IBM

9042 IBM 6 - COUNTRIES AND THEIR DIVISIONS

Despite many changes, IBM's major organisational structures were regionally based until the 1990s when it was split into rather more 'autonomous' business centres. On the other hand, for many years, until the end of 1985, the picture was complicated by a marketing split between customer or product groupings - with large multinationals or large mainframes handled by one division and smaller companies or small mainframes handled by another division, and yet another handling the 'also‑ran' products. At the highest level - again until 1993 (when the separate sales splits ceased top be published) - it was represented by the rigid split between the two 'equals' of 'Domestic' (the US) and 'World Trade' (the rest of the world). In practice the latter mainly comprised the five major countries ‑ Japan, Germany, France Italy and the UK. The last time it was reported separately World Trade accounted for nearly 60% of IBM revenue (but, it being 1993, only 30% of its loss!). For most of its history, though, the ratio was the other way round; 'Domestic' was the larger of the two, and set the pace on all fronts.

This is another chapter where there is little to be said about Microsoft. Where IBM was the leading transnational, and one of the few organisations to justify the term, Microsoft has remained determinedly based on the domestic operation in the US; only exporting its 'products' to the rest of the world.

The Countries

Despite its more recent problems, IBM's handling of its various individual country organisations still offers many lessons for other multi‑nationals. For, although IBM manufactured and developed on a world-wide basis, it still managed to offer each country a remarkable degree of autonomy; and in the process avoided many of the worst criticisms levelled at other multinationals. Although each country's management was typically headed up by a board comprising almost exclusively nationals of that country with perhaps one token foreign director, it ultimately had to defer to Armonk. Because of the integrated nature of the world-wide manufacturing it also had little say in product or pricing strategy. Even so, in most respects it ran its own country operation without any great outside interference. Notwithstanding this autonomy, the IBM culture then was so strong that IBM offices and plants across the world looked (and indeed were) remarkably similar. For the record, Microsoft only has the equivalent of marketing operations outside of the US.

Generic Multi-Nationals…despite all the hype about globalisation, our research shows that multi-nationals do not recognise themselves as part of a generic ('multi-nationals') category of companies. Indeed, despite the popular view, they do not see themselves as sharing any significant common features with other 'multi-nationals' outside of their own industry. On the other hand, management theory, in general, and that relating to international business, in particular, sees multi-nationals as a quite separate category of organisation - with very different features. One wonders which is right, the companies themselves or the academics?

Hofstede…an interesting aside, in this context, is that one of the most famous - and well-respected - pieces of research on national differences in management used IBM as its subject. This work by Hofstede[1] demonstrated some fascinating differences between countries; not least that between the Northern Europeans and the Southern Europeans - and their descendants around the rest of the world. What it did not do, however, was compare IBM's culture with that of other organisations. Thus, although I am certain that the subtle differences he found did exist, the major difference - that of IBM's own culture versus the rest - was not highlighted by that research.

Laboratories and Plants

IBM's development laboratories are organised on a world-wide basis, though even then they still have to fund their operations from their royalties on previous products and report, for pay and rations, to their 'host' country management. Until recently IBM cosseted them far more than other groups, giving them resources and locations - such as the Côte D'Azur - for which any university professor would sell his soul (and sometimes did); and then it gave them at least as much freedom as any university. Bill Gates has followed much the same route at Microsoft; whose 'product' is even more dependent on development work. His many thousands of 'developers' are just as cosseted; though they are based in the US, and almost exclusively in the Seattle area. To an extent this practice has now changed in IBM, where it has learned the Japanese lesson of locating its key development teams near to its plants in order to speed up the process of transferring developments into production.

The output of these laboratories was a plethora of invention, and for a long time IBM had the great advantage of choosing which of a number of possible products best suited the market needs and often allowed the market itself to choose ‑ by launching competing products. The description 'laboratory', though, conveys the wrong image, for the essence of IBM development is a large software team working at desks in conventional office buildings. However, many of IBM's most successful products, particularly in the all important software area, came from individual work ‑ and it was then IBM's genius to be able to absorb these unsolicited efforts and then build on them.

Cosset (R&D) Employees…this is true of all employees, but those in R&D may be even more than most susceptible to a productive environment. They are also the ones who in general can be located anywhere; so their location can be chosen solely for its productive environment. In management theory, providing a pleasant environment is low down on the list of factors affecting location.

Even before its conversion to market forces in the early 1980s, IBM, as a matter of principle (like the Japanese), bought in all those components that it believed could be produced at least as well by outside suppliers; but then exacted draconian quality standards on these suppliers. Despite this philosophy IBM still had to make many of its components ‑ which is an indication of just how far ahead its technology normally was in those days. The main difference from other companies (including the Japanese) is still that caused by the integration of its manufacturing world-wide. A computer on the assembly line may contain parts from IBM plants in half a dozen countries, as well as from hundreds of suppliers. Tracking and co-ordinating these items is a major activity (particularly where IBM now works to a computerised variation on the Japanese ‑ Toyota ‑ 'Just in Time' kanban system). The 'factory floor' is, thus, dominated by quality control and administrative staff; with very few 'conventional' workers indeed.

Admin Staff…the general administrative staff has long been as much of a problem for IBM as for any other company; particularly where they staff functions which other companies will never even have heard of. IBM's only successful solution up until the mid-1980s had been to hold the numbers down by making physical office space available to them late and growing the rest of the company faster than the natural (almost cancerous) growth of these potential bureaucrats. The downturn in the mid-1980s, however, meant that the latter solution would not work - indeed the trends then worked very strongly against IBM's control of the booming numbers in its administration. A series of frantic measures were enacted; which progressively saw administration staff drafted in to the field - as sales personnel - (a strange, and uniquely IBM policy, placing its least effective employees in its most important role - sales - and then early retirement plans were made ever more attractive. Finally, IBM's much vaunted lifetime employment policies were abandoned; first covertly, by increased stringency in its Appraisal and Counselling (A&C) procedures and by psychological pressure on vulnerable employees, and then overtly as, for the first time in almost three quarters of a century, it made its employees redundant! As yet, due to its very rapid growth which has far outweighed that of the 'administration', Bill Gates has not had to face up to these problems - and seems able to treat 'admin' as a peripheral issue, as do many other organisations!

World Trade

With the advent of the young Watsons the world had to be split into two; the United States ('Domestic' in IBM parlance) as the jewel in the crown was given to the eldest, Tom Jr, while the rest of the world ('World Trade') went as a consolation prize to the younger son, Arthur. The two brothers thus ruled separately their own more or less equally balanced empires. The only change was that World Trade, which started as the junior partner then grew to overtake the US in terms of revenue; though one suspects that - as with many other US transnationals -  it was still the junior partner in all other respects! The principle was maintained through to the early 1990s, when its other troubles swamped adherence to such principles. Up until then, however, the two halves of the company were largely independ­ent; only development was on a global scale. Each had its own complete set of manufacturing facilities, and there was remarkably little export, or import, between them. This provided an insurance that IBM was not exposed to insurmountable problems with a single source of supply On any product, there was always the fall‑back of the sister plant(s) in the other half of the corporation. The PC group's damaging early problems with the hard disk for the PC AT were, for instance, compounded by the fact that, unusually for IBM, it was single sourced on one supplier world-wide.

IBM was one of the few genuine transnationals. Where most other multinationals also operate world-wide, and allow their country operating units a great deal of autonomy and indeed independence, they are constrained by national boundaries. IBM, though, is truly transnational in that it carries out its business across boundaries - without really recognising that they exist.

Transnational Rarity…despite the intensity of debate over globalization, there are in fact relatively few genuine transnationals; and the multinationals, on the other hand, rarely commit to globalization in the full sense. Management theory tends not to differentiate to any great extent between the two; and is still often wedded to the 'export' model of international trade.

EMEA et al…within World Trade there was a further split between two, once more autonomous, halves; EMEA (Europe Middle East Africa) which in volume terms mainly comprised Europe, and AFE (Americas Far East) which was dominated by Japan. The latter was later split into Asia Pacific (dominated by Japan) and 'Americas' (with Brazil being the largest country in this region - since Canada has always been considered to be part of US Domestic!). These splits were, however, not balanced; for EMEA - in revenue terms - was double the size of 'Asia Pacific' and four times that of the 'Americas'.

Despite the regional splits, in organisational terms the most important layer below Armonk, outside of the US, was the country. The 'political' positions of the World Trade headquarters at White Plains and the regional headquarters, EHQ (European HeadQuarters) Paris in the case of EMEA, always were somewhat ano­malous. For a number of decades EHQ, at least in theory, provided a major focus for planning. A large, and high powered, staff was supposed to bring a significant degree of 'local' European expertise to strategic planning. In practice EHQ, as seen from the larger countries (in IBM parlance the 'majors'; Germany, France, Italy, UK), was rarely more than a further layer of bureaucracy; simply transmitting messages between Armonk and its colonies. Its success, such as it was, lay in its minimal real interference with the country operations. The undermining of even the formal role of EHQ started in the 1970's with the formation of GBG (General Business Group), which in essence reported direct to White Plains, and often had its closest links with the US divisional headquarters in Atlanta. The official recognition of the demise of EHQ as the prime focus for regional planning had however to wait nearly a decade until 1984.

In Microsoft, was with most US corporations, there is no confusion. If it does not originate in Seattle, and indeed in Bill Gates' office, it is not worthwhile!

Transnational Working…in order to make organisations work effectively on a genuinely transnational scale, a great deal of effort and resource - supported by especially strong organisational structures - must be invested. This does not happen by accident, it has to be created by design; and it most certainly is not an incremental development of the export model - whereby the world is run from the home (head) office (as it is with Microsoft), or even the 'home plant'! Unfortunately, management theory often seems to expect transnationals to emerge naturally as another phase of growth for the large organisations, and does not make any allowance for the very different processes involved in managing a genuine transnational - it took IBM the best part of half a century to get its own act together.

The Divisions

For more than a decade, up until the end of 1982, there was another organizational split layered over that inspired by geography. This was the 'divisional' split between the main operating division, DPD (Data Processing Division), which was responsible for IBM's large mainframes (which accounted for the largest part of its revenue and, especially, of its profit) and GBG (which handled everything else from the small mainframes down to typewriters). In Microsoft the equivalent, and even more important, split also seems to have been between its main products; or at least between the teams working on their development.

This was, at least in part, in response to the threat posed by the emergence of DEC to dominate the 'mini‑computer' market at the end of the 1960s. The very successful commer­cial computer the S/360 had, unfortunately, overshadowed the earlier 1130, which was in its time IBM's equally successful scientific computer. As this progressively became long in the tooth, DEC took the opportunity to effectively steal this whole market‑place from IBM. Indeed, DEC later moved on to build a strong position in distributed computing, until this became much more important; and attractive to others! It ultimately failed to respond effectively to the new entrants, such as Sun Microsystems, and was swallowed up by Compaq; which had developed ambitions beyond the PC market - with near disastrous consequences, though, for both! In the late 1970s, though, DEC produced a range of innovative small computers, in particular the PDP 11, which threatened the lower end of IBM's main, business computing, market. IBM's response was the System/3 range of small commercial computers. IBM located marketing of these machines in a new division, GSD (General Systems Division), which subsequently merged with OPD (Office Products Division; in other words typewriters).

It is clear that, whatever the reasons for its creation, Armonk subsequently used GSD as a balance to the much larger DPD (Data Processing Division) and as a stimulus to new business. In practice GSD did not halt the growth of DEC, but it ensured that this growth was largely limited to the scientific market. GSD did outstrip DEC in overall sales, albeit in the business market; to become the second largest computer 'business' in the world.

Microsoft seems to have developed the other way around. Its initial Windows products were very successfully aimed at the small (personal computer) end of the market. As it moved up into the larger systems, and networking, it found the differences were so great - and competition so much stronger - that it had to give this work to an entirely separate development group. The result has been an incompatibility between the various strands of development which has bedeviled it as much as it did IBM. It regularly claims the two sides are converging but, again like IBM, never seems to quite complete the convergence!

Divisional Power…the strongest divisions typically have the strongest reasons for their existence. They are, in particular, an effective method of focusing resources on key objectives. On the other hand, artificial creations, which are imposed for administrative convenience, rarely prosper. Divisional structures may be debated at length in theory, especially in economic (transaction cost) theory, but usually with little regard to the practical rationale for their existence.

GBG

A more cynical interpretation, which was prevalent within IBM at the time, held that GSD/GBG was a creation designed to produce a sacrificial lamb; should the US government anti‑trust suits appear to be approaching any degree of success. In the event the sacrifice was never needed.  Circumstantial support for this view came from the fact that it was created in the US when the suits started, was expanded world-wide as they reached a critical stage and was formally dispensed with when it became clear that a suitably inoffensive outcome was in the offing! Tom Watson Jr - in his autobiography - admitted that there was such a plan! Microsoft has very clearly taken the opposite approach, deliberately making any attempt at splitting it up as difficult as possible; embedding all the elements in one gigantic offering (and maybe this, as much as customer demand, is the reason for its stress on convergence).

This IBM split remained largely unchanged for the best part of a decade; and set the underlying structure for most of the later versions. Even in 1981, when the two divisions were merged again, to form 'Information Systems Group', from the customers point of view the main change was in names, since the sales-forces were still split in much the same way. ISAM (Information Systems Account Marketing) was much the same as the previous DPD and ISM (Information Systems Marketing) looked very like the previous GBG (with the addition of some of the lower end mainframes). Over the next years, the divisional structures were changed with monotonous regularity. This process might have been a continuation of Frank Cary's philosophy of constant change, but increasingly it came to look more like the search - by an increasingly desperate senior management - for the philosopher's stone of effective management. Even so, while these may have changed the faces which Armonk saw on a regular basis, in general they had little effect on IBM as a whole.

The major exception was in 1986, when the reorganisation merged the two sales-forces and this resulted in the neglect of the lower end mainframes - and the almost complete loss of that market. Two years later, in 1988, they were de-merged; to something resembling the previous situation, albeit hidden amongst five independent business groups and no less than seven 'lines of business'.

In 1991 the marketing and service groups were reduced by one in number, to four, but the manufacturing and development businesses increased by two, to nine in number! By 1992, however, the problems IBM faced were so great that John Akers was desperately trying to break the organisation up into even more 'autonomous' units; described in the 1992 Annual Report as "…redefining IBM from a single, centralised company into a network of more competitive businesses." By then, though, the structure of IBM represented the least of its problems - and changing it was certainly was not the answer to a better future. John Akers, in his desperation, however, moved IBM ever closer to break-up; for instance hiving off  the PC group to a separate company. Some parts, such as the typewriters and small printers division (becoming Lexmark) and Federal Systems Division, were actually sold ; the former to a leveraged buy-out and the latter to Loral Corporation.

Lou Gerstner, when he took over in 1993, reversed the process of break-up; and set about consolidating the whole business; though this time in the image not of the previous IBM but of the more traditional multinational. In reality, with the important exception of the emergence of Global Services, the 'divisional' picture was almost as confused at the end of the decade - since it proved near impossible to undo this aspect of the trouble Akers had caused! The wheel had turned full circle (several times, as it so happened!), but had been broken in the process!

Structure Thrash…wise managers eschew the temptation to play with organisational structure. If there is a naturally strong structure already in existence leave it alone. If there isn't then try to evolve one. Fashions as to organisation structure seem to change faster than almost any other aspect of management theory. The reality is that such successful designs as there are emerge from the individual needs of the organisations involved - and now, increasingly, from the rather different demands of the new communications networks.

The Countries

Not all the countries were handled in the same way. The main split was between the 'majors' and 'minors'. The 'majors' were the largest markets, which represented the biggest part of IBM's business, and in turn they were the countries with 'clout' at Armonk. They used to be known as the 'product‑line' countries, because only they were allowed to maintain 'product‑line' departments interfacing with the development labs to input national requirements. Indeed EHQ's most important real function was to provide central support for the minors on a wide range of specialised activities (though, paradoxically, funded in the main, by 'taxes' on revenue, from the majors). These majors in Europe were the countries you would expect, Germany France Italy and the UK. Outside of Europe, the only single country which mattered was Japan - since Canada was part of US Domestic. In Microsoft every country, except the US, seems to be treated as a 'minor'!

Majors And Minors…even the largest transnationals depend upon relatively few countries, their 'majors', for the largest part of their business; and must necessarily treat the rest, their 'minors', as also-rans. Management theory in general, and even that relating to international business, pays little attention to this aspect of the 80:20 Rule.

Minors…although IBM's 'minors', in EMEA for instance, were self supporting in terms of their main business they did require extensive technical support for the more special­ised products. The result was a rash of specialist centres, ranging from  Milan giving technical support for the smaller mainframes to Munich providing that for production industries. These were funded by a revenue based charge across all of these; which resulted in them being predominantly funded by the majors - though they mainly supported the minors by typically by providing the range of expertise they could not individ­ually afford. As times became harder, however, IBM - which had once proudly boasted that it had offices in most countries throughout the world - retreated; and handed over its 'presence' in many smaller countries to agents (often just selling PCs). By the early 1990's, for example, it reportedly did not have a single office of its own throughout the whole of Africa and the Middle East. Contacting what, in the telephone directory purported to be IBM, you found yourself talking to an IBM agent (often the only one in the country).

In the case of the majors, and to a lesser extent the minors, IBM's aim was not just to market in each country but also to manufacture; and where possible to also locate a suitable research or international support facility. Even before it retreated, to hide behind agents, in the case of all the other smaller countries this was not the case. These countries, mainly in the third world, were only ever a target for marketing activities. IBM's market‑place is disproportionately sensitive to the level of development, so that like many other multi‑nationals the business generated by the third world is, at least in IBM's terms, virtually neglig­ible; and accordingly is neglected - though Lou Gerstner, in common with the leaders of many other multi-nationals has recently made China (with its vast population and associated market potential) a prime target.

Countries…as was stated earlier, the kingpin of the World Trade (non‑US) system was the individual country. Although a country was involved, at least in theory, in negotiations with Armonk leading to a mutually agreed plan, in practice the country was very much the junior partner. Thus, a country had no autonomy in the key areas of: the products it could sell and their prices; the headcount it could deploy; its sales targets; its capital and expense budgets; and funding for any significant individual projects. In fairness to IBM it needs to be recognised that it was forced to run globally integrated business operations to support its mainframe sales. Unlike Microsoft, for example, it had to manage a vast range of products, all of which needed to be to some extent compatible with each other, and especially with earlier products - something which Bill Gates chooses to (profitably) ignore, even though it poses major problems for his customers. For IBM, it would have been quite impossible to have even attempted to control this complexity on a decentralised basis. The position to a degree changed with the advent of the PC. Its performance was guaranteed, and standardised, by the suppliers of its components (who became the world-wide producers in this field) and it is now possible to produce a perfectly respectable PC on an small assembly line in the corner of a warehouse.

Within the somewhat draconian constraints imposed by Armonk the countries were autonomous. Each was typically run by a board of its own nationals and chose its own officers. It usually had no more than one foreign board member; traditionally the head of World Trade or the Regional Grouping (EMEA, say, who was, in any case, conventionally the token European on the central management committee at Armonk). It was staffed and manned virtually exclusively by its own nationals; as compared with the Japanese who have a penchant for introducing a significant number of their own key managers. There were indeed very few US citizens anywhere in World Trade to intrude on national aspirations.

In most respects IBM did indeed behave, in terms of its relations with its employees and the outside world, exactly as if it was a national company. Many of the remaining rules that apparently limited its autonomy were also imposed by a wish to avoid any chance of being accused of 'laundering' profits through the most tax advantageous countries, or tax havens, by rigging 'inter‑company billing prices' (ICBP's in IBM jargon; that is the transfer prices between countries); which is usually the greatest criticism lev­elled by critics of multi-nationals.

National Pride…despite globalisation, the national staffs of multi-nationals still owe allegiance to their own nation - so management must be sensitive to this issue, no matter what other (financial) considerations may suggest. The acid test for a transnational, indeed, may well be the number of head office nominees placed on local boards! This is an issue normally dealt with under political theory rather than management theory.

The Laboratories

There is inevitably a degree of separation between the purely national activities and those groups which are in effect internationally controlled, such as the laboratories and plants; both of which still are int­egrated on a global scale. For the latter the country largely then acted as a 'host' (an often used IBM term) providing little more than pay and rations for the troops, and was not deeply involved in policy or management. The UK was not untypical in 'hosting' a major development laboratory at Hursley near Winchester. This was the largest in World Trade with some 1,500 staff, essentially under non‑UK control. Interestingly even the laboratories were, though integrated as part of a world-wide development function, in effect also run as autonomous businesses. They 'bid' for new development work, and funded their activities from the royalties Armonk passed back from their previous successful work; Hursley in the UK reportedly lived very well for a time, on the royalties from its very successful developments in the area of disk technology; though it is not clear how its involvement in the later work on software for the PC - with its very public development problems -  was rewarded! Despite the much lower costs now available in India, for example, Microsoft's equivalent (software) laboratories are not just concentrated in the US but almost exclusively around Seattle - where they are under the direct supervision of Bill Gates!

As was mentioned above, IBM's laboratories which were not integrated into the national structure had been, until the 1990s, in the enviable position of choosing their location purely with a view to the attractiveness to their employees. Thus Winchester is indeed one of the most pleasant UK locations, though not as pleasant as that of La Gaude on the Riviera (surely the most sought after European 'assignment'; posting). The brave attempt at support for the UK development areas which initially placed the UK Scientific Centre in the northern wastes of Peterlee (near Newcastle) had to be reversed, and that too is now cosily settled in Winchester.
 

The Electronic Cottage…the rest of IBM was tied down, by the need to be near to the markets or production centres, in locations which were less ideal - though even then the facilities staff worked in were the best money could buy.

Telework…now, though, the emergence of electronic communications has meant that work is becoming distance independent; so companies can locate most of their workers where the environment is more pleasant - and can even consider deploying homeworking on the large scale. Such moves will be leveraged by the emerging skills shortages, compounded by the move of staff away from the search for purely financial gain, so that - as evidenced by the developments in working practices in Silicon Valley - the environmental ambience (the offices where they work) may become more important. The electronic cottage has long been hyped, but the physical reality - set by the existing housing stock - does not as yet allow for the extra home office space for efficient teleworking. On the other hand, the displacement of software production, though not yet much development, to the developing countries - especially India - is well under way, and is posing a threat to the highly paid staff in the West.

The term 'laboratory' gives a misleading impression of how the inmates now work. The buildings they occupy are essentially standard office buildings, whose inhabitants have mainly desk jobs. There are very few additions which would be recognisable as conventional laboratories to a visiting scientist or engineer; though the Zurich laboratory, which does more basic research, has more exotic facilities. Indeed their only real concession to any difference from the office work that dominates the rest of IBM is the availability of much greater raw computing power. Even the hardware is now largely developed by software generated simulations. In any case the major development work, even that relating to so called 'hardware', lies in software. It is accordingly teams of software specialists, programmers, and in particular testers, that sit at the desks.

Technophobia…our own research indicates that senior managers in most organisations - unlike those in IBM -  suffer from technophobia. They know little about modern R&D - and often tend to think of it in terms of the chemistry laboratory they last saw at school; and are frightened of dealing with it. R&D is a subject which now is - surprisingly in view of its importance - poorly covered by management theory.

Research Teams…IBM's teams may range from as few as 100 people spending 5 years developing a specialised industry terminal system, such as that for the drinks trade, to the 2,000 people that were - in the 1970s - reportedly eventually set to work to rescue the S/38. A figure, on average - across all developments - of 200 each has been suggested. On the other hand, it is arguable that, on mainframe projects, it needs to be so because the real problem (taking up more than two thirds of the time) is ensuring the resulting prog­rammes are 'bug' free; something which until recently Microsoft had not copied . Even its teams are now getting much bigger, with more than 500 developers working on each of the latest versions of Windows.

The innovative efficiency of such teams probably decreases expon­entially with size. This was certainly the claim in the mid 1980s, when Microsoft's teams were made up of 20-30 programmers where IBM's equivalents were at least ten times the size and the flexibility of the smaller teams has been suggested as one reason they were able to move faster and more effectively - though even their work still took almost a decade to come to fruition and the real problem was that the IBM teams were even worse at meeting deadlines! The individual on the other hand can still often make quantum leaps that are impossible for such teams. At the very beginning of the 1980s, when such things were science fiction, I watched amazed as one particularly 'way out', but brilliant, researcher at the UK Scientific Centre supplied me with the key elements of a totally new design for a 200 mips processor literally on the back of a piece of scrap paper, in less than half an hour; and this was more than a decade ahead of such speed appearing in the market place. Over a somewhat longer timescale I have had a systems engineer write a complete, workable, Bill of Material Processor for a mini‑computer in only 3 days. Scale of resource may not be all in res­earch; though it certainly helps in the absence of genius!

The success during the 1980s of its competitors, especially those in the PC marketplace (and, in particular, Microsoft) came from being able to use teams which were much smaller; though their products were typically much less sophisticated and often shipped with major bugs. But Microsoft, for instance, then prided itself on having a team (working jointly with IBM on OS/2 in the late 1980s) which was just 200 strong to  IBM's 1,000 - doing equivalent work. Even within IBM, small teams did produce much faster results. The original PC was produced in just over a year, by a team of just 13; and a decade later, at the beginning of the 1990's (after the message of small teams had at last percolated through the system!), IBM's first laptop was also produced in a year - by a team of less than 20. Even the very successful AS/400 range was started by a 5-man 'skunkworks' team.

Bureaucratic R&D…despite the popular image of the individual inventor, and although the initial ideas may be generated by a small group, the full application of modern R&D typically is the province of large teams which require considerable skill to manage. Management of R&D teams is rarely considered as a separate topic in theory.

Product Choice

It was the richness of the competing products that was, in its earlier days, a great source of IBM's strength. At that time it was likely that for each of the 10 new products produced each week perhaps 2 others were offered by competing teams within IBM. Ken Olsen, then CEO of DEC, is reported as having made the assertion[2] about IBM that "...every department in the company, every group in the company, feels the rest of the company is their first competitor." Whatever the reason, IBM was then in the enviable position of being able to choose the best product, the one that most closely met the market needs, from a range of alternatives. Indeed this process was often taken one stage further, for amongst the 10 products actually launched each week there would probably be some duplicates; and IBM let the market directly decide which was the best runner. This is also a very Japanese approach, where they too launch many new products into the market - to let consumers decide which will live and which will die. It is even more paradoxical that the main criticism levelled recently at IBM has been that it has been producing too few new products, too slowly! Once more, Microsoft's great (monopoly) strength comes from exactly the opposite approach. It deliberately integrates its new products into the monolithic existing ones. This might be risky, but where the new developments are largely incremental or have been 'bought off the shelf', they work reasonably well - and its customers' are used to encountering far more bugs than IBM's would have accepted!

Product Thrash…perhaps the most effective, though the most expensive, form of R&D is to test the market by full-scale launches of the real product - to see which succeeds in practice. However, with the move to e-commerce, and intangible knowledge products, the costs of such launches may be reducing significantly - so that it may often now be cheaper to go straight to a full launch (and possible then fail) than it is to test-market a product. This may increasingly become the mainstream model even for the West.

It is also paradoxical that this approach - of developing multiple products in competition with each other - led to one of the most damaging earlier criticisms of IBM; in Regis McKenna's 'Who's Afraid of Big Blue'. Where he did points to a downside effect of this process - the researchers are de-motivated when their products are not the chosen ones - "Time and again, they watch as their best ideas are watered down or thrown out entirely by the bureaucracy." On the other hand, in view of Tom Watson Jr's campaign to protect 'wild ducks', I think Regis was wrong when he went on to say "In many respects, IBM's size and mind-boggling bureaucracy hold its most creative employees as prisoners." It certainly did not hold back its Nobel prize winning physicists!

In earlier and happier times, as a result of this multiplicity of potential products as well as its tight grip on control of the de facto industry standards, which Bill Gates has now taken on, IBM easily maintained its product leadership, and market domination. This was then against the odds as described by Peter Drucker, and was in spite of even the Japanese experience ‑  where over time (admittedly decades) the individual corporations did (as Drucker predicted) lose their grip on the product leadership and their market domination slipped. This was my one criticism of the guru, but I guess he has had the last laugh, in view of what has now happened to IBM! What is not clear, however, is why this happened - when IBM had previously survived nearly three quarters of a century of continuous change. The evidence is that IBM did not decline in old age, as Drucker was perhaps implying, but developed a form of auto-immune disease; where the body destroys itself.

Immortality is Possible, but Not Easy…IBM's extended success, over three quarters of a century against all the supposed odds, shows that corporate immortality is not impossible - but its subsequent rapid decline shows how difficult it may be to manage this process for ever. Management theory ignores, and often denies, the possibility of corporate immortality.

Silicon Valley Myths

On the other hand, at the other end of the scale, the myth of the Silicon Valley start‑up should not be swallowed in its entirety. It is true that Steve Wozniak, of Apple, did design some very innovative touches; particularly, perhaps, the Apple II floppy disk controller. It is also true that the initial 50 Apple II units, sold to the original Byte Shop, were produced in a garage and funded by the sale of Steve Job's Volkswagen. Their meteoric success only came when A C Markula raised nearly a million dollars of venture capital to market the product properly, and it almost immediately tapped a vein of previously unsatisfied demand. It was certainly the right product in the right place at the right time; but was hardly a garage start‑up in the true sense. Even earlier David Packard and William Hewlett also started HP in a Silicon Valley garage, but in their case sponsored by their Professor (Fred Terman from Stamford University, who was to a large degree the creator of Silicon Valley). They themselves were Fellows of this richest of universities. Even with such impressive backing, after 2 years they were still only a 200 employee company with a $2 million revenue; their large scale expansion took decades. So much for the myths of instant stardom! So great has become the 'garage' myth that some recent start-ups have - despite receiving multi-million dollar backing - 'started' in their garage; just so they could say this was the case!

In the days when it had confidence in itself, such individual contributions were not barred from the IBM development systems, even though the main effort and resources were reserved for the larger teams. The most obvious large scale example of this, before the PC (which followed a rather different process), was the VM software that became one of IBM's two large main­frame operating systems It was first offered as an unsupported alternative from IBM's Cambridge, Massachusetts, laboratory.

The other likely stimulus is the panic of dire necessity. The need to get a product to market in a very short timescale, because of a sudden realis­ation of a potential exposure is traditionally a great spur. The whole PC range was developed in this way to meet the newly recognised market needs. Much earlier, IBM's mainline operating system, DOS, was a 'temporary' response to the planned implementation of OS simply being far too large to fit on the smaller machines.

Development Lead Times

Since it ran into problems in the mid 1980s, IBM has been much criticised for the time it takes to get products into the market. Some of this has been justifiable. The separation of R&D from manufacturing (though not from marketing staff) was unwarranted - and IBM has more recently been integrating its development teams with manufacturing, in recognition of the problem of shorter development times its competitors were deploying.

Development Time-Management…the ideas fashionable in the 1990s, promoted by George Stalk, revolving around the importance of reducing development times do still have relevance; to the fast-moving high-tech industries at least. Even though he has since backtracked on some of the details, Stalk's ideas are well reported by management theorists.

IBM's public presentation of its incremental R&D - on the PC at least - was poor to non-existent. Along with most other Western organisations its publicity focused on 'radical' developments', that is large (discontinuous) changes which radically changed the nature of the product. The 360, and to a less obvious extent the original PC, came into this category. The Japanese, followed now by an increasing number of Western companies, progress instead by many much smaller steps. This is much easier to speed up, since the base technology is already there, and only requires modifying. As a result the process is much more predictable and less risky. Ironically, despite its poor image, incremental development had been very much to the forefront of 'new product' launches on IBM's mainframes. Over their life, these were regularly enhanced with new features. Microsoft's success has also come from a steady stream of incremental changes to its software, albeit that - along with the rest of the software industry - it bundles them together in new 'versions' or 'levels'

Such increments have become a way of life in the computer field, especially in the PC domain, over recent years. It is best evidenced by Intel's steady progression through its chip range; starting with truly revolutionary 8-bit processors, through the 8086 which powered the first PC (and was revolutionary not in its basic technology but in its production and application) though to the 486 and Pentium and now to speeds beyond a gigahertz. These latter developments required incremental developments in production technology which, impressive though they were, can hardly be described as radical (or even risky); but the customers see them as major changes, despite the fact that neither they, or most of the software they use, can really make use of them. Thus, to a large extent, IBM was not truly overtaken by product developments but by the image which others managed to attach to their relatively small incremental steps. This might have been unfair, but it is what counted; and in not dealing with this image problem IBM did fail.

Incremental is Complementary to Radical…incremental (Japanese) development offers short-term advantage. Radical (Western) development offers long-term advantage. But they are not mutually exclusive, as the West has often seemed to think, rather they are complementary development processes; and the wise organisation will fund both. Outside the field of operations management, these two aspects of R & D are too often confused by theory, and by practice, with unfortunate effects on the related strategy of Western organisations.

Nobel Laureates

The problem in real terms was that almost the whole of  IBM's R&D eventually came to be focused on radical developments. So radical, indeed, that in the 1980s - when it was losing the 'time to launch' race - its scientists won two Nobel Prizes for Physics in consecutive years! A much more important problem has been that - apart from its developments in the components which are now the basis for its fast-growing OEM business - IBM has not delivered any of these radical breakthroughs to the market in recent years.

In the case of IBM, the impacts of this 'radical' (development) viewpoint, which is still common-place in the West, appeared in a number of areas:

Customer Risk…the radical new systems to which customers will be exposed will, by definition be new and untried; where those in incremental development will be already in use - and only minor changes to them will need to be made. Microsoft is, at least as seen by its own customer, in this enviable position which results in major barriers to entry for potential competitors. This means that the 'product assurance' guarantees must be that much effective.

Testing…critics of IBM development processes imply that this process only revolves around designing systems and writing the programs to make them work. Unfortunately that is the least of the problem - though it may be less problematic in the case of incremental work where much of  programming already exits and is stable. The bulk of the real work in IBM, though - at least until recently when it fell into line with IBM - not Microsoft usually comes in testing the programs to ensure that they are 'bug-free' under every possible circumstance.

Upward Compatibility…IBM had on-going commitment to all its customers that they would be able to use their existing investments (especially in software) on new products. This was one of the things which killed its 'Future Series' (making it compatible with 360 developed application programs was so complex that it crippled the whole system). In a different direction it bedevilled IBM's development of OS/2 on the PC. Customers of the 'second generation' PC, the AT (which used Intel's 286 chip), were promised that OS/2 would work on it. Unfortunately, the 286 was badly designed. On the other hand, the 386 which followed, and which then rapidly took over from the 286 as the workhorse (and was in turn succeeded by the 486, Pentium etc), rectified this problem and was ideally suited to OS/2. But IBM developers still kept on trying to meet the promise and deliver OS/2 on the 286; and they kept on failing (because it was an impossible task). This process continued long after the sophisticated users (who would have been the customers for OS/2) had moved onto the 386 (and they were denied the OS/2 system they needed - and went instead to Microsoft Windows!).

Customer Protection…the value of existing customers is such that their business should never be jeopardized by new 'product' paths which are incompatible with their existing investments or pose undue risks. On the other hand, what is important is not the reality of this but customers' perceptions of it - and IBM, whilst behaving impeccably, lost the argument in terms of those perceptions - whilst Microsoft (behaving rather less impeccably) is still winning the image battle! Compatibility and bugs are topics not discussed in the boardroom!

Systems Sign-Off…because of the risk to the customer, and also because of the risk to IBM, it had a number of very sensible processes to minimize this risk. Unfortunately they took time. They revolved around a number of key documents, of which the most important at the stage of moving from development into production was 'Forecast Assumptions'. As the title implies, this was supposedly about the other parts of IBM (especially the marketing people - in the form of the new-product staff function called 'Product Line') commenting on the sales forecasts which had been derived from a number of assumptions. The document was, though, also a major test of the assumptions; and hence of the viability of the product design. As such, the debate over this document often rolled backwards and forwards around the world, as well as within the US, until everyone was happy (and 'signed-off' the product). This typically took 6 months or more.

The problem was in getting people or 'sign-off' (that is approve) the document (in exactly the same way that Japanese management send round a 'ringi' document for managers in their organisation to sign-off). Indeed, the 250 signatures reported, by Regis McKenna (in 'Who's Afraid of Big Blue'), and which my own personal experience of the process would indicate probably could have happened on the more important product launches, might seem excessive. This criticism was voiced even more strongly in the 1990s - especially in Paul Carroll's book ('Big Blues') which developed it as a main reason for IBM's decline.

On the other hand, this process provided an excellent safeguard against risk, and had previously cost little in time when the overall development cycle was 5 years or more, as it was in the 1970s, and had only fallen to 3 years by the early 1980s. By the late 1980s, however, the cycle time had fallen to under a year (and often much less); when the process did cause

delays!

Security…because of the highly confidential nature of the development work, because the anti-trust cases had made it almost criminal to let anyone 'pre-announce' a product before it was ready for delivery, all these communications were carried out under a cloak of secrecy which would have put the CIA to shame. The key documents (called 'candy-stripes' because their covers had these motifs so that they were instantly recognizable and issued only to a registered individual. I caused a major investigation when I handed some candy-stripes directly to a manager taking over my duties who, having finished them, conscientiously shredded them so they were lost to the system!

On a slightly more humorous note, at one stage I was running a small development team and, as was always the case in IBM, had to give this product a code name. We gave it a name which matched the theme of the series and one which was easy for all of us to remember. Regrettably, after a few weeks we all forgot it! The result was that the main development laboratory in the US refused to talk to us - even though we were all personally known to the staff there. It was a month later, when one of my team miraculously remembered the code name, before we could talk to them again!

Despite everything, however, IBM could still introduce major new developments. Its AS/400 product in the 'mini' range was a world beater in the mid 1980's - and even through to the 1990s; and it was developed in less than two years - despite the fact that it had to be launched in 27 different languages and with 2,500 application programmes available. Even though it lost the workstation market to Sun, for half a decade or so, IBM's own RISC based RS/6000 workstation was eventually launched in 1992. It too was a world beater; but Sun still - just - managed to hang onto leadership of the market, and the delay meant that competitors (such as Hewlett Packard) were able to launch against it within less than a year.

Stable Products…even though there can be rapid product churn in the new e-commerce markets, with product life sometimes measured in months before an incremental improvement outdates it, the stability of the underlying core is of paramount importance. This is illustrated by the problems Microsoft has visited on its customers in forcing them to move up to incompatible new offerings - something the old IBM would never have done! The problems of overcoming 'switching costs', the cost to users of switching to other suppliers' incompatible offerings, has long been recognised in the context of competition theory. It has not, though, been explored as a function of the continuing offerings from the same supplier, nor in terms of the stability of such products. Perhaps the assumption is that suppliers should, as IBM did, look after the interests of their customers; as Microsoft for instance too often does not, but still profitably holds onto its customers whose switching costs are much greater than those of staying with it!

Manufacturing

This was, apart from the PC, also integrated on an international scale; and it too followed the same broad reporting line as the laboratories. In a similar manner, as each new production requirement or 'mission' in the jargon, entered the plan it was put out to competitive tender to all qualifying IBM plants. The most attractive (that is the lowest price) tender won.

This became the time when IBM conducted its most important negotiations with host governments. This wheeling and dealing was, not unexpectedly, never made public. It was, however, a matter of public record that immediately prior to taking the decision to locate the European PC manufacturing plant at Greenock near Glasgow the whole IBM board met in London, and met Margaret Thatcher in Downing Street.

Although the prestigious PC plant was located in the UK it probably resulted in remarkably few new jobs, a few hundred at most. The PC was mainly an assembly of components bought world-wide, from the most suitable (cheapest) sources. Even assembly of the PC was then heavily automated, initially with extensive use of IBM's own robots; so even within the plant the numbers of additional personnel were limited (though these were increased when IBM found that, to its surprise, human beings were more efficient than robots!). The gross export figures for PC's, to help the UK balance of trade, no doubt looked impressive, probably in the billions of pounds sterling. The net figures, balanced off against the import of the various components, were probably less than half the gross.

Automation…at the end of the day, when all the political debates had been resolved, IBM built a plant which was as automated as was humanly, or at least robotically, possible; in the belief that this would ensure it maintained its position as the lowest cost supplier. After the late 1970's, when this policy was first enunciated, tens of billions of dollars were expend­ed on creating the most technologically advanced factories in the world. In addition the product itself represented leading edge technology of the very highest degree; so the combined result was a level of manufacturing sophist­ication that was several orders of magnitude in advance of virtually all other manufacturers. IBM once more hoped that this investment would make its future position unassailable, since no other company could afford to even contemplate the vast sums needed; it was the one classic example of IBM attempting to use its massive size to guarantee its future (and spectacularly failing, as we will see).

The potential payoff was supposedly shown by the IBM Lexington, Kentucky, plant where some $350 million was spent on automating a 25 year old plant. Before automation labour accounted for one third of manufacturing costs. The target for the automation was to cut this to only 5%; though, needless to say, IBM at that time spent some $5 million retraining the displaced workers rather than making them redundant! The cost was, however, not negligible. A strong rumour circulated in IBM to the effect that the first fully automatic factory never worked to specification, and as a result IBM lost a billion dollars. But as always with the IBM of that time this was affordable from 'petty cash', without even showing as a blip on the profits curve in the final accounts. The invaluable lessons learnt were supposed to have been incorporated in the later plants; and as a free spin‑off the technol­ogy for IBM's Robotics Group emerged. The real test of viability, though, came later - when IBM in effect abandoned its hopes for the plant; and it became the first of the management buy-outs, becoming a new company (funded by a leveraged buy-out), Lexmark

Robotics…whilst this approach has threatened on the one hand to create redundancies and on the other to vastly improve productivity, the reality has been that it has had relatively little impact outside of a few specialist areas such as automobile assembly. With improvements in global transport it is easier, and cheaper to source the 'labor' overseas. As part of production management, this is in any case an argument largely lost on management theorists!

JIT…at the same time IBM, though, also learned, and implemented, the core lessons of the 'Just in Time' methods of scheduling; first developed by Taiichi Ohno of Toyota in the 1950's, and a natural extension of IBM's own computerised systems. The evidence for this was most graphically demonstrated when the UK's Havant plant proudly opened its newest (and largest) clean room in what had previously been the warehouse holding the stocks of components (and which was clearly no longer needed for this function!). Like the Japanese themselves the confident IBM of the early 1980s still  searched the world for ideas to improve its business; and did not hesitate to put into practice the best of them, regardless of their origins.

Toyota…its experiences - often described by others but not Toyota itself as JIT- have been much investigated by Western organisations; but usually without a great deal of understanding. Toyota does respond to the term, but insists that it is a philosophy which embraces a number of techniques rather than a specific technique itself. The techniques include flexible manufacturing, mentioned above, as well as the Kanban; the order-card based 'pull system' its plants use. Above all, it is based on the culture - embraced by its staff - of kaisen; continuous improvement. Western 'experts', in general, do not understand the nuances of Toyota's all-important philosophies - usually mistaking them for the stand-alone techniques which the West favours.

One cost which emerged, especially from the massive investments in fully automated plants, was that of reduced flexibility. IBM previously had a history of creative improvisation on a massive scale whenever a quick solution to a major problem was needed; the launch of the PC was just the most obvious of a number of such sagas. With increasing fixed investment in plant, which was not flexible, IBM's options considerably narrowed. This showed most obviously in its continued commitment to technologies that had been largely superseded elsewhere. So, once it had committed its invest­ment to the ten year timescales of the automation of a given technology, even IBM could not easily afford to divert to newer technologies, no matter how attractive they might have appeared. The problem was, once more, illustrated by the example of the Lexington plant. The projected product costs might well, in theory, have been significantly reduced; but only when the plant was running at optimum capacity. In reality the demand for its products (typewriters and small printers) was less than forecast; and accordingly the low costs anticipated were not actually achieved. On the other hand, the $20 billion IBM had built up in current assets, of which the $10 billion net current assets could be switched almost at a moments notice, did then indicate a readiness for even the most disastrous emergencies. It is unfortunate (and perhaps negligent) that IBM subsequently wasted this cushion of resources, as it frantically - and with a remarkably un-IBM-like degree of inconsistency - chased after one ill-thought out (product) rescue package after another.

Flexible Manufacturing…the old view - to which IBM unfortunately subscribed in the 1980s - was that to obtain economies of scale you needed to reduce the number (variety) of product lines and invest in massively automated (dedicated) production. The more modern approach, led by Japanese practice which focused in particular in the reduction of set-up times so that production lines might be rapidly switched from one product to another, emphasises flexibility. This gives economies with variety. This development is well recognized by those theorists working in the field of operations management.

Quality Control

Learning from the Japanese, IBM also set itself the target of 'zero‑defects'. It had previously been the tenet of the 'traditional' quality control department that there had to be a trade‑off between the percentage of defects and the cost of removing these, and this was an exponential relationship. There was no possibility of achieving 100% defect free product, and even achieving 99% was usually prohibitively expensive. The key to success was thus tradition­ally seen to be in achieving the correct balance via the AQL (Acceptable Quality Level). The Japanese - following the lead set by the American P. R. Crosby (then working on the US Pershing missile project) - revolutionised this by refusing to accept that 100% was impossible; the solution was simply in designing the products so that they could achieve 100%. IBM rationalized this about face by a rule of thumb that said that the cost of removing defects rises by an order of magnitude at each stage as work progresses from design, through production to instal­lation and then use (a ratio of 1000:1 overall).

There was a degree of irony in IBM learning from the Japanese. It is true that the Japanese took their quality control practices from the US; and even, at the beginning of the 1950's, recognised this by creating the 'Deming Prize' - named after the American professor. As popularly seen by the West, at least, their breakthrough came with their development of 'Quality Circles'. The irony is that, as described earlier, it is very possible that the original model for these 'circles' was IBM itself. The description of IBM in Peter Druckers very influential book ('The Practice of Management' ‑ a best-seller in Japan) included admiring reference to its job enrichment programmes, and in particular to its very productive involvement of the plant staff in the design of new products; both concepts that the Japanese seized on and (as usual) developed to their logical conclusion. There is no indication that IBM, in later imitating the Japanese, realised that it was itself probably the original source; but IBM too would readily incorporate any improvement, from whatever unlikely source. There was an even greater irony, perhaps, that in becoming so obsessed with these Japanese philosophies IBM lost its way (especially in terms of trying to become ' the lowest cost producer') and abandoned its greatest strengths (especially its Three Beliefs). This is especially ironic, where Toyota - the leader in quality - stresses, above all, its overall relations with its workforce - its commitment to HRS in general - as the reason for its success!

Quality Circles…the first lesson is the general one, that ideas transferred from Japan often lose considerably in translation - where the 'consultants' delivering them often do not fully understand the concepts behind them. The second lesson is that true quality typically emerges, over the long-term, as an outcome of improved HRS in general not from specific techniques. Management theory, however, generally places the emphasis on the specific techniques.

Yet another irony, perhaps, is that Toyota - which led the way on most developments in quality - actually places relatively little stress on Quality Circles, though Kono & Clegg[3] report that 80 per cent of Japanese companies employ them (compared with 40 per cent in the US).Toyota itself emphasises, instead, its much larger 'suggestion scheme' (which it proudly reported had 2.65 million suggestions, 48 per employee, in 1986); which, also as described earlier, actually focuses on detecting problems. Thus, one of the most powerful techniques for continuous improvement (kaizen) was actually developed by an American, as part of his '14 Steps to Quality'. In essence, it is a suggestion scheme, but with an important difference. Whereas its Western counterparts - which usually only result in dust-laden boxes decorating the office wall - insist that the employee finds the answer Crosby's (Japanese) version simply requires that they ask the question; and it is then up to management to develop the answer - recognizing that the most difficult part of the process is finding the question. It may be that the reason it is unheard of in the West is that it is called Error Cause Removal (ECR), a terrible title - but a process to which Toyota ascribe much of their success. In confirmation of its importance there, Kono & Clegg[4], reporting the results of their 1995 survey, said "…80 per cent of Japanese corporations employ a suggestion system and between 3.5 and 24 ideas per person per year are presented."

Error Cause Removal - ECR…this is, according to Toyota, the most important technique in implementing quality - yet it is unknown in the West. It is also simple to implement, once you have the right relationship with your workers! You just ask them to tell you what the problems are. Toyota's management  implements solutions to at least 95% of the literally millions of these 'suggestions' it receives each year. Even when quality circles are discussed, the much more important topic of ECRs is rarely mentioned in the West!

Buying-In

Even before its adoption of the fashionable 'market forces' philosophy, IBM only manufactured components itself where it had unique expertise, which could not be matched outside; for example in production of substrates and the most sophisticated chips, and of course in very specific areas such as disk technology. One of the great mistakes, however, in the case of the PC was to continue to buy in the chip (which was the heart of the PC, and a technology where IBM then led the world) from an outside supplier; in this way effectively giving away its (product) birthright to Intel! All other components wherever possible are sourced on outside suppliers. The quality demands are not, however, relaxed. Even the metal‑bashers who produce cabinets are not excused the demanding quality standards that should make any supplier think twice. In my previous company, BTR, we struggled for 5 years in the late 1960's to produce the most basic handmade rubber tube which simply held a wiring loom in place. We never actually succeeded in meeting IBM's quality standards; but as with many IBM suppliers the totally disproportionate costs were justified on the basis of the cachet, and 'reference' potential, possible if we succeeded. Paradoxically, about the first thing I realised on reaching IBM was that the part had long since become obsolete in terms of current production!

Global Production…the other major difference in IBM mainframe manufacturing is its global scale. Apart from the PC plants, which largely contained assembly lines, IBM factories only produced parts of a complete computer system. Thus, at the time when it reached its peak, a given system, ordered by a European customer, the water‑cooled CPU - now a fossilised technology - might come from France (but with ceramic multi‑layer substrates - again now fossilised by IBM's move to more conventional chip technology)  - and memory modules from Germany, and power supplies from Spain. The disk drives might come from Germany and the tape drives from Spain, the printers from Sweden, and the display units (VDU's) from the UK but with network mini‑computers (as controllers) coming from Italy! The AS/400, for example, was produced at 37 sites on 3 continents. The result was that IBM in effect had to run a 'single factory' with a number of separate departments producing a large range of components that were eventually bolted together to produce the final product. The fact that these individual departments were separated by thousand of miles was, in IBM terms, easily ignored. Even now, physical transport, even by air, is relatively cheap compared with the very high cost components; and is rapid. Electronic communication, which is essential for integrating control is, of course for IBM, second nature. In essence the 'logical factory', the 'electronic shop floor', was held on computers; for example, in the case of EMEA at ISL (IBM Information Systems Limited), which was IBM's own computerised production control centre, located at Portsmouth, England. Scheduling of all EMEA plants was accomplished there; even if the final destination was in South Africa and the plant was in Italy.

The Virtual Factory…increasingly, especially at the ERP (Enterprise Resource Planning) end of Business-To-Business (B2B) e-commerce, components are sourced, and manufactured, world-wide; and production (control) then lies inside the computer which arranges how they are to be brought together. Production is now often a function of IT, rather than the other way round. Management theory no longer overly concerns itself with production - even in the case of ERP - and, even if it does, it follows the traditional view of this.

Administration

This naturally leads on to the IBM function - Administration and General (A&G) - which has, from the early 1980s onwards, threatened to consume all the profits. In IBM the 'sharp end', be it sales at one extreme or manufacturing at the other, was rigorously measured ; and as such was contained. A & G though was essentially unmeasurable and IBM had as much difficulty as any other company in finding out just what this bureaucracy thought it was doing, let alone in exerting control. It was only when IBM's problems became life threatening, in the early 1990s, that some sort of control was regained; and the proportion of costs accruing to A&G declined.

A & G is, though, a global description for a wide range of IBM functions. At its most obvious it contains the administrators who process all the paperwork, the many thousands of different forms, that make any large organisation tick. Such material is now largely held in electronic form, which reduces the paper mountains to somewhat more manageable proportions. Since the 1980s almost all IBM's desk workers, even temporary ones, have had their own dedicated terminal; with - even in the mid 1980s - more than a quarter of a million terminals in use world-wide. The main problem this causes is that IBM is its own most complex customer, with the most advanced leading edge requirements.

A & G covers a multitude of functions, which is in itself one of the prob­lems of control; how do you control a function which only its members understand? As an example of its diversity at the peak of its success, A & G even included a range of departments handling property development. IBM was then, in its own right, one of the world's major property developers; a paradox for a company dedicated to making the office building obsolete!

A & G may well have been a black hole in absorbing ever increasing resource from IBM, but it was probably better controlled than most bureaucracies, simply because the calibre of the individual personnel and their motivation were once more significantly higher than in other companies. The problem too often was that this bureaucracy was more effective than most in implementing the formal rules of IBM to the very letter, and indeed adding further very profession­ally produced rules to protect the best interests of IBM as the group (somewhat parochially) saw it. This was, unfortunately, totally unsuited to the very rapidly moving business environment it found itself in from the 1980s onwards!

Temps…the problem of geometric growth of the administrative bureaucracy may have been, even at its peak, greater than IBM was willing to admit even to itself, because all of its headcount figures were based on permanent employees. In the early 1980s, for instance, parts of IBM employed, on a regular basis, temporary and contract staff accounting for in excess of a third of the total workforce. The cut-backs of the late 1980s presumably fell first on these temporary workers, but that probably only exacerbated the problem - as permanent IBMers would have had to be drafted to take over at least some of the roles. It never was a matter of controlling the numbers of people. IBM did not have  a problem of 'window-gazers' - the Japanese description for those personnel who had no jobs to do but were still employed because of 'lifetime employment' policies. It was that of controlling unnecessary work.

Over the decade of the 1980s, IBM tried desperately to reduce the numbers in A&G - typically by switching them to more productive tasks - almost all of which failed. At the end of the decade, the chosen solution was probably the least successful of all. It involved drafting such staff into the field support functions; especially into the sales-force. It is not obvious that this reduced the workload of A&G, though its annual reports did show a reduction in the share of costs taken by A&G, but it diluted the quality of the sales-force! IBM seemed incapable of dealing with this problem - until - in the 1990s - it finally abandoned its cherished philosophies. The move from 'customer service' meant that failures resulting from under-resourced administration no longer mattered. In particular, however, the removal of the guarantee of 'lifetime employment' meant that surplus personnel could simply be made redundant! The ultimate weapon had been deployed.

Explosive Admin…in a rapidly expanding organisation, admin may grow faster still. This is partly because there is seen to be a greater need for (admin) control, and partly because there is the money to fund this. When overall growth stops, however, admin growth may not - and control of this may be very difficult, and certainly must be very different in style. Control of admin costs, as opposed to operating costs,  is not normally a focus of attention in management theory.

Portents in the 1980s

In a previous book, written in the mid-1980s when the shape of the new IBM hard barely begun to emerge, I wrote "IBM is, however, in a state of flux…IBM is starting to think the previously unthinkable…the question of how much IBM can get away with is being asked more frequently and more strongly; and occ­asionally it is now also being asked by Business Practices rather than of it! Its role has changed subtly from that of gamekeeper to poacher. IBM no longer has the threat of anti‑trust action hanging over it, and a decade and a half of repressed aggression against its competitors is threatening to surface." My fears were fully justified, as later history graphically demonstrated.

[back]     [home]

hits