The Innovator's Solution

by Clayton Christensen

  Modularity has a profound impact on industry structure because it enables independent, nonintegrated organizations to sell, buy, and assemble components and subsystems.8 Whereas in the interdependent world you had to make all of the key elements of the system in order to make any of them, in a modular world you can prosper by outsourcing or by supplying just one element. Ultimately, the specifications for modular interfaces will coalesce as industry standards. When that happens, companies can mix and match components from best-of-breed suppliers in order to respond conveniently to the specific needs of individual customers.

  As depicted in figure 5-1, these nonintegrated competitors disrupt the integrated leader. Although we have drawn this diagram in two dimensions for simplicity, technically speaking they are hybrid disruptors because they compete with a modified metric of performance on the vertical axis of the disruption diagram, in that they strive to deliver rapidly exactly what each customer needs. Yet, because their nonintegrated structure gives them lower overhead costs, they can profitably pick off low-end customers with discount prices.

  From Interdependent to Modular Design—and Back

  The progression from integration to modularization plays itself out over and over as products improve enough to overshoot customers’ requirements.9 When wave after wave of sequential disruptions sweep through an industry, this progression repeats itself within each wave. In the original mainframe value network of the computer industry, for example, IBM enjoyed unquestioned dominance in the first decade with its interdependent architectures and vertical integration. In 1964, however, it responded to cost, complexity, and time-to-market pressure by creating a more modular design starting with its System 360. Modularization forced IBM to back away from the frontier of functionality, shifting from the left to the right trajectory of performance improvement in figure 5-1. This created space at the high end for competitors such as Control Data and Cray Research, whose interdependent architectures continued to push the bleeding edge of what was possible.

  Opening its architecture was not a mistake for IBM: The economics of competition forced it to take these steps. Indeed, modularity reduced development and production costs and enabled IBM to custom-configure systems for each customer. This created a major new wave of growth in the industry. Another effect of modularization, however, was that nonintegrated companies could begin to compete effectively. A population of nonintegrated suppliers of plug-compatible components and subsystems such as disk drives, printers, and data input devices enjoyed lower overhead costs and began disrupting IBM en masse.10

  This cycle repeated itself when minicomputers began their new-market disruption of mainframes. Digital Equipment Corporation initially dominated that industry with its proprietary architecture when minicomputers really weren’t very good, because its hardware and operating system software were interdependently designed to maximize performance. As functionality subsequently approached adequacy, however, other competitors such as Data General, Wang Laboratories, and Prime Computer that were far less integrated but much faster to market began taking significant share.11 As happened in mainframes, the minicomputer market boomed because of the better and less-expensive products that this intensified competition created.

  The same sequence occurred in the personal computer wave of disruption. During the early years, Apple Computer—the most integrated company with a proprietary architecture—made by far the best desktop computers. They were easier to use and crashed much less often than computers of modular construction. Ultimately, when the functionality of desktop machines became good enough, IBM’s modular, open-standard architecture became dominant. Apple’s proprietary architecture, which in the not-good-enough circumstance was a competitive strength, became a competitive liability in the more-than-good-enough circumstance. Apple as a consequence was relegated to niche-player status as the growth explosion in personal computers was captured by the nonintegrated providers of modular machines.

  The same transition will have occurred before long in the next two waves of disruptive computer products—notebook computers and hand-held wireless devices. The companies that are most successful in the beginning are those with optimized, interdependent architectures. Companies whose strategy is prematurely modular will struggle to be performance-competitive during the early years when performance is the basis of competition. Later, architectures and industry structures will evolve toward openness and dis-integration.

  Figure 5-2 summarizes these transitions in the personal computer industry in a simplified way, showing how the proprietary systems and vertically integrated company that was strongest during the industry’s initial not-good-enough years gave way to a nonintegrated, horizontally stratified population of companies in its later years. It almost looks like the industry got pushed through a bologna slicer. The chart would look similar for each of the value networks in the industry. In each instance, the driver of modularization and dis-integration was not the passage of time or the “maturation” of the industry per se.12 What drives this process is this predictable causal sequence:

  The pace of technological improvement outstrips the ability of customers to utilize it, so that a product’s functionality and reliability that were not good enough at one point overshoot what customers can utilize at a later point.

  This forces companies to compete differently: The basis of competition changes. As customers become less and less willing to reward further improvements in functionality and reliability with premium prices, those suppliers that get better and better at conveniently giving customers exactly what they want when they need it are able to earn attractive margins.

  As competitive pressures force companies to be as fast and responsive as possible, they solve this problem by evolving the architecture of their products from being proprietary and interdependent toward being modular.

  Modularity enables the dis-integration of the industry. A population of nonintegrated firms can now outcompete the integrated firms that had dominated the industry. Whereas integration at one point was a competitive necessity, it later becomes a competitive disadvantage.13

  FIGURE 5 - 2

  The Transition from Vertical Integration to Horizontal Stratification in the Microprocessor-Based Computer Industry

  Figure 5-2 is simplified, in that the integrated business model did not disappear overnight—rather, it became less dominant as the trajectory of performance improvement passed through each tier of each market and the modular model gradually became more dominant.

  We emphasize that the circumstances of performance gaps and performance surpluses drive the viability of these strategies of architecture and integration. This means, of course, that if the circumstances change again, the strategic approach must also change. Indeed, after 1990 there has been some reintegration in the computer industry. We describe one factor that drives reintegration in the next section, and return to it in chapter 6.

  The Drivers of Reintegration

  Because the trajectory of technological improvement typically outstrips the ability of customers in any given tier of the market to utilize it, the general current flows from interdependent architectures and integrated companies toward modular architectures and nonintegrated companies. But remember, customers’ needs change too. Usually this happens at a relatively slower pace, as suggested by the dotted lines on the disruption diagram. On occasion there can be a discontinuous shift in the functionality that customers demand, essentially shifting the dotted line in figure 5-1 upward. This flips the industry back toward the left side of the diagram and resets the clock into an era in which integration once again is the source of competitive advantage.

  For example, in the early 1980s Apple Computer’s products employed a proprietary architecture involving extensive interdependence within the software and across the hardware–software interface. By the mid-1980s, however, a population of specialized firms such as WordPerfect and Lotus, whose products plugged into Microsoft’s DOS operating system through a well-defined interface,
had arisen to dethrone Apple’s dominance in software. Then in the early 1990s, the dotted lines of functionality that customers needed in PC software seemed to shift up as customers began demanding to transfer graphics and spreadsheet files into word processing documents, and so on. This created a performance gap, flipping the industry to the not-good-enough side of the world where fitting interdependent pieces of the system together became competitively critical again.

  In response, Microsoft interdependently knitted its Office suite of products (and later its Web browser) into its Windows operating system. This helped it stretch so much closer to what customers needed than could the population of focused firms that the nonintegrated software companies, including WordPerfect and Lotus’s 123 spreadsheet, vaporized very quickly. Microsoft’s dominance did not arise from monopolistic malfeasance. Rather, its integrated value chain under not-good-enough conditions enabled it to make products whose performance came closer to what customers needed than could nonintegrated competitors under those conditions.14

  Today, however, things may be poised to flip again. As computing becomes more Internet-centric, operating systems with modular architectures (such as Linux), and modular programming languages (such as Java) constitute hybrid disruptions relative to Microsoft. This modularity is enabling a population of specialized firms to begin making incursions into this industry.

  In a similar way, fifteen years ago in optical telecommunications the bandwidth available over a fiber was more than good enough for voice communication; as a consequence, the industry structure was horizontally stratified, not vertically integrated. Corning made the optical fiber, Siemens cabled it, and other companies made the multiplexers, the amplifiers, and so on. As the screams for more bandwidth intensified in the late 1990s, the dotted line in figure 5-1 shifted up, and the industry flipped into a not-good-enough situation. Corning found that it could not even design its next generation of fiber if it did not interdependently design the amplifier, for example. It had to integrate across this interface in order to compete, and it did so. Within a few years, there was more than enough bandwidth over a fiber, and the rationale for being vertically integrated disappeared again.

  The general rule is that companies will prosper when they are integrated across interfaces in the value chain where performance, however it is defined at that point, is not good enough relative to what customers require at the next stage of value addition. There are often several of these points in the complete value-added chain of an industry. This means that an industry will rarely be completely nonintegrated or integrated. Rather, the points at which integration and nonintegration are competitively important will predictably shift over time.15 We return to this notion in greater detail in chapter 6.

  Aligning Your Architecture Strategy to Your Circumstances

  In a modular world, supplying a component or assembling outsourced components are both appropriate “solutions.” In the interdependent world of inadequate functionality, attempting to provide one piece of the system doesn’t solve anybody’s problem. Knowing this, we can predict the failure or success of a growth business based on managers’ choices to compete with modular architectures when the circumstances mandate interdependence, and vice versa.

  Attempting to Grow a Nonintegrated Business

  When Functionality Isn’t Good Enough

  It’s tempting to think you can launch a new-growth business by providing one piece of a modular product’s value. Managers often see specialization as a less daunting path to entry than providing an entire system solution. It costs less and allows the entrant to focus on what it does best, leaving the rest of the solution to other partners in the ecosystem. This works in the circumstances in the lower-right portions of the disruption diagram. But when functionality and reliability are inadequate, the seemingly lower hurdle that partnering or outsourcing seems to present usually proves illusory, and causes many growth ventures to fail. Modularity often is not technologically or competitively possible during the early stages of many disruptions.

  To succeed with a nonintegrated, specialist strategy, you need to be certain you’re competing in a modular world. Three conditions must be met in order for a firm to procure something from a supplier or partner, or to sell it to a customer. First, both suppliers and customers need to know what to specify—which attributes of the component are crucial to the operation of the product system, and which are not. Second, they must be able to measure those attributes so that they can verify that the specifications have been met. Third, there cannot be any poorly understood or unpredictable interdependencies across the customer–supplier interface. The customer needs to understand how the subsystem will interact with the performance of other pieces of the system so that it can be used with predictable effect. These three conditions—specifiability, verifiability, and predictability—constitute an effective modular interface.

  When product performance is not good enough—when competition forces companies to use new technologies in nonstandard product architectures to stretch performance as far as possible—these three conditions often are not met. When there are complex, reciprocal, unpredictable interdependencies in the system, a single organization’s boundaries must span those interfaces. People cannot efficiently resolve interdependent problems while working at arm’s length across an organizational boundary.16

  Modular Failures in Interdependent Circumstances

  In 1996 the United States government passed legislation to stimulate competition in local telecommunication services. The law mandated that independent companies be allowed to sell services to residential and business customers and then to plug into the switching infrastructure of the incumbent telephone companies. In response, many nonintegrated competitive local exchange carriers (CLECs) such as Northpoint Communications attempted to offer high-speed DSL access to the Internet. Corporations and venture capitalists funneled billions of dollars into these companies.

  The vast majority of CLECs failed. This is because DSL service was in the interdependent realm of figure 5-1. There were too many subtle and unpredictable interdependencies between what the CLECs did when they installed service on a customer’s premises and what the telephone company had to do in response. It wasn’t necessarily the technical interface that was the problem. The architecture of the telephone companies’ billing system software, for example, was interdependent—making it very difficult to account and bill for the cost of a “plugged-in” CLEC customer. The fact that the telephone companies were integrated across these interdependent interfaces gave them a powerful advantage. They understood their own network and IT system architectures and could consequently deploy their offerings more quickly with fewer concerns about the unintended consequences of reconfiguring their own central office facilities.17

  Similarly, in the eagerly anticipated wireless-access-to-data-over-the-Internet industry, most European and North American competitors tried to enter as nonintegrated specialists, providing one element of the system. They relied prematurely on industry standards such as Wireless Applications Protocol (WAP) to define the interfaces between the handset device, the network, and the new content being developed. Companies within each link in the value chain were left to their own devices to determine how best to exploit the wireless Internet. Almost no revenues and billions in losses have resulted. The “partnering” theology that had become de rigueur among telecommunications investors and entrepreneurs who had watched Cisco succeed by partnering turned out to be misapplied in a different circumstance in which it couldn’t work—with tragic consequences.

  Appropriate Integration

  In contrast, Japan’s NTT DoCoMo and J-Phone have approached the new-market disruptive opportunity of the wireless Internet with far greater integration across stages of the value chain. These growth ventures already claim tens of millions of customers and billions in revenue.18 Although they do not own every upstream or downstream connection in the value chain, DoCoMo and J-Phone carefully manage the interfaces with their c
ontent providers and handset manufacturers. Their interdependent approach allows them to surmount the technological limitations of wireless data and to create user interfaces, a revenue model, and a billing infrastructure that make the customer experience as seamless as possible.19

  The DoCoMo and J-Phone networks comprise competing, proprietary systems. Isn’t this inefficient? Executives and investors indeed are often eager to hammer out the standards before they invest their money, to preempt wasteful duplication of competing standards and the possibility that a competitor’s approach might emerge as the industry’s standard. This works when functionality and reliability and the consequent competitive conditions permit it. But when they do not, then having competing proprietary systems is not wasteful.20 Far more is wasted when huge sums are spent on an architectural approach that does not fit the basis of competition. True, one system ultimately may define the standard, and those whose standards do not prevail may fall by the wayside after their initial success, or they may become niche players. Competition of this sort inspired Adam Smith and Charles Darwin to write their books.

  Parenthetically, we note that in some of its ventures abroad, such as in its partnership with AT&T Wireless in the United States, DoCoMo has followed its partners’ strategy of adopting industry standards with less vertical integration and has stumbled badly, just like its American and European counterparts. It’s not DoCoMo that makes the difference. It’s employing the right strategy in the right circumstances that makes the difference.