As global competition grows ever fiercer in manufacturing industries, American managers are adopting a new battle cry: “Beat ’em with technology or move—over there.” Indeed, since 1975, the boom in information-intensive processing technologies has been explosive. A close look at how U.S. managers are actually using these technologies, however, silences their battle cry in a hurry. Yes, they are buying the hardware of flexible automation—but they are using it very poorly. Rather than narrowing the competitive gap with Japan, the technology of automation is widening it further.

With few exceptions, the flexible manufacturing systems installed in the United States show an astonishing lack of flexibility. In many cases, they perform worse than the conventional technology they replace. The technology itself is not to blame; it is management that makes the difference. Compared with Japanese systems, those in U.S. plants produce an order-of-magnitude less variety of parts. Furthermore, they cannot run untended for a whole shift, are not integrated with the rest of their factories, and are less reliable. Even the good ones form, at best, a small oasis in a desert of mediocrity.

Lest this sound unduly harsh, consider the facts summarized in Exhibit I. In 1984 I conducted a focused study of 35 flexible manufacturing systems (FMSs) in the United States and 60 in Japan, a sample that represented more than half the installed systems in both countries. The kinds of products they made—large housings, crankcases, and the like—were comparable in size and complexity, and required similar metal-cutting times, numbers of tools, and precision of parts. The U.S. systems had an average of seven machines and the Japanese, six.

Exhibit I Comparison of FMSs studied in the United States and Japan

Here the similarities end. The average number of parts made by an FMS in the United States was 10; in Japan the average was 93, almost ten times greater. Seven of the U.S. systems made just 3 parts. The U.S. companies used FMSs the wrong way—for high-volume production of a few parts rather than for high-variety production of many parts at low cost per unit. Thus the annual volume per part in the United States was 1,727; in Japan, only 258. Nor have U.S. installations exploited opportunities to introduce new products. For every new part introduced into a U.S. system, 22 parts were introduced in Japan. In the critical metal-working industries, from which these numbers come, the United States is not using manufacturing technology effectively. Japan is.

I have spent several years examining the experiences of companies that have installed FMSs. (See the insert entitled “Primary Research” for details of the study.) The object has been to observe the most sophisticated form of information-intensive technology in manufacturing. Flexible systems resemble miniature factories in operation. They are natural laboratories in which to study computer-integrated manufacturing, which is rapidly becoming the battleground for manufacturing supremacy around the globe.

The battle is on, and the United States is losing badly. It may even lose the war if it doesn’t soon figure out how better to use the new technology of automation for competitive advantage. This does not mean investing in more equipment; in today’s manufacturing environment, it is how the equipment is used that is important. Success comes from achieving continuous process improvement through organizational learning and experimentation.

Technology Leadership

The FMS installations surveyed in Exhibit I were, as noted, technically alike. They had similar machines and did similar types of work. The difference in results was mainly due to the extent of the installed base of machinery, the work force’s technical literacy, and management’s competence. In each of these areas, Japan was far ahead of the United States.

In the last five years, Japan has outspent the United States two to one in automation. During that time, 55% of the machine tools introduced in Japan were computer numerically controlled (CNC) machines, key parts of FMSs. In the United States, the figure was only 18%. Of all these machines installed worldwide since 1975, more than 40% are in Japan. What’s more, over two-thirds of the CNC machines in Japan went to small and medium-sized companies.

Just counting how much of this technology companies use is not enough. Because software development lies at the heart of this increasingly information-intensive manufacturing process, the technological literacy of a company’s workers is critical. In the Japanese companies I studied, more than 40% of the work force was made up of college-educated engineers, and all had been trained in the use of CNC machines. In the U.S. companies studied, only 8% of the workers were engineers, and less than 25% had been trained on CNC machines. Training to upgrade skills was 3 times longer in Japan than in the United States. Compared with U.S. plants, Japanese factories had an average of 2½ times as many CNC machines, 4 times as many engineers, and 4 times as many people trained to use the machines.

Management’s role

A skilled work force and a large installed base of equipment build the foundation for technological leadership. It is the competence of managers, however, that makes such leadership happen. To understand why, we should look more closely at recent experience with FMS technology.

A flexible manufacturing system is a computer-controlled grouping of semi-independent work stations linked by automated material-handling systems. The purpose of an FMS is to manufacture efficiently several kinds of parts at low to medium volumes. All activities in the system—metal cutting, monitoring tool wear, moving parts from one machine to another, setup, inspection, tool adjustment, material handling, scheduling, and dispatching—are under precise computer control. In operation, an FMS is a miniature automated factory.

The system at one prominent Midwestern heavy-equipment producer consisted of 12 machines that made just 8 different parts for a total volume of 5,000 units a year. Once the FMS went on line, management prevented workers from making process improvements by encouraging them not to make any changes. “If it ain’t broke, don’t fix it” became the watchword.

The FMS boosted machine uptime and productivity, but it did not come close to realizing its full—and distinctive—strategic promise. The technology was applied in a way that ignored its huge potential for flexibility and for generating organizational learning.

Management treated the FMS as if it were just another set of machines for high-volume, standardized production—which is precisely what it is not. Captive to old-fashioned Taylorism and its principles of scientific management, these executives separated the establishment of procedures from their execution, replaced skilled blue-collar machinists with trained operators, and emphasized machine uptime and productivity. In short, they mastered narrow-purpose production on expensive FMS technology designed for high-powered, flexible usage.

This is no way to run a railroad. Certainly, Frederick W. Taylor’s work still applies—but not to this environment. Managing an FMS as if it were the old Ford plant at River Rouge is worse than wrong; it is paralyzing. In this case there was little, if any, attention given to process or program flexibility and almost no support for software improvement. Management failed to utilize the FMS’s improved capabilities, from which even greater improvements might have flowed over time.

Not surprisingly, the flexibility achieved by this FMS was much less than that of a stand-alone CNC machining center. And that’s the good news. The system had four operators per shift, each of whom was responsible for checking gauges, changing hydraulic fluid and parts like drill bits, and making simple diagnoses when something went wrong. These tasks, as specified by management, were very procedural, and no operator had the discretion to change procedures. If anything, the complexity of the FMS forced operators to stick more rigidly to procedure than they did at the stand-alone CNC machining centers.

Goals for management

How, then, should managers look at FMSs? About what should they ask? For one thing, development time. The systems in the United States take 2½ to 3 years and about 25,000 man-hours to conceive, develop, install, and get running. Japanese systems take 1¼ to 1¾ years and 6,000 man-hours. Here, again, the difference is management. U.S. project teams are usually large groups made up of specialists who design systems for a much greater level of flexibility than their companies are prepared to use. This greater complexity means that projects not only take longer but have plenty of bugs when finished. Delays create enormous pressure on software engineers to take shortcuts and seek hard-wired fixes.

At the end of a project, as a rule, the team is disbanded. The engineers assigned to maintain a system, who are usually not its developers, are reluctant to make any changes. They know about all the bugs but are unwilling to tinker with things because “you never know what may happen.” The result: inflexibility.

By contrast, the FMS installations in Japan are remarkably flexible. This would not be so troublesome for the United States if the old-fashioned productivity of its systems, for which flexibility gets sacrificed, were better than that of Japanese systems. But it is not. The average utilization rate (metal-cutting time as a percentage of total time) of U.S. flexible manufacturing systems over two shifts was 52%, as opposed to 84% in Japan. Over three shifts, because of reliable untended operations, the figure in Japan was even higher.

Where does so huge a difference come from? In a word, the reliability designed into the system. In Japan, system designers strive to create operations that can run untended. Of the 60 FMSs I studied, 18 ran untended during the night shift. Such systems take more time and resources to develop than those that require even a single attendant, because designers have to anticipate all possible contingencies. But the additional costs are well worth it. So demanding a design objective leads in practice to a great deal of advance problem solving and process improvement. The entire project team remains with the system long after installation, continually making changes. Learning occurs throughout—and learning gets translated into ongoing process mastery and productivity enhancement. This learning is what gives rise to, and sustains, competitive advantage.

Most of the systems built in Japan after 1982 have achieved untended operations and system uptime of an astonishing 90% to 99%. Operators on the shop floor make continual programming changes and are responsible for writing new programs for both parts and systems as a whole. They are highly skilled engineers with multifunctional responsibilities. Like the designers, they work best in small teams. Most important, managers see FMS technology for what it is—flexible—and create operating objectives and protocols that capitalize on this special capability. Not bound by outdated mass-production assumptions, they view the challenge of flexible manufacturing as automating a job shop, not simply making a transfer line flexible. The difference in results is enormous, but the vision that leads to it is in human scale. No magic here—just an intelligent process of thinking through what new technology means for how work should be organized.

FMS On Line

To find out more about this “job shop” approach, I examined more closely 22 FMS installations at Hitachi-Seiki, Yamazaki Mazak, Okuma, Murata, Mori-Seiki, Makino, and Fanuc. As Exhibit II shows, these systems far outperformed the conventional CNC equipment they replaced.

Exhibit II Comparison of Japanese FMSs with the systems they replaced

Both systems produced the same variety of parts. But the FMSs did it with five times fewer workers than the conventional systems. Moreover, it took only half as many flexible machines to produce the same volume of parts as conventional machines. The CNC machines used in both systems were identical; the FMSs, however, also employed robots, special material-handling equipment, automated storage systems, and tool-handling equipment. These support devices added another 30% to hardware costs, but they helped boost average uptime from 61% to 92% and made untended operations possible. These benefits alone more than justified the extra cost; better quality and reduced inventories were a bonus.

Potential FMS users often worry that the systems are difficult to justify strictly in economic terms. Based on the experiences of Japanese companies, these fears are groundless. All 22 systems I studied in Japan met their companies’ ROI criterion of a three-year payback.

Even so, the impact of flexible manufacturing on the performance of a company reaches far beyond simple productivity rates and investment calculations. FMSs take on strategic importance when the installed base of flexible systems in a factory reaches a critical mass. Only when separate “islands of automation” in a plant start to link does management realize the possibilities for new kinds of competitive advantage via manufacturing.

Of the six Japanese companies that used flexible automation extensively, three had fully automated fabrication plants. At the time I visited them, they were the only flexible manufacturing factories in the world. Their productivity was stupendous.

Exhibit III compares the performance of one such factory before and after the introduction of total flexible automation, and Exhibit IV shows the effect of such performance on cost structure and competition in an industry. Specifically, the exhibit compares the manpower requirements of various manufacturing systems for metal-cutting operations: if it took 100 people in a conventional Japanese factory to make a certain number of machine parts, it would take 194 people in a conventional U.S. factory—but only 43 in a Japanese FMS-equipped factory. If U.S. companies mastered flexible automation as the Japanese have, they would have more than a fourfold increase in labor productivity. This efficiency in labor is part of the reason that smaller companies in Japan have been able to use FMS technology so effectively.

Exhibit III Performance of one factory before and after automation

Exhibit IV Manpower requirements for metal-cutting operations to make the same number of identical parts Note: There is no column here for FMSs in the United States because, at the time of this study, no domestic machine tool producer had an FMS on line.

Perhaps even more interesting than such aggregate improvements are their components. The largest manpower reduction in the exhibit is in manufacturing overhead, where an FMS cuts the number of workers from 64 to 5. In engineering, an FMS cuts the number of workers from 34 to 16. One consequence of these reductions (92% in manufacturing overhead, but only 53% in engineering) is to change the composition of the work force: engineers now outnumber production workers three to one. This may not sound like much at first, but it signals a fundamental change in the environment of manufacturing.

Flexible automation shifts the arena of competition from manufacturing to engineering, from running the plant to planning it. In the FMS environment, engineering innovation and engineering productivity hold the keys to success. Engineering now performs the critical line function. Manufacturing has become, by comparison, a staff or support function.

Managing Above the Line

Picture a “lights out” factory operating untended, with general-purpose CNC machines that make a wide variety of parts and are capable of adapting easily to new demands. If two such factories compete with similar products, competition will focus on price. This is so because all costs in the development of tools, fixtures, and programs are sunk before the first unit is produced. The only variable costs are those of materials and energy, which usually amount to less than 10% of total costs.

Each factory’s profits will erode over time as other companies acquire the same operating capabilities. How, then, would a company stay ahead? One way is by creating new physical assets in the form of better programmed and better managed equipment. Each plant’s competitive fate would rest heavily on its ability to create facilities that generate performance advantages—and to do it faster than the competition. When the lion’s share of costs are sunk before production starts, the creation and management of intellectual assets becomes the prime task of management.

This is manufacturing’s new competitive environment. It may sound like something from the distant future, but the Japanese are doing it now. The crucial variable in this kind of environment is automation—the ability of an FMS to run untended. And Japanese manufacturing companies are becoming increasingly expert in that field.

Exhibit V summarizes my findings from 20 of the 22 Japanese FMSs on the extraordinary degree of automation reflected in different production activities. Exhibit VI presents data from these 20 systems on the amount of manual labor time spent on the factory floor to support such levels of automation. Average system losses took 16.6% of total operating time, about a third the figure in U.S. systems. Each 144 hours of metal-cutting time took only 26 hours of manual effort, which included direct labor as well as required activities usually associated with manufacturing overhead. In the United States, manufacturing overhead activities are separated from direct labor and take about ten times longer.

Exhibit V Degree of automation in production activities of 20 Japanese FMSs in numbers of systems

Exhibit VI Production activities in an average FMS in Japan*

In most plants, 26 hours of manual effort translate into two workers per system for each of two shifts. In all 22 of these FMSs, however, there was a third person on each shift, whose work accounted for part of the 26 hours of manual effort. By dividing the work among three people, the companies that had these systems purposely created extra time for such process-improvement activities as additional test cutting of new parts, observing machine behavior, and examining statistics on performance. In all 22 systems, each of the workers did these nonrequired—but immensely valuable—tasks. The number of people required to do all this in conventional systems making the same parts in the same companies was four times greater.

The distribution of the 26 hours of manual effort is also instructive. More than half were spent loading and unloading pallets. The other major activity, which took 7 hours, was mounting tools and qualifying them on machines. Together, these efforts accounted for 80% of the time spent on manual labor. Workers loaded pallets and mounted tools during the day shift, and the machines ran untended at night. Production planning, a weekly activity, took only one hour of a person’s time. Systems making a large variety of parts also had automated methods for production planning. Those with a low variety of parts did it manually.

The FMS installations performed exceptionally well. Delivery performance in each system, tracked during a three-month period, was 100%. The high reliability of individual machines and of the system itself kept the variance in unscheduled downtime to only 2%. The scheduled slack for software testing and process experiments ranged from 4% to 9% of capacity and was more than enough to accommodate any variation in machine reliability. Each system met its production schedule, as long as the schedule observed the constraints of capacity. In addition, only six pieces in a thousand had a quality problem. Of these, three were reworked, usually by the operators themselves. The other three were scrapped. Tool breakage caused most of these quality problems, and the machine operators could make the necessary adjustments.

With such impressive levels of performance, few contingencies demanded management’s attention. In fact, executives were largely absent from day-to-day operations. Instead of concerning themselves with internal operations, they focused their attention on how to meet competitive pressures on product performance. In the United States, on the other hand, managers spend so much time on routine problems with quality and production delivery schedules that they have virtually no time left over to plan for long-term process improvement.

As noted before, the prime task of management once the system has been made reliable is not to categorize tasks or regiment workers but to create the fixed assets—the systems and software—needed to make products. This calls for intellectual assets, not just pieces of hardware. Thus the new role of management in manufacturing is to create and nurture the project teams whose intellectual capabilities produce competitive advantage. What gets managed is intellectual capital, not equipment.

The technology of flexible manufacturing has led managers into a drastically altered competitive landscape. This new landscape has a number of important features:

  • A sharp focus on intellectual assets as the basis for a company’s distinctive competence.

  • A heightened emphasis on the selection of the portfolio of projects a company chooses to manage.
  • A close attention to the market and to the special competence of process engineers.
  • A steady adjustment of product mix and price in order to maintain full capacity utilization.
  • A pointed emphasis on reducing fixed manufacturing costs and the time required to generate new products, processes, and programs.
  • An intensification of cost-based competition for manufactured products.

I am convinced that the heart of this new manufacturing landscape is the management of manufacturing projects: selecting them, creating teams to work on them, and managing workers’ intellectual development. In company after company in Japan, systems engineers with a thorough knowledge of several disciplines have proved the key to the success of flexible manufacturing systems. One rigidly organized Japanese company, recognizing the importance of such versatile teams, now rotates experienced engineers through all manufacturing departments. Another, which already had job rotation, has begun to keep its engineers longer in each area so they can learn more from their FMS experience.

In contrast with the traditional Japanese approach of involving a large number of people in decision making, small teams of highly competent, engineering-oriented people have been most successful with flexible manufacturing. These groups have succeeded because they are given responsibility for both design and operations. They remain on a project until the FMS achieves 90% uptime and untended operations. Perhaps most important, in all the Japanese companies I studied, the teams came entirely from engineering and were given line responsibility for day-to-day operations.

New Mission Statement

The management of FMS technology is taking place in a different manufacturing environment, and thus consists of new imperatives:

Build small, cohesive teams. Very small groups of highly skilled generalists show a remarkable propensity to succeed.

Manage process improvement, not just output. FMS technology fundamentally alters the economics of production by drastically reducing variable labor costs. When these costs are low, little can be gained by reducing them further. The challenge is to develop and manage physical and intellectual assets, not the production of goods. Choosing projects that develop intellectual and physical assets is more important than monitoring the costs of day-to-day operations. Old-fashioned, sweat-of-the-brow manufacturing effort is now less important than system design and team organization.

Broaden the role of engineering management to include manufacturing. The use of small, technologically proficient teams to design, run, and improve FMS operations signals a shift in focus from managing people to managing knowledge, from controlling variable costs to managing fixed costs, and from production planning to project selection. This shift gives engineering the line responsibilities that have long been the province of manufacturing.

Treat manufacturing as a service. In an untended FMS environment, all of the tools and software programs required to make a part have to be created before the first unit is produced. While the same is true of typical parts and assembly operations, the difference in an FMS is that there are no allowances for in-the-line, people-intensive adjustments. As a result, competitive success increasingly depends on management’s ability to anticipate and respond quickly to changing market needs. With FMS technology, even a small, specialized operation can accommodate shifts in demand. Manufacturing now responds much like a professional service industry, customizing its offerings to the preferences of special market segments.

Making flexibility and responsiveness the mission of manufacturing flies in the face of Taylor’s view of the world, which for 75 years has shaped thinking about manufacturing. FMS technology points inevitably toward a new managerial ethos—an ethos dedicated to the building of knowledge in the flexible service of markets, not merely to the building of things. Scale is no longer the central concern. Size no longer provides barriers to entry. The minimum efficient scale for FMS operations is a cell of roughly six machines and fewer than a half a dozen people. That’s the new reality.

Going to FMS-based operations does not require lots of money or people. It can be done—at its best, it is done—on a small scale. The critical ingredient here is nothing other than the competence of a small group of people. There is no Eastern mystery in this, no secrets known only to the Japanese. We can do it too—if we will.

What, after all, is a manufacturing company? Today, no artist would represent a factory as a huge, austere building with bellowing smokestacks. The behemoth is gone. The efficient factory is now an aggregation of small cells of electronically linked and controlled FMSs. New technology enables these operating cells to be combined in nonlinear ways. No shared base of infrastructure mandates large-scale production integration. The days of Taylor’s immense, linear production systems are largely gone.

Unless U.S. managers understand the implications of Japan’s mastery of FMS technology, their companies will fall further behind. Flexible manufacturing systems are no longer a theory, a pipe dream. They exist. And the leverage they provide on continuous process improvement is immense. Making automation work means a whole new level of process mastery. A large number of Japanese factories demonstrate its reality every day. They lead the way; we linger behind at our own peril.

A version of this article appeared in the November 1986 issue of Harvard Business Review.