The Technology Industry and Clayton Christensen’s Innovator’s Dilemma
(Draft)
by Dave Shaw

Abstract
This paper argues that Clayton Christensen’s influential book The Innovator’s Dilemma misinterprets many of the experiences of the technology companies described in the book.  It argues that, properly understood, the experiences of many of these companies do not support, but rather sharply undermine the thesis of The Innovator’s Dilemma.  The paper further argues that Christensen badly misrepresents the critical role that market research can play in helping companies navigate technology shifts.

Introduction and Overview
It has now been over ten years since the publication of Clayton Christensen’s extraordinarily influential book The Innovator’s Dilemma.  The essence of this dilemma, as articulated by Christensen, is that by following “effective” management practices and by doing the right thing by listening to their customers, firms actually increase the likelihood that they will fail when confronted with “disruptive innovations.”  Christensen’s articulation of this innovator’s dilemma has sparked a tremendous amount of thinking by academics and practitioners.  In some cases, writers question the notion of an innovator’s dilemma itself; in others, the notion is accepted but writers raise different viewpoints regarding practical approaches for resolving the dilemma.

In the introduction to The Innovator’s Dilemma, Christensen nicely summarizes his thesis.  Referring to a list of firms that includes Digital Equipment, IBM, Apple, and Xerox, he writes: “The research reported in this book … shows that in the cases of well-managed firms such as those cited above, good management was the most powerful reason they failed to stay atop their industries.  Precisely because these firms listened to their customers, invested aggressively in new technologies that would provide their customers more and better products of the sort they wanted, and because they carefully studied market trends and systematically allocated investment capital to innovations that promised the best returns, they lost their positions of leadership (1997, xv).” 

Given that many writers have explored Christensen’s ideas, one might ask what I believe I can add to what is already a large body of work.  In short, the answer is that many of the examples Christensen uses are technology industry developments in the 1980s and 1990s, and, at least up to a point, I was there.  For example, Christensen writes about the difficulties faced by minicomputer companies when they tried to achieve success in the PC market.  I worked for a minicomputer company, HP, at a time when HP was trying to correct major mistakes from its initial entry into the PC market and gain a foothold in that market.  Christensen writes at great length about the efforts of Digital Equipment Corp. (DEC) to succeed in PCs, and as a competitive analyst for HP’s PC business, I monitored DEC’s efforts.  In addition, I had friends and colleagues who worked at other firms Christensen mentions, such as Apple.  (I do not, by the way, want to overstate my role at HP.  I was simply one of many working on the issue described above.) 

A key challenge that an academic like Christensen faces is to examine the experiences of many organizations and then draw useful generalizations about those experiences.  In part, Christensen is trying to help us the learn the lessons of business history.  He has made a great effort to do this in The Innovator’s Dilemma, and I applaud this effort.  However, in his effort to draw lessons from the experiences of dozens of companies, it is not clear how well he understands any one of them. 

The central argument of this paper is that Christensen largely misinterprets a number of developments in the technology industry in formulating his thesis.  The key point I wish to make is that, properly understood, many of the examples Christensen cites do not support his claim that firms failed because they did the “right thing” and listened to their customers.  In fact, a number of these firms failed for precisely the opposite reason: because they arrogantly refused to listen to their customers.  Furthermore, Christensen’s suggestion that traditional market research techniques are often harmful when confronting disruptive innovations is directly contradicted, not supported, by an accurate analysis of many of these examples.  I base my claim that Christensen largely misinterprets technology industry events on both information that was widely available to those who closely followed the industry, as well as information that was less widely available to the general public, but accessible to those of us who worked directly on some of these issues.

In concluding my analysis, I will briefly speculate on whether Christensen’s analysis is sufficiently flawed to significantly undermine his overall thesis of an “innovator’s dilemma.” 

Historical Analysis

The Experience of DEC
One of the primary examples Christensen utilizes is that of Digital Equipment Corporation, better known as DEC.  Before I discuss this example in detail, I would like to acknowledge DEC’s numerous contributions to the computer industry.  While I, like many, am critical of many decisions DEC made, one can never ignore the contributions of DEC or those of its co-founder, Ken Olsen (pictured below).

Having said this, Christensen’s suggestion that DEC failed because it followed the requests of its best customers is largely wrong.  DEC’s best customers were largely Fortune 1000 firms, and on the whole these firms were extremely vocal about their preference for non-proprietary computer systems based on (relatively) open standards such as Unix and DOS.  To cite just one example, Lou Gerstner has described how his intense frustration with proprietary systems as an IBM customer caused him to move IBM towards open standards when he later became IBM’s CEO (L. Gerstner, Who Says Elephants Can’t Dance, 2002, p. 125.)

Clearly this customer preference for open systems represented a challenge for DEC, which was generating tremendous profits from its proprietary VAX system.  Moreover, some DEC customers were certainly so “hooked” by their investments in VAX that their greatest near-term needs were enhancements to VAX, so there is a grain of truth to the suggestion that non-proprietary architectures were “disruptive innovations,” as Christensen defines the this term.  However, Christensen cannot deny that leading Fortune 1000 customers were vocal about their preference for non-proprietary systems.  Nor can he deny that DEC’s preference, regardless of what these customers wanted or asked for, was to sell them the VAX machines that commanded such high margins. 

In addition, the U.S. Federal Government, sometimes described as the largest purchaser of computing equipment in the world, had long insisted on non-proprietary systems, usually favoring UNIX.  There was no secret here; the Federal Government included open systems, such as UNIX, in its requirements lists for virtually all major purchases.  Many other governments in Europe and Asia did the same. 

There was, then, absolutely nothing mysterious about the customer preference for non-proprietary architectures, and the challenge this represented for DEC.  Christensen’s notion that even a superbly managed company would have had difficulty understanding or responding to this customer preference is flatly wrong.    For example, a 1989 article in the Wall St. Journal stated that “Although its VAX line has been enormously popular, fueling Digital’s explosive growth since 1985, more and more customers don’t want to get hooked on computers based on proprietary software.  ‘In the VAX prison, the food’s great, but you’re still locked in,’ says George Colony, a consultant at Forrester Research Inc.” (“At Digital Equipment, Slowdown Reflects Industry’s Big Changes,” by John R. Wilke, Wall St. Journal, 9/15/89)

Instead of listening to and responding to these requests, DEC’s management developed the reputation of having an arrogant, engineering-centric, and unresponsive mindset.  This mindset was, for many observers, perfectly typified in DEC CEO Ken Olsen’s description of UNIX as “snake oil,” and of the PC as a toy.  In fact, in their book on the rise and fall of DEC, Schein et. al. discuss how DEC’s engineering-centric culture led it to sometimes view customers with an attitude that bordered on contempt; some customers were viewed as simply too “dumb’ to appreciate the technical elegance of DEC’s products (E. Schein et. al., DEC is Dead, Long Live DEC, 2003, p. 42-43). 

Christensen claims that DEC failed in PCs because resources were taken from this disruptive innovation and diverted to other development efforts at DEC.  But he does not provide a shred of evidence to support this claim, and a much stronger argument can be made that DEC failed in PCs because its proud, engineering-centric culture, led by Olsen, refused to provided the “99%” IBM-compatible machines that customers clearly wanted.  Gordon Bell has written that, as late as 1987, Olsen sent him a DEC PC that “failed to run standard software, and even though its cabling was simple and elegant, it was ‘better’ but incompatible.”  (See “What Happened, a Postscript,” by G. Bell in Schein et. al, p. 299.)

On this issue, I can bring some personal insight to bear on the problem.  Christensen might reply that in the early 1980s, even the best managers could not have fully anticipated that IBM-compatible machines would win the day, and that embracing such a “commodity” business ran contrary to the views of leading management thinkers.  Such a reply might excuse the initial failure of DEC of introduce a 99% IBM-compatible machine in the early 1980s.  Indeed, HP made the same mistake; its first generation Touchscreen 150 PCs were “semi-IBM-compatible.”  However, the customers of the day stridently articulated their requirement that a PC in this market run software designed for IBM-compatibles as seamlessly as possible.  Customers were also clear that while value-added enhancements, such as a touchscreen or terminal keyboard features, might be welcome, this was only the case if such enhancements did not compromise IBM-compatibility.

HP eventually learned this lesson, not because HP employees were particularly insightful, but because most customers were persistent in not buying PCs that were not “99% compatible.”  After Compaq set the “99% compatible” standard, most PC (non-Apple) customers effectively stated “if your machine is not as IBM-compatible as Compaq’s and you are priced at rough parity with Compaq, I won’t even consider your product; I’ll just buy Compaq.”  Christensen cannot possibly attribute DEC’s extreme reluctance to meet this standard to good management or listening to customers.  It reflected precisely the opposite.

On this topic, Christensen’s argument that traditional market research techniques cannot help companies manage disruptive innovations is flatly wrong.  In the very early days of the PC market, one could argue that the typical user did not understand PCs well enough to articulate product preferences.  However, by the mid-80s and late-80s, user certainly could do so – and they did.  In many market research studies, a number of which were in the public domain, customers stated that IBM-compatibility provided far more value than proprietary enhancements.  This was Market Research 101, plain and simple.

The Experience of Wang Labs
Another example Christensen utilizes, albeit briefly, is that of Wang Laboratories.  Wang Labs is cited as one of five other “aggressively managed” minicomputer companies that also “missed the desktop personal computer market (1997, p. xiii).”  In the next paragraph, Christensen states that “many of these leading computer manufacturers were at one time regarded as among the best-managed companies in the world,” although, to be fair, he does not explicitly mention Wang Labs by name.

In fact, there is compelling evidence that Wang Labs, at the time it failed to navigate the shift to PCs, was a poorly-managed company that not only failed to respond to customer needs, but also failed to take even basic, commonsense measures.  Once again, honestly requires me to be critical of both individuals and an organization, but first I would like to acknowledge the contributions to the industry of An Wang, the brilliant engineer who founded Wang Labs and is pictured below.  Mr. Wang’s contributions should certainly not be minimized. 

However, one must also acknowledge that An Wang made a number of major mistakes that are not consistent with Christensen’s description of a well-managed firm.  An Wang made the enormous mistake of insisting that his son Fred be groomed to become CEO after he retired.  In order to accomplish this, he displaced the successful head of R&D to place the woefully-unprepared Fred in this position.  Moreover, An Wang himself cemented the foolish decision to introduce a product championed by Fred, the Wang Office Assistant, as a dedicated word processor (Charles Kenney, Riding the Runaway Horse: The Rise and Decline of Wang Laboratories, 1992, p. 170).

He did so in spite of the fact that IBM PC compatibility could have been added to the machine at a minimal cost, and An Wang’s motivation appears to have largely been a stubborn pride that refused to acknowledge the fact that most customers valued IBM compatibility, even over Wang’s word processing software.  Mr. Wang may have also been motivated by the fact that Wang’s proprietary systems had been extremely profitable in the past, and acknowledging that that customers preferred the relatively “open” IBM PC would have required him to accept the unpleasant truth that the salad days of proprietary minicomputer systems were coming to an end.

Wang Labs had also suffered, for many years, from the reputation of having an extremely poor, unresponsive customer service department.  While Wang Labs had originally provided excellent word processing solutions, the evidence suggests that the company had, over time, become increasingly indifferent to customer needs (Kenney, 1992).

All things considered, there is no need to turn to the notion of the innovator’s dilemma to understand the general failure of minicomputer firms to succeed in PCs.  Their customers clearly disliked the lock-in associated with proprietary systems, and the minicomputer firms generally ignored these customers because such firms could not shake off the seductive mirage of continuing to sell proprietary systems with 60% gross margins.  Even when the writing was on the wall, arguably in neon orange paint, some of these firms refused to read it.  This wasn't’t good management; it wasn’t even mediocre management.  It was bad management, and Christensen never offers any concrete evidence to dispute this obvious conclusion.

The Experience of Apple with Newton
Christensen also misinterprets Apple’s experience with the Newton PDA, introduced in 1993, to lend spurious support to his theoretical position.  Christensen argues that the Newton was “a market-creating, disruptive product targeted at an indefinable set of users whose needs were unknown to either themselves or Apple (p. 150).”  This is, to say the least, an overstatement; some of those user needs, like an effective data input system, were known to anybody in the industry who was paying the slightest bit of attention.

Since the introduction of the Compaq “luggable” machine in 1983, ten years before the Newton, it had been clear that PC users valued the ability to carry at least a subset of PC functionality around with them.  Calculator and electronics firms had offered some simple “organizer” capabilities in palmtops since the mid-80s, and HP had introduced the 95LX palmtop in 1991, with some success.  It had also long been clear that the “input system” of such a device had to work effectively, an easily-identifiable challenge since a highly portable device could not use a standard, full-size computer keyboard as its primary input system.  Newton’s handwriting recognition clearly failed to meet this very basic test.  Moreover, even though any reasonable form of quality testing would have easily identified the problems with Newton’s handwriting recognition, Apple made the decision to introduce the product anyhow.  This decision subjected the product to widespread ridicule – witness the Doonesbury cartoons mocking Newton’s poor handwriting recognition – and helped create the negative environment that led Apple to eventually give up on the product.

There was no justification for Apple’s poor decision-making on the key issue of Newton’s handwriting recognition and input system.  A strong case can be made that by combining sound management with good market research, Apple could have achieved success with the Newton. The challenges were less formidable than Christensen suggests.

Common Elements: A Powerful, “Hands-On” CEO

DEC’s PCs, Wang’s Office Assistant, and Apple’s Newton all shared a characteristic that Christensen does not sufficiently emphasize.  All of these products received a good deal of direct, personal attention from the CEO of the respective company.  Moreover, in the cases of DEC and Wang, the CEO was also a company founder.   While the situation with the Newton is less clear, the failure of DEC, Wang, and Apple to respond effectively to disruptive technologies raises the question of whether direct CEO involvement, especially when the CEO helped create the technology that is being displaced, is more harmful than helpful.  Perhaps these companies failed because, when it came down to a contest between clearly-articulated customer needs and CEO preferences, none of these companies were ultimately able to take a stand in favor of the customer.  If one accepts this view, then the failure of these companies to manage disruptive innovations was simply due to plain old poor management – nothing more.

Conclusion

In his introduction to The Innovator’s Dilemma, Christensen offers the key observation that a number of firms made enormous errors in judgment regarding disruptive technologies at the precise time when they being frequently cited as extremely well-managed firms.  

Christensen notes that this fact raises an obvious question: is it possible that these firms were widely viewed as being well-managed, but actually weren’t?  Perhaps they did not deserve their reputations; perhaps they had just been lucky?  Ultimately, he rejects this possibility in favor of the notion of the innovator’s dilemma, that “good management was the most powerful reason they failed to stay atop their industries (p. xv).”  In fact, his statement of his thesis includes the very strong claim that “these failed firms were about as well-run as one could expect a firm managed by mortals to be …(p. xv).”

I do not believe that Christensen’s argument holds water in the examples I have discussed above.  Instead, I believe these examples illustrate several management problems that have been discussed for years: 1) decision-making under conditions of uncertainty, particularly when the stakes are very high 2) the challenge that corporate founders and CEOs face in deciding when to “let go” and allow subordinates to make decisions they might not agree with, and 3) the need for strong boards to manage the role of founders and CEOs in public companies.  Again, these are issues that have been discussed at great length in the business literature for many years.  They are difficult issues, to be sure, but there is no need to resort to the concept of an “innovator’s dilemma” to understand the problems of DEC, Wang, and Apple that I have explored in this paper.

One might ask, then, whether I am prepared to argue that Christensen’s entire concept of an innovator’s dilemma is spurious, at least with respect to the technology industry.  At this point I am not, primarily because I have not yet analyzed in great detail one of the major technology examples Christensen uses, that of the hard disk drive industry.  I will note, however, that this industry was somewhat unique in that, for the time period Christensen examines, it was almost entirely an OEM industry.  Hard drive manufacturers sold the vast majority of their products to systems manufacturers, and some of these system manufacturers, e.g. IBM and DEC, could survive for significant periods of time without being fully responsive to end-user customer needs.  In other words, an end-user customer often had to accept a suboptimal hard drive solution if that was all his/her system manufacturer offered.  I believe the special characteristics of an OEM industry, in an overall environment dominated by a couple of firms, raise a unique set of issues that may not apply to management decision making in general.

In closing, then, I will note that while I have not yet seen a single example of Christensen’s that supports his theory of an innovator’s dilemma, I believe it would be premature to argue that the theory should be thrown out.  There is also the intriguing possibility that while Christensen’s underlying theory may be incorrect, some of his proposed management solutions are, nonetheless, often quite useful.  I will reserve the exploration of this topic for a later paper.