically facilitate component-level innovations,
because most product development organizations consist of subgroups
that correspond to a product’s components. Such systems work very well
as long as the product’s fundamental architecture does not require change.
But, say the authors, when architectural technology change is required,
this type of structure impedes innovations that require people and groups
to communicate and work together in new ways.
This notion has considerable face validity. In one incident recounted in
Tracy Kidder’s Pulitzer Prize–winning narrative, The Soul ofa New Machine,
Data General engineers developing a next-generation minicomputer
intended to leapfrog the product position of Digital Equipment Corporation
were allowed by a friend of one team member into his facility in the
middle of the night to examine Digital’s latest computer, which his company
had just bought. When Tom West, Data General’s project leader
and a former long-time Digital employee, removed the cover of the DEC
minicomputer and examined its structure, he saw ‘‘Digital’s organization
chart in the design of the product.’’2
Because an organization’s structure and how its groups work together
may have been established to facilitate the design of its dominant product,
the direction of causality may ultimately reverse itself: The organization’s
structure and the way its groups learn to work together can then affect
the way it can and cannot design new products.
CAPABILITIES AND RADICAL TECHNOLOGY
AS AN EXPLANATION
In assessing blame for the failure of good companies, the distinction is
sometimes made between innovations requiring very different technological
capabilities, that is, so-called radical change, and those that build
upon well-practiced technological capabilities, often called incremental
innovations.3 The notion is that the magnitude of the technological change
relative to the companies’ capabilities will determine which firms triumph
after a technology invades an industry. Scholars who support this view
find that established firms tend to be good at improving what they have
long been good at doing, and that entrant firms seem better suited for
exploiting radically new technologies, often because they import the techValue
Networks and the Impetus to Innovate 31
nology into one industry from another, where they had already developed
and practiced it.
Clark, for example, has reasoned that companies build the technological
capabilities in a product such as an automobile hierarchically and experientially.
4 An organization’s historical choices about which technological
problems it would solve and which it would avoid determine the sorts of
skills and knowledge it accumulates. When optimal resolution of a product
or process performance problem demands a very different set of knowledge
than a firm has accumulated, it may very well stumble. The research of
Tushman, Anderson, and their associates supports Clark’s hypothesis.5
They found that firms failed when a technological change destroyed the
value of competencies previously cultivated and succeeded when new
technologies enhanced them.
The factors identified by these scholars undoubtedly affect the fortunes
of firms confronted with new technologies. Yet the disk drive industry
displays a series of anomalies accounted for by neither set of theories.
Industry leaders first introduced sustaining technologies of every sort,
including architectural and component innovations that rendered prior
competencies irrelevant and made massive investments in skills and assets
obsolete. Nevertheless, these same firms stumbled over technologically
straightforward but disruptive changes such as the 8-inch drive.
The history of the disk drive industry, indeed, gives a very different
meaning to what constitutes a radical innovation among leading, established
firms. As we saw, the nature of the technology involved (components
versus architecture and incremental versus radical), the magnitude of the
risk, and the time horizon over which the risks needed to be taken had
little relationship to the patterns of leadership and followership observed.
Rather, if their customers needed an innovation, the leading firms somehow
mustered the resources and wherewithal to develop and adopt it.
Conversely, if their customers did not want or need an innovation, these
firms found it impossible to commercialize even technologically simple
innovations.
VALUE NETWORKS AND NEW PERSPECTIVE
ON THE DRIVERS OF FAILURE
What, then, does account for the success and failure of entrant and established
firms? The following discussion synthesizes from the history of the
disk drive industry a new perspective on the relation between success or
32 WHY GREAT COMPANIES CAN FAIL
failure and changes in technology and market structure. The concept of the
value network—the context within which a firm identifies and responds to
customers’ needs, solves problems, procures input, reacts to competitors,
and strives for profit—is central to this synthesis.6Within a value network,
each firm’s competitive strategy, and particularly its past choices of markets,
determines its perceptions of the economic value of a new technology.
These perceptions, in turn, shape the rewards different firms expect to
obtain through pursuit of sustaining and disruptive innovations.7 In established
firms, expected rewards, in their turn, drive the allocation of resources
toward sustaining innovations and away from disruptive ones.
This pattern of resource allocation accounts for established firms’ consistent
leadership in the former and their dismal performance in the latter.
Value Networks Mirror Product Architecture
Companies are embedded in value networks because their products generally
are embedded, or nested hierarchically, as components within other
products and eventually within end systems of use.8 Consider a 1980svintage
management information system (MIS) for a large organization,
as illustrated in Figure 2.1. The architecture of the MIS ties together
various components—a mainframe computer; peripherals such as line
printers and tape and disk drives; software; a large, air-conditioned room
with cables running under a raised floor; and so on. At the next level,
the mainframe computer is itself an architected system, comprising such
components as a central processing unit, multi-chip packages and circuit
boards, RAM circuits, terminals, controllers, and disk drives. Telescoping
down still further, the disk drive is a system whose components include
a motor, actuator, spindle, disks, heads, and controller. In turn, the disk
itself can be analyzed as a system composed of an aluminum platter,
magnetic material, adhesives, abrasives, lubricants, and coatings.
Although the goods and services constituting such a system of use may
all be produced within a single, extensively integrated corporation such
as AT&T or IBM, most are tradable, especially in more mature markets.
This means that, while Figure 2.1 is drawn to describe the nested physical
architecture of a product system, it also implies the existence of a nested
network ofproducers and markets through which the components at each
level are made and sold to integrators at the next higher level in the
system. Firms that design and assemble disk drives, for example, such as
Quantum and Maxtor, procure read-write heads from firms specializing
Value Networks and the Impetus to Innovate 33
Figure 2.1 A Nested, or Telescoping, System of Product Architectures
Architecture of Disk
Platter
material
Magnetic
media
Adhesives
Application
process
Platter lapping
techniques
Architecture of
Disk Drive
Motor
Actuator
Servo
system
Recording
codes
Power
dissipation
Cache
Disks
Spindle
design
Physical size
and weight
Read-write
heads
Architecture of
Mainframe Computer
Cooling
system
Central
processing
unit
Terminals
Controllers
Back-up tape
storage
Operating
system
IC packaging
technology
Disk drive
Interface
technology
Random access
memory
Architecture of Management
Information System
Design of MIS
reports for
management
Line
printers
Card
readers
Proprietary
software
Protective
abrasives
Read-only
memory
Commercially purchased
software
Network design Data collection
systems
Service and
repair
requirements
Careers,
training, and
unique
language of
EDP staff
Configuration of
remote terminals
Physical environment–large, air-conditioned,
glass-front rooms with raised floors
Source: Reprinted from Research Policy 24, Clayton M. Christensen and Richard S. Rosenbloom, ‘‘Explaining the Attacker’s Advantage: Technological
Paradigms, Organizational Dynamics, and the Value Network,’’ 233–257, 1995 with kind permission of Elsevier Science—NL, Sara Burgerhartstraat 25, 1055
KV Amsterdam, The Netherlands.
34 WHY GREAT COMPANIES CAN FAIL
in the manufacture of those heads, and they buy disks from other firms
and spin motors, actuator motors, and integrated circuitry from still
others. At the next higher level, firms that design and assemble computers
may buy their integrated circuits, terminals, disk drives, IC packaging,
and power supplies from various firms that manufacture those particular
products. This nested commercial system is a value network.
Figure 2.2 illustrates three value networks for computing applications:
Reading top to bottom they are the value network for a corporate MIS
system-of-use, for portable personal computing products, and for
computer-automated design (CAD). Drawn only to convey the concept
of how networks are bounded and may differ from each other, these
depictions are not meant to represent complete structures.
Metrics ofV alue
The way value is measured differs across networks.9 In fact, the unique
rank-ordering of the importance of various product performance attributes
defines, in part, the boundaries of a value network. Examples in
Figure 2.2, listed to the right of the center column of component boxes,
show how each value network exhibits a very different rank-ordering of
important product attributes, even for the same product. In the top-most
value network, disk drive performance is measured in terms of capacity,
speed, and reliability, whereas in the portable computing value network,
the important performance attributes are ruggedness, low power consumption,
and small size. Consequently, parallel value networks, each
built around a different definition of what makes a product valuable, may
exist within the same broadly defined industry.
Although many components in different systems-of-use may carry the
same labels (for example, each network in Figure 2.2 involves read-write
heads, disk drives, RAM circuits, printers, software, and so on), the nature
of components used may be quite different. Generally, a set of competing
firms, each with its own value chain,10 is associated with each box in a
network diagram, and the firms supplying the products and services used
in each network often differ (as illustrated in Figure 2.2 by the firms listed
to the left of the center column of component boxes).
As firms gain experience within a given network, they are likely to
develop capabilities, organizational structures, and cultures tailored to
their value network’s distinctive requirements. Manufacturing volumes,
the slope of ramps to volume production, product development cycle
Figure 2.2 Examples of Three Value Networks
Metal-in-Gap
Ferrite Heads
Applied Magnetics
Cost
Availablity in high
unit volumes
Zenith
Toshiba
Dell
Connor
Quantum
Western Digital
Light and compact
Rugged
Easy to use
Ruggedness
Low power consumption
Low profile
Modems, etc.
Displays, etc.
AT/SCSI embedded
interface, etc.
Word processing and
spreadsheet software
CISC
microprocessor
2.5-inch Disk
Drives
Notebook
Computers
Portable Personal
Computing
Thin film
disks
Sun Microsystems
Hewlett-Packard
Maxtor
Micropolis
Read-Rite
Speed (MIPS)
Fits on desktop
Capacity
Speed
Size
Areal density
Simulation and
graphics software, etc.
Power supplies, etc.
ESDI embedded
interface, etc.
High-resolution
color monitors
RISC
microprocessor
Thin-Film
Heads
5.25-inch Disk
Drives
Engineering
Workstations
Computer-Automated
Design and
Manufacturing
Thin-film
disks
Corporate
Management
Information System
IBM
Amdahl
Unisys
StorageTech
Control Data
IBM
Capacity
Speed
Reliabilty
Recording
density
Accounting
software, etc.
Multi-chip IC
packaging, etc.
Actuators, etc.
Line printers
Central
processing unit
Read/Write Heads
(Captive supply)
Disk Drives
Mainframe
Computers
Particulate
oxide disks
Capacity
Speed
Reliabilty
Source: Reprinted from Research Policy 24, Clayton M. Christensen and Richard S. Rosenbloom,
‘‘Explaining the Attacker’s Advantage: Technological Paradigms, Organizational Dynamics, and
the Value Network,’’ 233–257, 1995 with kind permission of Elsevier Science—NL, Sara
Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.
36 WHY GREAT COMPANIES CAN FAIL
times, and organizational consensus identifying the customer and the
customer’s needs may differ substantially from one value network to the
next.
Given the data on the prices, attributes, and performance characteristics
of thousands of disk drive models sold between 1976 and 1989, we can
use a technique called hedonic regression analysis to identify how markets
valued individual attributes and how those attribute values changed over
time. Essentially, hedonicregression analysis expresses the total price of
a product as the sum of individual so-called shadow prices (some positive,
others negative) that the market places on each of the product’s characteristics.
Figure 2.3 shows some results of this analysis to illustrate how
different value networks can place very different values on a given
performance attribute. Customers in the mainframe computer value network
in 1988 were willing on average to pay $1.65 for an incremental
megabyte of capacity; but moving across the minicomputer, desktop, and
portable computing value networks, the shadow price of an incremental
megabyte of capacity declines to $1.50, $1.45, and $1.17, respectively.
Figure 2.3 Difference in the Valuation of Attributes Across Different Value Networks
Shadow Price (Dollars) of an Incremental
Megabyte of Capacity
Shadow Price (Dollars) of an Incremental Reduction of
One Cubic Inch in Size
1.40
0.20
1.30
1.10
–0.15 –0.10 –0.05 0.05 0.10 0.15
1.50
1.20
Mainframe computing
value network
1.60
1.70
Minicomputer
value network
Portable computing
value network
Desktop personal computer
value network
0.00 0.25
Value Networks and the Impetus to Innovate 37
Conversely, portable and desktop computing customers were willing to
pay a high price in 1988 for a cubic inch of size reduction, while customers
in the other networks placed no value on that attribute at all.11
Cost Structures and Value Networks
The definition of a value network goes beyond the attributes of the physical
product. For example, competing within the mainframe computer network
shown in Figure 2.2 entails a particular cost structure. Research, engineering,
and development costs are substantial. Manufacturing overheads
are high relative to direct costs because of low unit volumes and customized
product configurations. Selling directly to end users involves significant
sales force costs, and the field service network to support the
complicated machines represents a substantial ongoing expense. All these
costs must be incurred in order to provide the types of products and
services customers in this value network require. For these reasons, makers
of mainframe computers, and makers of the 14-inch disk drives sold to
them, historically needed gross profit margins of between 50 percent and
60 percent to cover the overhead cost structure inherent to the value
network in which they competed.
Competing in the portable computer value network, however, entails
a very different cost structure. These computer makers incur little expense
researching component technologies, preferring to build their machines
with proven component technologies procured from vendors. Manufacturing
involves assembling millions of standard products in low-laborcost
regions. Most sales are made through national retail chains or by
mail order. As a result, companies in this value network can be profitable
with gross margins of 15 percent to 20 percent. Hence, just as a value
network is characterized by a specific rank-ordering of product attributes
valued by customers, it is also characterized by a specific cost structure
required to provide the valued products and services.
Each value network’s unique cost structure is illustrated in Figure 2.4.
Gross margins typically obtained by manufacturers of 14-inch disk drives,
about 60 percent, are similar to those required by mainframe computer
makers: 56 percent. Likewise, the margins 8-inch drive makers earned
were similar to those earned by minicomputer companies (about 40 percent),
and the margins typical of the desktop value network, 25 percent,
also typified both the computer makers and their disk drive suppliers.
The cost structures characteristic of each value network can have a
38 WHY GREAT COMPANIES CAN FAIL
Figure 2.4 Characteristic Cost Structures of Different Value Networks
Characteristic Gross Margins (Percent)
Mainframe
Computing
Value Network
Desktop
OEMs
14-inch disk drive
makers
5.25-inch disk
drive makers
Mainframe
OEMs
Minicomputer
OEMs
8-inch disk
drive makers
34%
25-30%
41% 40%
56%
60%
20
10
0
30
40
50
60
70
80
90
100
Minicomputer
Value Network
Desktop PC
Value Network
Source: Data are from company annual reports and personal interviews with executives from several
representative companies in each network.
powerful effect on the sorts of innovations firms deem profitable. Essentially,
innovations that are valued within a firm’s value network, or in a
network where characteristic gross margins are higher, will be perceived
as profitable. Those technologies whose attributes make them valuable
only in networks with lower gross margins, on the other hand, will not
be viewed as profitable, and are unlikely to attract resources or managerial
interest. (We will explore the impact of each value network’s characteristic
cost structures upon the established firms’ mobility and fortunes more
fully in chapter 4.)
In sum, the attractiveness of a technological opportunity and the degree
Value Networks and the Impetus to Innovate 39
of difficulty a producer will encounter in exploiting it are determined by,
among other factors, the firm’s position in the relevant value network.
As we shall see, the manifest strength of established firms in sustaining
innovation and their weakness in disruptive innovation—and the opposite
manifest strengths and weaknesses of entrant firms—are consequences
not of differences in technological or organizational capabilities between
incumbent and entrant firms, but of their positions in the industry’s different
value networks.
TECHNOLOGY S-CURVES AND VALUE NETWORKS
The technology S-curve forms the centerpiece of thinking about technology
strategy. It suggests that the magnitude of a product’s performance improvement
in a given time period or due to a given amount of engineering
effort is likely to differ as technologies mature. The theory posits that in
the early stages of a technology, the rate of progress in performance
will be relatively slow. As the technology becomes better understood,
controlled, and diffused, the rate of technological improvement will accelerate.
12 But in its mature stages, the technology will asymptotically approach
a natural or physical limit such that ever greater periods of time
or inputs of engineering effort will be required to achieve improvements.
Figure 2.5 illustrates the resulting pattern.
Many scholars have asserted that the essence of strategic technology
management is to identify when the point of inflection on the present
technology’s S-curve has been passed, and to identify and develop whatever
successor technology rising from below will eventually supplant the present
approach. Hence, as depicted by the dotted curve in Figure 2.5, the
challenge is to successfully switch technologies at the point where S-curves
of old and new intersect. The inability to anticipate new technologies
threatening from below and to switch to them in a timely way has often
been cited as the cause of failure of established firms and as the source
of advantage for entrant or attacking firms.13
How do the concepts of S-curves and of value networks relate to each
other?14 The typical framework of intersecting S-curves illustrated in Figure
2.5 is a conceptualization of sustaining technological changes within
a single value network, where the vertical axis charts a single measure of
product performance (or a rank-ordering of attributes). Note its similarity
to Figure 1.4, which measured the sustaining impact of new recording
head technologies on the recording density of disk drives. Incremental
40 WHY GREAT COMPANIES CAN FAIL
Figure 2.5 The Conventional Technology S-Curve
Time or Engineering Effort
Product Performance
Third technology
Second technology
First technology
Source: Clayton M. Christensen, ‘‘Exploring the Limits of the Technology S-Curve. Part I: Component
Technologies,’’ Production and Operations Management 1, no. 4 (Fall 1992): 340. Reprinted by
permission.
improvements within each technology drove improvements along each of
the individual curves, while movement to new head technologies involved
a more radical leap. Recall that there was not a single example in the
history of technological innovation in the disk drive industry of an entrant
firm leading the industry or securing a viable market position with a
sustaining innovation. In every instance, the firms that anticipated the
eventual flattening of the current technology and that led in identifying,
developing, and implementing the new technology that sustained the overall
pace of progress were the leading practitioners of the prior technology.
These firms often incurred enormous financial risks, committing to new
technologies a decade or more in advance and wiping out substantial
bases of assets and skills. Yet despite these challenges, managers of the
industry’s established firms navigated the dotted line course shown in
Figure 2.5 with remarkable, consistent agility.
Value Networks and the Impetus to Innovate 41
A disruptive innovation, however, cannot be plotted in a figure such
as 2.5, because the vertical axis for a disruptive innovation, by definition,
must measure different attributes of performance than those relevant
in established value networks. Because a disruptive technology gets its
commercial start in emerging value networks before invading established
networks, an S-curve framework such as that in Figure 2.6 is needed to
describe it. Disruptive technologies emerge and progress on their own,
uniquely defined trajectories, in a home value network. If and when
they progress to the point that they can satisfy the level and nature of
performance demanded in another value network, the disruptive technology
can then invade it, knocking out the established technology and its
established practitioners, with stunning speed.
Figures 2.5 and 2.6 illustrate clearly the innovator’s dilemma that precipitates
the failure of leading firms. In disk drives (and in the other industries
covered later in this book), prescriptions such as increased investments
in R&D; longer investment and planning horizons; technology scanning,
forecasting, and mapping; as well as research consortia and joint ventures
are all relevant to the challenges posed by the sustaining innovations
Figure 2.6 Disruptive Technology S-Curve
Performance as Defined in Application "A"
Time or Engineering Effort
Application (Market) "A"
Technology 2
Technology 1
Technology 2
Application (Market) "B"
Performance as Defined in Application "B"
Source: Clayton M. Christensen, ‘‘Exploring the Limits of the Technology S-Curve. Part I: Component
Technologies,’’ Production and Operations Management 1, no. 4 (Fall 1992): 361. Reprinted by
permission.
42 WHY GREAT COMPANIES CAN FAIL
whose ideal pattern is depicted in Figure 2.5. Indeed, the evidence suggests
that many of the best established firms have applied these remedies and
that they can work when managed well in treating sustaining technologies.
But none of these solutions addresses the situation in Figure 2.6, because
it represents a threat of a fundamentally different nature.
MANAGERIAL DECISION MAKING AND DISRUPTIVE
TECHNOLOGICAL CHANGE
Competition within the value networks in which companies are embedded
defines in many ways how the firms can earn their money. The network
defines the customers’ problems to be addressed by the firm’s products
and services and how much can be paid for solving them. Competition
and customer demands in the value network in many ways shape the
firms’ cost structure, the firm size required to remain competitive, and
the necessary rate of growth. Thus, managerial decisions that make sense
for companies outside a value network may make no sense at all for those
within it, and vice versa.
We saw, in chapter 1, a stunningly consistent pattern of successful
implementation of sustaining innovations by established firms and their
failure to deal with disruptive ones. The pattern was consistent because
the managerial decisions that led to those outcomes made sense. Good
managers do what makes sense, and what makes sense is primarily shaped
by their value network.
This decision-making pattern, outlined in the six steps below, emerged
from my interviews with more than eighty managers who played key roles
in the disk drive industry’s leading firms, both incumbents and entrants,
at times when disruptive technologies had emerged. In these interviews I
tried to reconstruct, as accurately and from as many points of view as
possible, the forces that influenced these firms’ decision-making processes
regarding the development and commercialization of technologies either
relevant or irrelevant to the value networks in which the firms were at
the time embedded. My findings consistently showed that established
firms confronted with disruptive technology change did not have trouble
developing the requisite technology: Prototypes of the new drives had
often been developed before management was asked to make a decision.
Rather, disruptive projects stalled when it came to allocating scarce resources
among competing product and technology development proposals
(allocating resources between the two value networks shown at right and
Value Networks and the Impetus to Innovate 43
left in Figure 2.6, for example). Sustaining projects addressing the needs
of the firms’ most powerful customers (the new waves of technology
within the value network depicted in Figure 2.5) almost always preempted
resources from disruptive technologies with small markets and poorly
defined customer needs.
This characteristic pattern of decisions is summarized in the following
pages. Because the experience was so archetypical, the struggle of Seagate
Technology, the industry’s dominant maker of 5.25-inch drives, to successfully
commercialize the disruptive 3.5-inch drive is recounted in detail to
illustrate each of the steps in the pattern.15
Step 1: Disruptive Technologies Were First Developed
within Established Firms
Although entrants led in commercializing disruptive technologies, their
development was often the work of engineers at established firms, using
bootlegged resources. Rarely initiated by senior management, these architecturally
innovative designs almost always employed off-the-shelf components.
Thus, engineers at Seagate Technology, the leading 5.25-inch drive
maker, were, in 1985, the second in the industry to develop working
prototypes of 3.5-inch models. They made some eighty prototype models
before the issue of formal project approval was raised with senior management.
The same thing happened earlier at Control Data and Memorex, the
dominant 14-inch drive makers, where engineers had designed working 8-
inch drives internally, nearly two years before the product appeared in
the market.
Step 2: Marketing Personnel Then Sought Reactions
from Their Lead Customers
The engineers then showed their prototypes to marketing personnel, asking
whether a market for the smaller, less expensive (and lower performance)
drives existed. The marketing organization, using its habitual
procedure for testing the market appeal of new drives, showed the prototypes
to lead customers of the existing product line, asking them for an
evaluation.16 Thus, Seagate marketers tested the new 3.5-inch drives with
IBM’s PC Division and other makers of XT- and AT-class desktop personal
computers—even though the drives had significantly less capacity than
the mainstream desktop market demanded.
Not surprisingly, therefore, IBM showed no interest in Seagate’s disrup44
WHY GREAT COMPANIES CAN FAIL
tive 3.5-inch drives. IBM’s engineers and marketers were looking for 40
and 60MBdrives, and they already had a slot for 5.25-inch drives designed
into their computer; they needed new drives that would take them further
along their established performance trajectory. Finding little customer
interest, Seagate’s marketers drew up pessimisticsales forecasts. In addition,
because the products were simpler, with lower performance, forecast
profit margins were lower than those for higher performance products,
and Seagate’s financial analysts, therefore, joined their marketing colleagues
in opposing the disruptive program. Working from such input,
senior managers shelved the 3.5-inch drive—just as it was becoming firmly
established in the laptop market.
This was a complex decision, made in a context of competing proposals
to expend the same resources to develop new products that marketers
felt were critical to remaining competitive with current customers and
achieving aggressive growth and profit targets. ‘‘We needed a new model,’’
recalled a former Seagate manager, ‘‘which could become the next ST412
[a very successful product generating $300 million sales annually in the
desktop market that was near the end of its life cycle]. Our forecasts for
the 3.5-inch drive were under $50 million because the laptop market was
just emerging, and the 3.5-inch product just didn’t fit the bill.’’
Seagate managers made an explicit decision not to pursue the disruptive
technology. In other cases, managers did approve resources for pursuing
a disruptive product—but, in the day-to-day decisions about how time
and money would actually be allocated, engineers and marketers, acting
in the best interests of the company, consciously and unconsciously starved
the disruptive project of resources necessary for a timely launch.
When engineers at Control Data, the leading 14-inch drive maker, were
officially chartered to develop CDC’s initial 8-inch drives, its customers
were looking for an average of 300 MB per computer, whereas CDC’s
earliest 8-inch drives offered less than 60 MB. The 8-inch project was
given low priority, and engineers assigned to its development kept getting
pulled off to work on problems with 14-inch drives being designed for
more important customers. Similar problems plagued the belated launches
of Quantum’s and Micropolis’s 5.25-inch products.
Step 3: Established Firms Step Up the Pace ofSustaining
Technological Development
In response to the needs of current customers, the marketing managers
threw impetus behind alternative sustaining projects, such as incorporatValue
Networks and the Impetus to Innovate 45
ing better heads or developing new recording codes. These gave customers
what they wanted and could be targeted at large markets to generate
the necessary sales and profits for maintaining growth. Although often
involving greater development expense, such sustaining investments appeared
far less risky than investments in the disruptive technology: The
customers existed, and their needs were known.
Seagate’s decision to shelve the 3.5-inch drive in 1985 to 1986, for
example, seems starkly rational. Its view downmarket (in terms of the
disk drive trajectory map) was toward a small total market forecast for
1987 for 3.5-inch drives. Gross margins in that market were uncertain,
but manufacturing executives predicted that costs per megabyte for 3.5-
inch drives would be much higher than for 5.25-inch drives. Seagate’s
view upmarket was quite different. Volumes in 5.25-inch drives with
capacities of 60 to 100 MB were forecast to be $500 million by 1987.
Companies serving the 60 to 100 MB market were earning gross margins
of between 35 and 40 percent, whereas Seagate’s margins in its highvolume
20 MB drives were between 25 and 30 percent. It simply did not
make sense for Seagate to put its resources behind the 3.5-inch drive when
competing proposals to move upmarket by developing its ST251 line of
drives were also being actively evaluated.
After Seagate executives shelved the 3.5-inch project, the firm began
introducing new 5.25-inch models at a dramatically accelerated rate. In
1985, 1986, and 1987, the numbers of new models annually introduced
as a percentage of the total number of its models on the market in the
prior year were 57, 78, and 115 percent, respectively. And during the same
period, Seagate incorporated complex and sophisticated new component
technologies such as thin-film disks, voice-coil actuators,17 RLL codes,
and embedded SCSI interfaces. Clearly, the motivation in doing this was
to win the competitive wars against other established firms, which were
making similar improvements, rather than to prepare for an attack by
entrants from below.18
Step 4: New Companies Were Formed, and Markets for the
Disruptive Technologies Were Found by Trial and Error
New companies, usually including frustrated engineers from established
firms, were formed to exploit the disruptive product architecture. The
founders of the leading 3.5-inch drive maker, Conner Peripherals, were
disaffected employees from Seagate and Miniscribe, the two largest 5.25-
inch manufacturers. The founders of 8-inch drive maker Micropolis came
46 WHY GREAT COMPANIES CAN FAIL
from Pertec, a 14-inch drive manufacturer, and the founders of Shugart
and Quantum defected from Memorex.19
The start-ups, however, were as unsuccessful as their former employers
in attracting established computer makers to the disruptive architecture.
Consequently, they had to find new customers. The applications that
emerged in this very uncertain, probing process were the minicomputer,
the desktop personal computer, and the laptop computer. In retrospect,
these were obvious markets for hard drives, but at the time, their ultimate
size and significance were highly uncertain. Micropolis was founded before
the emergence of the desk-side minicomputer and word processor
markets in which its products came to be used. Seagate began when
personal computers were simple toys for hobbyists, two years before IBM
introduced its PC. And Conner Peripherals got its start before Compaq
knew the potential size of the portable computer market. The founders
of these firms sold their products without a clear marketing strategy—
essentially selling to whoever would buy. Out of what was largely a trialand-
error approach to the market, the ultimately dominant applications
for their products emerged.
Step 5: The Entrants Moved Upmarket
Once the start-ups had discovered an operating base in new markets, they
realized that, by adopting sustaining improvements in new component
technologies,20 they could increase the capacity of their drives at a faster
rate than their new market required. They blazed trajectories of 50 percent
annual improvement, fixing their sights on the large, established computer
markets immediately above them on the performance scale.
The established firms’ views downmarket and the entrant firms’ views
upmarket were asymmetrical. In contrast to the unattractive margins and
market size that established firms saw when eyeing the new, emerging
markets for simpler drives, the entrants saw the potential volumes and
margins in the upscale, high-performance markets above them as highly
attractive. Customers in these established markets eventually embraced
the new architectures they had rejected earlier, because once their needs for
capacity and speed were met, the new drives’ smaller size and architectural
simplicity made them cheaper, faster, and more reliable than the older
architectures. Thus, Seagate, which started in the desktop personal computer
market, subsequently invaded and came to dominate the minicomputer,
engineering workstation, and mainframe computer markets for disk
Value Networks and the Impetus to Innovate 47
drives. Seagate, in turn, was driven from the desktop personal computer
market for disk drives by Conner and Quantum, the pioneering manufacturers
of 3.5-inch drives.
Step 6: Established Firms Belatedly Jumped on the Bandwagon to
Defend Their Customer Base
When the smaller models began to invade established market segments,
the drive makers that had initially controlled those markets took their
prototypes off the shelf (where they had been put in Step 3) and introduced
them in order to defend their customer base in their own market. By this
time, of course, the new architecture had shed its disruptive character
and become fully performance-competitive with the larger drives in the
established markets. Although some established manufacturers were able
to defend their market positions through belated introduction of the new
architecture, many found that the entrant firms had developed insurmountable
advantages in manufacturing cost and design experience, and
they eventually withdrew from the market. The firms attacking from
value networks below brought with them cost structures set to achieve
profitability at lower gross margins. The attackers therefore were able to
price their products profitably, while the defending, established firms
experienced a severe price war.
For established manufacturers that did succeed in introducing the new
architectures, survival was the only reward. None ever won a significant
share of the new market; the new drives simply cannibalized sales of older
products to existing customers. Thus, as of 1991, almost none of Seagate’s
3.5-inch drives had been sold to portable/laptop manufacturers: Its 3.5-
inch customers still were desktop computer manufacturers, and many of
its 3.5-inch drives continued to be shipped with frames permitting them
to be mounted in XT- and AT-class computers designed to accommodate
5.25-inch drives.
Control Data, the 14-inch leader, never captured even a 1 percent share
of the minicomputer market. It introduced its 8-inch drives nearly three
years after the pioneering start-ups did, and nearly all of its drives were
sold to its existing mainframe customers. Miniscribe, Quantum, and Micropolis
all had the same cannibalistic experience when they belatedly
introduced disruptive technology drives. They failed to capture a significant
share of the new market, and at best succeeded in defending a portion
of their prior business.
48 WHY GREAT COMPANIES CAN FAIL
The popular slogan ‘‘stay close to your customers’’ appears not always
to be robust advice.21 One instead might expect customers to lead their
suppliers toward sustaining innovations and to provide no leadership—or
even to explicitly mislead—in instances of disruptive technology change.22
FLASH MEMORY AND THE VALUE NETWORK
The predictive power of the value network framework is currently being
tested with the emergence of flash memory: a solid-state semiconductor
memory technology that stores data on silicon memory chips. Flash differs
from conventional dynamic random access memory (DRAM) technology
in that the chip retains the data even when the power is off. Flash memory
is a disruptive technology. Flash chips consume less than 5 percent of the
power that a disk drive of equivalent capacity would consume, and because
they have no moving parts, they are far more rugged than disk memory.
Flash chips have disadvantages, of course. Depending on the amount of
memory, the cost per megabyte of flash can be between five and fifty
times greater than disk memory. And flash chips are not as robust for
writing: They can only be overwritten a few hundred thousand times
before wearing out, rather than a few million times for disk drives.
The initial applications for flash memory were in value networks quite
distant from computing; they were in devices such as cellular phones, heart
monitoring devices, modems, and industrial robots in which individually
packaged flash chips were embedded. Disk drives were too big, too fragile,
and used too much power to be used in these markets. By 1994, these
applications for individually packaged flash chips—‘‘socket flash’’ in industry
parlance—accounted for $1.3 billion in industry revenues, having
grown from nothing in 1987.
In the early 1990s, the flash makers produced a new product format,
called a flash card: credit card–sized devices on which multiple flash chips,
linked and governed by controller circuitry, were mounted. The chips on
flash cards were controlled by the same control circuitry, SCSI (Small
Computer Standard Interface, an acronym first used by Apple), as was
used in disk drives, meaning that in concept, a flash card could be used
like a disk drive for mass storage. The flash card market grew from $45
million in 1993 to $80 million in 1994, and forecasters were eyeing a
$230 million flash card market by 1996.
Will flash cards invade the disk drive makers’ core markets and supplant
magneticmemory? If they do, what will happen to the disk drive makers?
Value Networks and the Impetus to Innovate 49
Will they stay atop their markets, catching this new technological wave?
Or will they be driven out?
The Capabilities Viewpoint
Clark’s concept of technological hierarchies (see note 4) focuses on the
skills and technological understanding that a company accumulates as
the result of the product and process technology problems it has addressed
in the past. In evaluating the threat to the disk drive makers of flash
memory, someone using Clark’s framework, or the related findings of
Tushman and Anderson (see note 5), would focus on the extent to which
disk drive makers have historically developed expertise in integrated circuit
design and in the design and control of devices composed of multiple
integrated circuits. These frameworks would lead us to expect that drive
makers will stumble badly in their attempts to develop flash products if
they have limited expertise in these domains and will succeed if their
experience and expertise are deep.
On its surface, flash memory involves radically different electronics
technology than the core competence of disk drive makers (magnetics and
mechanics). But such firms as Quantum, Seagate, and Western Digital
have developed deep expertise in custom integrated circuit design through
embedding increasingly intelligent control circuitry and cache memory in
their drives. Consistent with the practice inmuch of the ASIC (applicationspecific
integrated circuit) industry, their controller chips are fabricated
by independent, third-party fabricators that own excess clean room semiconductor
processing capacity.
Each of today’s leading disk drive manufacturers got its start by designing
drives, procuring components from independent suppliers, assembling
them either in its own factories or by contract, and then selling them.
The flash card business is very similar. Flash card makers design the card
and procure the component flash chips; they design and have fabricated
an interface circuit, such as SCSI, to govern the drive’s interaction with
the computing device; they assemble them either in-house or by contract;
and they then market them.
In other words, flash memory actually builds upon important competencies
that many drive makers have developed. The capabilities viewpoint,
therefore, would lead us to expect that disk drive makers may not stumble
badly in bringing flash storage technology to the market. More specifically,
the viewpoint predicts that those firms with the deepest experience in IC
50 WHY GREAT COMPANIES CAN FAIL
design—Quantum, Seagate, and Western Digital—will bring flash products
to market quite readily. Others, which historically outsourced much
of their electronic circuit design, may face more of a struggle.
This has, indeed, been the case to date. Seagate entered the flash market
in 1993 via its purchase of a 25 percent equity stake in Sundisk Corporation.
Seagate and SunDisk together designed the chips and cards; the chips
were fabricated by Matsushita, and the cards were assembled by a Korean
manufacturer, Anam. Seagate itself marketed the cards. Quantum entered
with a different partner, Silicon Storage Technology, which designed the
chips that were then fabricated and assembled by contract.
The Organizational Structure Framework
Flash technology is what Henderson and Clark would call radical technology.
Its product architecture and fundamental technological concept are
novel compared to disk drives. The organizational structure viewpoint
would predict that, unless they created organizationally independent
groups to design flash products, established firms would stumble badly.
Seagate and Quantum did, indeed, rely on independent groups and did
develop competitive products.
The Technology S-Curve Framework
The technology S-curve is often used to predict whether an emerging
technology is likely to supplant an established one. The operative trigger
is the slope of the curve of the established technology. If the curve has
passed its point of inflection, so that its second derivative is negative (the
technology is improving at a decreasing rate), then a new technology may
emerge to supplant the established one. Figure 2.7 shows that the S-curve
for magnetic disk recording still has not hit its point of inflection: Not
only is the areal density improving, as of 1995, it was improving at an
increasing rate.
The S-curve framework would lead us to predict, therefore, that whether
or not established disk drive companies possess the capability to design
flash cards, flash memory will not pose a threat to them until the magnetic
memory S-curve has passed its point of inflection and the rate of improvement
in density begins to decline.
Insights from the Value Network Framework
The value network framework asserts that none of the foregoing frameworks
is a sufficient predictor of success. Specifically, even where estabValue
Networks and the Impetus to Innovate 51
Figure 2.7 Improvements in Areal Density of New Disk Drives (Densities in Millions
of Bits per Square Inch)
Areal Density
Engineering Effort
100
1,000,000
10
0.1
1 10 100 1,000 10,000 100,000
1,000
10
1970
1973
1977
1981
1985
1989
1995
Source: Data are from various issues of Disk/Trend Report.
lished firms did not possess the requisite technological skills to develop
a new technology, they would marshal the resources to develop or acquire
them if their customers demanded it. Furthermore, the value network
suggests that technology S-curves are useful predictors only with sustaining
technologies. Disruptive technologies generally improve at a parallel
pace with established ones—their trajectories do not intersect. The
S-curve framework, therefore, asks the wrong question when it is used
to assess disruptive technology. What matters instead is whether the disruptive
technology is improving from below along a trajectory that will
ultimately intersect with what the market needs.
The value network framework would assert that even though firms
such as Seagate and Quantum are able technologically to develop competitive
flash memory products, whether they invest the resources and managerial
energy to build strong market positions in the technology will depend
52 WHY GREAT COMPANIES CAN FAIL
on whether flash memory can be initially valued and deployed within the
value networks in which the firms make their money.
As of 1996, flash memory can only be used in value networks different
from those of the typical disk drive maker. This is illustrated in Figure
2.8, which plots the average megabytes of capacity of flash cards introduced
each year between 1992 and 1995, compared with the capacities of
2.5- and 1.8-inch drives and with the capacity demanded in the notebook
computer market. Even though they are rugged and consume little power,
Figure 2.8 Comparison of Disk Drive Memory Capacity to Flash Card
Memory Capacity
Year
Average Capacity (Megabytes)
Flash memory
1.8-inch drives
2.5-inch drives
Capacity demanded in notebook computers
92 93 94 95
Source: Data are from various issues of Disk/Trend Report.
Value Networks and the Impetus to Innovate 53
flash cards simply don’t yet pack the capacity to become the main mass
storage devices in notebook computers. And the price of the flash capacity
required to meet what the low end of the portable computing market
demands (about 350 MB in 1995) is too high: The cost of that much
flash capacity would be fifty times higher than comparable disk storage.23
The low power consumption and ruggedness of flash certainly have no
value and command no price premium on the desktop. There is, in other
words, no way to use flash today in the markets where firms such as
Quantum and Seagate make their money.
Hence, because flash cards are being used in markets completely different
from those Quantum and Seagate typically engage—palmtop computers,
electronic clipboards, cash registers, electronic cameras, and so
on—the value network framework would predict that firms similar to
Quantum and Seagate are not likely to build market-leading positions in
flash memory. This is not because the technology is too difficult or their
organizational structures impede effective development, but because their
resources will become absorbed in fighting for and defending larger chunks
of business in the mainstream disk drive value networks in which they
currently make their money.
Indeed, the marketing director for a leading flash card producer observed,
‘‘We’re finding that as hard disk drive manufacturers move up to
the gigabyte range, they are unable to be cost competitive at the lower
capacities. As a result, disk drive makers are pulling out of markets in
the 10 to 40 megabyte range and creating a vacuum into which flash can
move.’’24
The drive makers’ efforts to build flash card businesses have in fact
floundered. By 1995, neither Quantum nor Seagate had built market
shares of even 1 percent of the flash card market. Both companies subsequently
concluded that the opportunity in flash cards was not yet substantial
enough, and withdrew their products from the market the same year.
Seagate retained its minority stake in SunDisk (renamed SanDisk), however,
a strategy which, as we shall see, is an effective way to address
disruptive technology.
IMPLICATIONS OF THE VALUE NETWORK FRAMEWORK
FOR INNOVATION
Value networks strongly define and delimit what companies within them
can and cannot do. This chapter closes with five propositions about the
54 WHY GREAT COMPANIES CAN FAIL
nature of technological change and the problems successful incumbent
firms encounter, which the value network perspective highlights.
1. The context, or value network, in which a firm competes has a
profound influence on its ability to marshal and focus the necessary
resources and capabilities to overcome the technological and organizational
hurdles that impede innovation. The boundaries of a value network
are determined by a unique definition of product performance—a rankordering
of the importance of various performance attributes differing
markedly from that employed in other systems-of-use in a broadly defined
industry. Value networks are also defined by particular cost structures
inherent in addressing customers’ needs within the network.
2.Akey determinant of the probability of an innovative effort’s commercial
success is the degree to which it addresses the well-understood needs
of known actors within the value network. Incumbent firms are likely to
lead their industries in innovations of all sorts—architecture and components—
that address needs within their value network, regardless of intrinsic
technological character or difficulty. These are straightforward
innovations; their value and application are clear. Conversely, incumbent
firms are likely to lag in the development of technologies—even those in
which the technology involved is intrinsically simple—that only address
customers’ needs in emerging value networks. Disruptive innovations are
complex because their value and application are uncertain, according to
the criteria used by incumbent firms.
3. Established firms’ decisions to ignore technologies that do not address
their customers’ needs become fatal when two distinct trajectories interact.
The first defines the performance demanded over time within a given
value network, and the second traces the performance that technologists
are able to provide within a given technological paradigm. The trajectory
of performance improvement that technology is able to provide may have a
distinctly different slope from the trajectory of performance improvement
demanded in the system-of-use by downstream customers within any given
value network. When the slopes of these two trajectories are similar, we
expect the technology to remain relatively contained within its initial
value network. But when the slopes differ, new technologies that are
initially performance-competitive only within emerging or commercially
remote value networks may migrate into other networks, providing a
vehicle for innovators in new networks to attack established ones. When
such an attack occurs, it is because technological progress has diminished
the relevancce of differences in the rank-ordering of performance attributes
Value Networks and the Impetus to Innovate 55
across different value networks. For example, the disk drive attributes of
size and weight were far more important in the desktop computing value
network than they were in the mainframe and minicomputer value networks.
When technological progress in 5.25-inch drives enabled manufacturers
to satisfy the attribute prioritization in the mainframe and
minicomputer networks, which prized total capacity and high speed, as
well as that in the desktop network, the boundaries between the value
networks ceased to be barriers to entry for 5.25-inch drive makers.
4. Entrant firms have an attacker’s advantage over established firms in
those innovations—generally new product architectures involving little
new technology per se—that disrupt or redefine the level, rate, and direction
of progress in an established technological trajectory. This is so
because such technologies generate no value within the established network.
The only way established firms can lead in commercializing such
technologies is to enter the value network in which they create value. As
Richard Tedlow noted in his history of retailing in America (in which
supermarkets and discount retailing play the role of disruptive technologies),
‘‘the most formidable barrier the established firms faced is that they
did not want to do this.’’25
5. In these instances, although this ‘‘attacker’s advantage’’ is associated
with a disruptive technology change, the essence of the attacker’s advantage
is in the ease with which entrants, relative to incumbents, can identify
and make strategic commitments to attack and develop emerging market
applications, or value networks. At its core, therefore, the issue may be
the relative flexibility of successful established firms versus entrant firms
to change strategies and cost structures, not technologies.
These propositions provide new dimensions for analyzing technological
innovation. In addition to the required capabilities inherent in new technologies
and in the innovating organization, firms faced with disruptive
technologies must examine the implications of innovation for their relevant
value networks. The key considerations are whether the performance
attributes implicit in the innovation will be valued within the networks
already served by the innovator; whether other networks must be addressed
or new ones created in order to realize value for the innovation;
and whether market and technological trajectories may eventually intersect,
carrying technologies that do not address customers’ needs today to
squarely address their needs in the future.
These considerations apply not simply to firms grappling with the most
modern technologies, such as the fast-paced, complex advanced electronic,
56 WHY GREAT COMPANIES CAN FAIL
mechanical, and magnetics technologies covered in this chapter. Chapter
3 examines them in the context of a very different industry: earthmoving
equipment.