accueil/homeretour/backGérard Dréan

A review of
Broken Promises
An unconventional view of what went wrong at IBM

D Quinn Mills, G Bruce Friesen (Harvard Business School Press, 1996)

(published in Sociétal n°7, avril 1997)

The fall of IBM in 1991 remains an economic enigma. In almost 80 years of existence, this company had regularly known growth rates of 15% and profits after taxes higher than 10% of its revenue. In three years, the enterprise reputed the most invincible, the model of management, the scarecrow of the profession was going to post losses of 25 billion dollars, to lose 75 billion dollars of its market value and to terminate almost half of its 400000 employees. One spoke of a national disaster, and forecasts concerning its survival were most gloomy.

This reversal raises questions that exceed by far the particular case. If such a disaster can happen to IBM, who is protected ? Does it only result from local management errors, or are big firms systematically doomed  ? How to avoid a similar fate ? What are the mechanisms that decide the success or failure of an enterprise ? And now that IBM is back with profits and growth, and stock prices zooming up, can one anticipate that it will resume the dominant position that it had lost ?

Among the books that investigate that period, the one by D Quinn Mills and G Bruce Friesen aims higher than the anecdote by offering replies to those questions, while recognizing as soon as the prologue that their work is " one perspective among many ". One will see indeed that it is possible not to share totally their interpretation of events, nor the conclusions that they draw.

A cyclical history

But let us look first at the analysis. The authors recall rightfully that IBM has already known crises in its history, and has overcome them by being able to redefine itself. In the beginning of the fifties, IBM entered the computer market only thanks to the obstinacy of Tom Watson Jr, against the will of his father and of the rest of IBM executive management. Similarly, between 1961 and 1963, the same Tom Watson and his general manager Vince Learson had to fight to impose System/360 against general reluctance - and IBM almost succumbed in 1965 /1966 under the difficulties  of the project. At the end of the eighties, IBM was able to meet the challenge of minicomputers by splitting in order to adjust to different market dynamics, and later by accepting the cannibalization of its existing product lines by the 4300, that introduced a revolutionary level of price/performance. Finally, the launching of the PC in 1981 reproduced the pattern of the launching of computers 30 years earlier, with a pirate project decided by Frank Cary, then President, against the advice of the other executives, carefully isolated from the rest of IBM and respecting none of  its  rules. 

Mills and Friesen rightfully note the cyclical character of those events : a long phase of sleepiness resulting from success, from which IBM emerges under the personal impetus of one of its executives. Those internal organizational cycles of approximately 9 years, where a phase of bureaucracy follows a brief surge of innovation, are superimposed with the general business cycle of expansion and recession and with the technological cycle of integration and disintegration specific to the computer industry. Our authors recall the unceasing struggle of IBM leaders against the invasion of bureaucracy, quoting virulent memoranda by Watson in 1963, Learson in 1972, and John Opel in 1981.

What happened therefore in 1991 ? Was the capacity of management exceeded ? Did management techniques become obsolete ? Or did the enterprise choose a wrong strategic direction ? Most observers think that it is its conservatism that have almost killed IBM. Our authors think on the contrary that IBM has stumbled because it has moved away from its founding principles : singlenessand loyalty, which form what they call the  megastrategy . They logically conclude that if IBM goes back to those two principles, its future is bright. 

The singleness principle concerns the relationship with customers. It aims to make the company their only source of computer technologies. It assumes on the one hand listening attentively and permanently to customers, on the other hand a consistent and complete offer of products and services. It translated into a promise to customers : " we will always make all possible efforts to solve your problems ". The loyalty principle concerns the social contract between the enterprise and its employees. It aims to build a relationship of confidence in the long term, of which the full employment policy is the cornerstone. Only in this relationship of confidence will the employees in turn build long term cooperative relationships with customers, where IBM appears as a preferential service provider and not as a simple supplier of products. The loyalty principle is essential for the implementation of the singleness principle, insofar as " success in a service business depends on management’s ability to motivate those who deliver the service ".

Why did IBM lose sight of those two principles, and betrayed both its   promise to customers - excellence of service - and its promise to the employees - job security ?

An inexorable sequence

The " mother of all errors " was committed in 1981 under the presidency of John Opel : seeking to beat the Japanese by becoming the lowest cost producer, through the combination of high sales volumes and investments in mass production techniques. Errors have then inexorably followed one another. To finance its investments, IBM had to obtain cash rapidly by selling its equipment instead of renting it. The traditional sales relationship, centered on the long term and on the needs of customers, left place to one-shot sales governed by the needs of IBM. By putting pressure on salespeople to increase sales volumes, management pushed them to sell to their customers no longer what they needed, but what satisfied the immediate financial needs of IBM. From being cooperative, the commercial relationship has become contradictory. Having lost their preferred counselor, customers then turned without hesitation to less expensive competitors.     

The traditional relationship with customers being thus destroyed, IBM remained blind to the transformation  of the  market that followed the introduction of the new microcomputing technologies, and concentrated on the large centralized systems. At the same time, the periodic cycle of innovation revival against bureaucracy broke down, partly because of an access of timidity following the failure of the Future Systems project (FS) and partly because of antitrust pressures, two factors on which we will return, and a lot by the paralysis of the decision system by contention and consensus that had prevailed so far.

That decision system combined very centralized power with a collective elaboration process for major decisions. It encouraged competitive projects and brought to the attention of executives a large number of objections to decisions under study, that had to be resolved before the decision could be taken. Such a system identified a maximum number of potential problems in advance and prepared for a smooth execution, but delayed decision-making and encouraged conflicts, that could be overcome only by a strong singleness and loyalty culture.

Cut off from the judgment of the market and solicited by a multitude of problems, the system started to operate in a vacuum. IBM executive management managed an increasingly complex process without having enough time or skills to spend on the substance of the problems processed. The strategy was crushed by the process.

Then sales started to stagnate. While the strategic plan established in 1981 aimed explicitly at 100 billion dollars of revenue for 1990, actual sales actually reached only two-thirds of that objective. Meanwhile, IBM was going from approximately 300000 persons in 1981 to more than 400000 in 1986, without counting production capacities that had been  built  at a high expense. It became necessary to reduce those excess staffs, which was made particularly difficult by a drift of the full employment policy. The euphoria of years 1981-1985 had induced excessive tolerance towards the poor performers, and the full employment policy was no longer perceived as a commitment to offer a new job when a position would disappear, but  to  maintain the  job eternally. For example, in 1990, Armonk refused to authorize sales management to cut 15% of its staff. The social contract had become inconsistent with the market.

Finally, Wall Street entered the stage. The gradual profit erosion prevented IBM from obtaining the cash needed to undo what it had done wrongly, that is to say to close factories and to indemnify the personnel leaving, so that shareholders gained power. Now, even if firms have balanced obligations towards their shareholders, customers, suppliers and employees, Wall Street favors shareholders and the short term. As the fad of the period was to dismantle big enterprises into small specialized firms, the financial community advocated a splitting of IBM into 13 independent companies, which John Opel announced in the beginning of 1991. "A firm that survived seven decades by largely ignoring capital markets may meet its demise as a result of finally acceding to their demands ".

When that last action appeared slow in carrying its hoped-for fruits, John Akers was removed in the beginning of 1993 and replaced by Louis Gerstner, who hurried on the one hand to stop the implementation of that reorganization, on the other hand to formally announce the end of the full employment policy. A fortunate decision according to our authors, who assert that " the advice to break up IBM was dead wrong " and that "  Akers’ great failure was to try to solve a problem of basic strategy through an organizational route ".

Today, Louis Gerstner has returned IBM to its roots : singleness and listening to customers. And while in the beginning of the nineties, the three cycles - of the economy, of the industry and of the enterprise - were in their unfavorable phase, all three have now entered a rising phase. The period from 1991 to 1994 will therefore have only been a temporary stumble, which seems to be confirmed by the facts : losses have been entirely due to provisions for restructuring, indeed roughly underestimated originally, but IBM has never known operational losses. On the market, IBM has remained number one in seven segments out of nine, second and third in the other two, and  contrarily  to rumors has never suffered serious technological delay. The future appears potentially brilliant for IBM, if after re-establishing the bond with customers, it succeeds in reconstructing a social contract. 

Avoidable problems

From that history, Mills and Friesen conclude that IBM’s problems are due to management errors, most of which were avoidable. They admit indeed that " some of the causes were external ... given similar circumstances, any firm might have slipped ", but  they quote examples of other firms such as Hewlett Packard, packaged software firms and Intel, that have not slipped in the same manner when confronted with the same environment,. On the other hand, they refuse to play the prosecutors by noting that in an enterprise like IBM, strategic decisions only give their fruit - suave or bitter - after several years. Each chief executive pays for the mistakes of his predecessors and works for the benefit of his successors. " Akers’ predecessors had taken the steps that made him blind "

But this reassuring vision of history leads to statements that one hesitates to quote, so unsupported they seem : " had IBM’s executives aggressively backed the personal computer, ... , marketed an IBM version of the Intel microprocessor and had they quickly displaced Microsoft’s operating system, IBM’s share price would never have taken the dive it did ". Or still " if IBM had been left alone by the markets after 1990, it would have maintained its course ". 

The general lessons for the large corporations, currently under criticism, are in the same vein : " other large firms may hope that with the right choice of leadership they can remain successful ". Ten pieces of advice are offered, that, apart from the most original one " a Chief executive should not be part of the management team ", range from " never forget the customer " and " pay attention to the employee "  to " struggle unceasingly against bureaucracy " and " opportunities lost are the biggest consequence of a failure ". That being told, "  these lessons learned, large companies can survive and prosper ". In short, the opposite of the unconventional views announced in the subtitle, as if an otherwise penetrating analysis had stopped short and ended in the most trivial conclusions.

A different reading

Yet the penetrating descriptions of the inner workings of IBM seemed to lead to a different interpretation and to considerably more interesting conclusions. Yes, the historical pre-eminence of IBM is due to the exceptionally harmonious alliance of a singleness strategy with an organization, a management system, a social contract and a company culture consistent with that strategy. It is the fine-tuned adjustment of those elements that has made IBM a model, and that explains its exceptional success.

But in a complex system with that degree of integration, when one subsystem is no longer working, whether the organization or the decision system, it is likely that the whole system must be revised. If the organization that has allowed the success of a strategy breaks down, it may be a signal that the strategy is no longer viable. 

For example, in an enterprise of that size, the contention system cannot be divorced from a singleness strategy. The extreme case told by Mills and Friesen, where a president with a purely commercial background has to arbitrate a technological conflict between two vice-presidents with PhD in physics, is by necessity the daily bread and butter of the system. The contention system ensures that proposals for action do surface at executive level, and the managers, ignorant of the substance, then count on the system to reveal the weaknesses of competing positions, and on the ritual of meeting preparation for arguments to be expressed in a manner accessible to the non-specialists. Proper working of the process is a prerequisite to the quality of decisions made on substance. 

Many criticisms thus fall flat, even if they rely on accurate observations. Mills and Friesen are horrified that " IBM had no separate profit-loss  measure for some of its largest units ". They forget that in an enterprise as integrated as IBM, and committed to the singleness principle, profit cannot be expressed in a significant manner at any level other than the entire firm : each unit is only a specialized part that contributes to the overall profit and loss account in a specific and limited manner. Any local concept of "profit" could only be fallacious and misleading, and carry more perverse effects that motivation. 

Similarly when Mills and Friesen blame Akers for having " decentralized strategic decision-making  and centralized operating decisions instead of the opposite ". A strategy is only words (or dreams.. .) as long as it is not embodied in operational decisions. Among other objectives, the contention system aimed at confronting those responsible for making strategy decisions with the nitty-gritty of their implementation, both before and after the decision. Otherwise, strategies would be unrealistic or poorly executed. In real life, there is no discontinuity between strategic decisions and operational decisions.

Even the so much criticized bureaucracy has reasons for being. If the number of levels is high, it is because the number of subordinates at  each  level  must be low enough for managers to devote sufficient attention to each of them. In the social contract of IBM, local management was assumed to play the role held elsewhere by the Personnel staff, and even to some extent by union representatives. If the staffs are plethoric, it is to correctly prepare tradeoffs in the contention system. If assistant positions are numerous, it is also to train future executives. In short, such clumsiness of the system is part of the cost of the singleness and loyalty strategy.

As our authors underline, strategy, organization, management system and social contract form a whole. It follows that one cannot hope to solve problems of the magnitude under consideration by changing one element and by keeping the rest. If the system broke down, it is of course, at the surface, because it had become too complex ; but in depth, it is probably because the singleness strategy had exceeded its limits. It was mandatory first to « throw a fragmentation grenade » at the organization to enable each enterprise in the new federation to adopt the principles, strategy, organization, management system and social contract suitable for its activity and competitive environment. The answer to the problem of strategy was indeed in the organization, since the era of the single megastrategy was over.. 

The FS project revisited

Let us look again at the alternation of innovation and bureaucracy phases identified by Mills and Friesen. Phases of innovation appear in 1952, 1963, 1980 and 1990, that is at intervals of 11 years, 17 years and 10 years. Where is therefore the peak of innovation that should have appeared in 1974, consistent with the arrival of Frank Cary to the presidency in 1972 ? The missing cycle is easy to identify : it is the aborted FS episode that our authors blame for being the remote origin of the bureaucratic sleepiness of IBM.

Nevertheless, this phase of innovation is in its place in the calendar and attempted to solve real problems. I recall a presentation given at the very beginning of the Task Force convened by John Opel in 1971, describing what IBM would become if it accepted to fight competition on the basis of price and performance of each unit taken separately (the famous " box strategy "). The company would have to align its way of life on firms that had neither its activities of R and D and market development, nor its lifestyle in general ; it would have to gradually limit its marketing and planning activities, to reduce resources accordingly, and finally renounce its management system and social policy . Instead of that, the key objective of FS was to place the competitiveness of IBM in the coherence of the global system - the implementation of a singleness or "  systems " strategy.

The FS battle has never been fought on the market, and has been " lost " only by decision of IBM, that has shied away from the apocalyptic risks of the project. True, the context was hardly favorable to risk taking.. Still remembering the struggles to win the "5 billion dollar gamble" of 1964, IBM refused to take a 50 billion dollar gamble ten years later. In the meantime, the innovation in business practices introduced in 1969 had been the occasion of a major stumble in handling systems engineering assistance in the US, corrected just at the edge of disaster. Finally, during the entire decade of the seventies, the specter of ongoing trials maintained a panic fear of any action that could be interpreted as a step towards dominating the market. 

But the alternative is simple : either one thinks that IBM should have persevered, and then its renunciation to FS means that bureaucratic forces had definitively taken over in 1973, a time when internal opposition to the project began to score decisive points. Or, like most outside observers, one admits that IBM was right in stopping, and then one must also admit that, as early as the middle of the seventies, the mechanisms required for implementing singleness could no longer  work at the scale of the entire firm.

Other routes were possible. For example, Mills and Friesen rightfully remark that singleness does not mean " to be all things to all customers ", but " to be all things for some customers ". That strategy could therefore remain viable for limited markets, as shown in the same period by the success of mini vendors and of Cray, and even of IBM in the case of System/38 and its successor the AS/400, that were nothing else than FS limited to a single niche. IBM could have carried on the 1974 separation between GBG (small systems) and DPG (large systems), and split into units focused on one market segment each, where they would have implemented singleness strategies analogous to that of the S/38, but largely independent of each other.

But as is often the case, by renouncing the solution, IBM thought that it had gotten rid of the problem. It preferred to forgo any large global project and to manage its offer on a product by product basis, thus adopting the " box strategy ", the consequences of which had been announced in 1971 and eventually materialized. The strategic error of 1981 is indeed a consequence of abandoning FS in 1975, but the unavoidable collapse of profits could still be managed in two ways : retaining similar volumes but reducing costs and staffs, or attempting to maintain resources flat by vigorously increasing volumes. In retrospect, it is obvious that the chosen way, to increase volumes by increasing expenses, had no chance of  succeeding . But who could have admitted it at the time ? 

The aborted FS project played therefore a decisive role, but more profound than Mills and Friesen imagine. Its real message is that, as early as the middle of the 70s, the IBM system, including not only its organization and management methods, but also the principles embodied in that organization and in those methods in an indivisible way, had ceased to be adapted to the size of the enterprise and to the reality of the market. IBM leaders of that period confusedly realized that radical changes in the structure and rules of the game were under way in the computer industry, but failed to draw the consequences. Their successors, and with them our two authors, seem to have forgotten the lesson that John Akers had rediscovered.

Singleness in question

Crucial by the obsolescence of the singleness system, the years 1975 to 1985 were going moreover to see the emergence of a new fragmented industry structure. It is not an accidental coincidence that all firms built on the integrated model have known serious difficulties in that same period : IBM, but also Control Data, Unisys, Bull, followed by Prime, Digital Equipment and finally  Apple.  It is no more an accident that the most blatant success stories are now those of highly specialized firms like Microsoft or Intel, but also Dell or Cisco. The authors do mention that a fragmented industry is in the process of definitively superseding the integrated firms, but they fail to draw the obvious conclusions.

The rules of the game are no longer the same as in 1980. To fight Microsoft, one must operate like Microsoft. To struggle against Dell, one has to work like Dell. To struggle against Intel, one must function like Intel. As long as the whole industry was dominated by a few generalists, they imposed their global dynamics in each sector . Now that each sector is dominated by a majority of specialized firms, the specific dynamics of each sector imposes themselves on all players, including the surviving generalists.

The same unit can not operate like Microsoft and like Dell and like Intel, neither simultaneously nor alternately. If IBM wants to remain strongly present in all sectors, it must differentiate its organization and its systems and adapt  its  components to the diversity of local business dynamics.  Can one imagine that autonomous units operating in radically different manners can coexist within the same firm without being legally distinct ? The example of Hewlett Packard seems to indicate that the answer is yes, but such a situation nevertheless requires from executives a lot of self-discipline to leave the necessary autonomy to business units and to refrain from exercising their power. 

Our authors almost ignore that upheaval of the structure of the computer industry. Instead of that major cycle, they propose a technological cycle where the need for integration would now become preponderant thanks to the client/server architecture. But as much as their concept of an internal cycle of innovation and bureaucracy is convincing, as much their description of this so-called technological cycle is inconsistent, along with their statement that the success of Intel and Microsoft comes from the client/server architecture !

Where do they see that the market is massively demanding integrated offerings, other than probably in the self-justifying statements of IBM ? A certain number of customers, including many of the large traditional IBM customers, are wedded to that principle. But what are the respective sizes of the markets that prefer singleness as compared to those that prefer variety, low price and vendor independence ? And since one vendor cannot go after all of the above, one must admit that the singleness principle provides an advantage only in narrow niches, and probably shrinking ones at that.

A different future for a different IBM

Although I do believe, like Mills and Friesen, that the misfortunes of IBM are due to a crisis of the management system and in no way to a technological crisis, the rest of my conclusions are the opposite of theirs.

First, the crisis is structural and not circumstantial, because it expresses a mandatory mutation from one industry model to another. For that reason, the misfortunes of IBM were unavoidable - like those of Unisys yesterday, of Digital Equipment or Apple today, that have the same profound cause. IBM was in any event condemned to see its profits drop from 15% to 5% and its market share from 70% to 20%, because such is the new nature of competition. It could no more have avoided reducing its staffs to their current level, nor foregoing the luxurious attributes of its grandeur, for instance the full employment policy.

At the very most, it might not have compounded the problem through the strategic mistake of 1981, and then it might have managed more effectively what could be nothing else than an ordeal by definition. We all know now what Napoleon should have done to win at Waterloo. But in the storm, John Akers behaved as well as it was humanly possible to expect, and finally appeared the most perceptive and the most courageous of recent IBM leaders.     

John Akers renounced singleness, but attempted to preserve the essentials of the social contract until the bitter end. Upon his arrival, Gerstner hurried to denounce the social contract, by formally abandoning the full employment principle, while returning to the singleness principle. One has to believe that, even rusty, the IBM management system remains too impressive a machinery for a newcomer to destroy it happily before tasting its seduction. Gerstner probably did not want to deny himself the instruments of centralized power which Akers had understood that he had to give up. 

The press celebrates Gerstner as an activist supersalesman who spends most of his time visiting customers (see Business Week, December 1996). Without minimizing the virtue of example, that can not suffice very long to manage an enterprise of the size of IBM. The unavoidable ground work has only been deferred by the psychological effects of the arrival of a new management. Sooner or later, personal example will no longer be sufficient, and it will be necessary to put the whole enterprise in battle order to confront the competition of the 21stcentury. Despite his own declarations and despite the opinion of Mills and Friesen, Gerstner, or his successor, will sooner or later have to resume the organization works started by Akers.

New rules of the game

There are fewer differences concerning general lessons for the other firms, so difficult it is to be at odds with the ten lessons of the last chapter. From the IBM case, one could conclude that an integrated enterprise cannot efficiently implement a strategy of singleness when its size significantly exceeds 300000 employees, but that concerns rather few enterprises.

One can also derive from it a view of the real world mechanisms that prevent the existence of monopolies : " attaining the scale sufficient to dominate a market leads to a loss  of flexibility sufficient to lose competitiveness ". Those same mechanisms are at work in the process of dis-integration of an industry. As long as the dominant enterprise remains effective, its integrative dynamics prevail and specialized firms remain marginal. When, by growing, it ceases to function, specialized competitors are those that benefit most, more than similarly integrated competitors. The industry fragments itself into separate sectors, and in the end even the dominant firm has to split into autonomous enterprises, each operating in one sector.

Broken Promises is thus a mixture of penetrating analyses of the functioning of IBM, hasty statements about the market, thought-provoking judgments and debatable interpretations. Whether one agrees or not with the analyses and conclusions, this book triggers unlimited comments. If one judges the quality of a book by the thought processes that it starts, here is an excellent piece of work indeed!