What Is a Single Line of Poetry Called

• Tweet

• Post

• Share

• Save

• Get PDF

• Buy Copies

• Print

Analyses of what has gone wrong with American industry have returned, time and again, to the poor quality of American-made products and to the management philosophy responsible for that quality. To date, most of the available evidence has been largely impressionistic, and few managers have felt the need to question familiar, long-established approaches to the work of manufacturing. We no longer have that excuse.

Mr. Garvin has spent several years studying the quality performance of virtually every competitor, American and Japanese, and every plant in a single but broadly representative manufacturing industry: room air conditioners. His findings document a bitter but inescapable truth. The competitively significant variation in levels of quality performance is immense. But the truth is also heartening: superior levels of performance come not from national traits or cultural advantages but from sound management practices deliberately and systematically applied. If nothing else, the data reported here should force American executives to rethink their approaches to product quality.

When it comes to product quality, American managers still think the competitive problem much less serious than it really is. Because defining the term accurately within a company is so difficult (is quality a measure of performance, for example, or reliability or durability), managers often claim they cannot know how their product quality stacks up against that of their competitors, who may well have chosen an entirely different quality "mix." And since any comparisons are likely to wind up as comparisons of apples with oranges, even a troubling variation in results may reflect only a legitimate variation in strategy. Is there, then, a competitive problem worth worrying about?

I have recently completed a multiyear study of production operations at all but one of the manufacturers of room air conditioners in both the United States and Japan (details of the study are given in the insert, Research Methods). Each manufacturer uses a simple assembly-line process; each uses much the same manufacturing equipment; each makes an essentially standardized product. No apples versus oranges here: the comparison is firmly grounded. And although my data come from a single industry, both that industry's manufacturing process and its managers' range of approaches to product quality give these findings a more general applicability.

This article is based mainly on data collected in 1981 and 1982 from U.S. and Japanese manufacturers of room air conditioners. I selected that industry for study for a number of reasons: it contains companies of varying size and character, which implies a wide range of quality policies and performance; its products are standardized, which facilitates inter-company comparisons; and it employs a simple assembly-line process, which is representative of many other mass production industries.

Nine of the ten U.S. companies in the industry and all seven of the Japanese companies participated in the study. They range in size from small air conditioning specialists with total sales of under $50 million to large home appliance manufacturers with annual sales of more than $200 million in this product line alone. Taken together, they account for approximately 90% of U.S. industry shipments and 90% of Japanese industry shipments. I have collected data separately for each plant (two of the American companies operate two plants apiece; otherwise, each company employs only a single plant). Of the 18 plants studied, 11 are American and 7 Japanese.

Once U.S. companies had agreed to participate in the study, I sent them a questionnaire requesting background information on their product line, production practices, vendor management practices, quality policies, and quality performance. I then visited them all in order to review the questionnaire results, collect additional data, tour the factories, and conduct interviews with key personnel. The interviews were open-ended and unstructured, although I posed similar questions at each company. A typical visit included interviews with managers in the quality, manufacturing, purchasing, engineering, and service departments, as well as several hours spent walking the production floor.

Preliminary analysis of the interviews and questionnaires showed that companies neither employed the same conventions in reporting data nor answered questions in the same degree of detail. I therefore sent each company its own set of follow-up questions to fill in these gaps and to make the data more comparable across companies. In addition, I requested each company to administer a brief questionnaire on quality attitudes to each of its first-line production supervisors.

I followed a similar approach with the Japanese manufacturers, although time constraints limited the amount of information that I could collect. All questionnaires were first translated into Japanese and mailed to the participating companies. Six of the seven companies completed the same basic quality questionnaire as did their American counterparts; the same companies also administered the survey on quality attitudes to a small number of their first-line supervisors. With the aid of a translator, I conducted on-site interviews at all the companies and toured six of the plants.

The shocking news, for which nothing had prepared me, is that the failure rates of products from the highest-quality producers were between 500 and 1,000 times less than those of products from the lowest. The "between 500 and 1,000" is not a typographical error but an inescapable fact. There is indeed a competitive problem worth worrying about.

Measuring Quality

Exhibit I presents a composite picture of the quality performance of U.S. and Japanese manufacturers of room air conditioners. I have measured quality in two ways: by the incidence of "internal" and of "external" failures. Internal failures include all defects observed (either during fabrication or along the assembly line) before the product leaves the factory; external failures include all problems incurred in the field after the unit has been installed. As a proxy for the latter, I have used the number of service calls recorded during the product's first year of warranty coverage because that was the only period for which U.S. and Japanese manufacturers offered comparable warranties.

Exhibit I Quality in the room air conditioning industry 1981–1982*

Measured by either criterion, Japanese companies were far superior to their U.S. counterparts: their average assembly-line defect rate was almost 70 times lower and their average first-year service call rate nearly 17 times better. Nor can this variation in performance be attributed simply to differences in the number of minor, appearance-related defects. Classifying failures by major functional problems (leaks, electrical) or by component failure rates (compressors, thermostats, fan motors) does not change the results.

More startling, on both internal and external measures, the poorest Japanese company typically had a failure rate less than half that of the best U.S. manufacturer. Even among the U.S. companies themselves, there was considerable variation. Assembly-line defects ranged from 7 to 165 per 100 units—a factor of 20 separating the best performer from the worst—and service call rates varied by a factor of 5.

For ease of analysis, I have grouped the companies studied according to their quality performance (see the Appendix). These groupings illustrate an important point: quality pays. Exhibit II, for example, presents information on assembly-line productivity for each of these categories and shows that the highest-quality producers were also those with the highest output per man-hour. On the basis of the number of direct labor hours actually worked on the assembly line, productivity at the best U.S. companies was five times higher than at the worst.

To identify patterns of behavior, I first grouped U.S. plants into categories according to their quality performance on two dimensions—internal quality (defect rates in the factory) and external quality (failure rates in the field).

Table A presents the basic data on external quality. I measured field performance in two ways: by the service call rate for units under first-year warranty coverage (the total number of service calls recorded in 1981 divided by the number of units in the field with active first-year warranties) and by the service call rate for units under first-year warranty coverage less "customer instruction calls" (only those service calls that resulted from a faulty unit, not from a customer who was using the unit improperly or had failed to install it correctly).

Table A Field performance for U.S. plants in 1981

The second measure was necessary because companies differed in their policies toward customer instruction calls. Some reimbursed repairmen for these calls without argument; others did their best to eliminate such calls completely. An accurate assessment of product performance required the separation of these calls from problems that reflect genuinely defective units.

I classified plants on the basis of their rankings on the second of the two measures in Table A, and then grouped them according to their actual levels of field failures. In most cases, the dividing lines were clear, although there were some borderline cases. Plant 8, for example, had a total service call rate well above the industry median, yet after subtracting customer instruction calls, its failure rate differed little from the other average performers. Because this second figure more accurately reflects the rate of product malfunction, I treated Plant 8 as having average, rather than poor, external quality. A number of companies with high failure rates did not break out customer instruction calls. I have treated them as having poor external quality because their customer instruction calls would have to have been two or three times as frequent as the highest rate recorded in 1981 for them to have warranted an average ranking.

I followed a similar procedure in classifying plants on internal quality. Because companies differed in how they defined and recorded defects (some noted every single product flaw; others were interested only in major malfunctions), I employed several indexes to ensure consistency. The results are displayed in Table B. I ranked companies first by their total assembly-line defect rates (every defect recorded at every station along the assembly line divided by the number of units produced) and then by the number of defects requiring off-line repair. The second index compensates for the different definitions just noted, for it more accurately reflects the incidence of serious problems. Minor adjustments and touch-ups can generally be made without pulling a unit off the line; more serious problems normally require off-line repair. Measured on this basis, the high total defect rates of Plants 1 and 9 appear to be much less of a problem.

Table B Internal quality for U.S. plants in 1981

Because several companies had to estimate the off-line repair rate, I used a third index, the number of repairmen per assembly-line direct laborer, to measure defect seriousness. The proportion of the work force engaged in repair activities, including workers assigned to separate rework lines and to rework activities in the warehouse, is likely to correlate well with the incidence of serious defects, for more serious problems usually require more time to correct and necessitate a larger repair staff. This measure provides important additional information, confirming the conclusions about Plant 1 (its high total defect rate appears to include a large number of minor problems) but contradicting those about Plant 9 (its large number of repairmen suggests that defects are, in fact, a serious problem, despite the small proportion of units requiring off-line repair).

I assigned plants to groups using much the same procedure as before. I first computed a composite ranking for each plant by averaging together the three rankings of Table B. Dividing lines between groups followed the absolute levels of the indexes for each plant. Once again, some judgment was involved, particularly for Plants 4,5, and 9. Plants 5 and 9 were borderline cases, candidates for ranking as either average or poor internal quality. I classified the former as average, even though its overall rank was low, because its absolute scores on the first two measures were quite close to the median. I classified the latter as poor because its absolute scores on both the first and the third measures were so high. Plant 4 presented a different problem, for it provided no information at all on assembly-line defects. Rather than classifying the plant on the basis of the third index alone, I employed supplementary data. Based on its defect rate at the end-of-the-line quality audit and its rework and scrap costs as a percentage of sales, both of which were quite close to figures reported by other companies with average internal quality, Plant 4 showed up as an average performer.

Table C combines the results of the previous two tables. Overall quality rankings appear for each plant. In most cases, success on internal quality implied success on external measures, although the correlation is not perfect, as Plants 1,7, and 8 demonstrate. The Japanese plants are in a category of their own, for on both internal and external measures they are at least twice as good as the best U.S. plant.

Table C Classification of plants on internal and external quality

Exhibit II Quality and productivity

Measuring productivity by "standard output" (see Exhibit II) blurs the picture somewhat. Although the Japanese plants maintain a slight edge over the best U.S. plants, categories of performance tend to overlap. The figures based on standard output, however, are rather imperfect indicators of productivity—for example, they fail to include overtime or rework hours and so overstate productivity levels, particularly at the poorer companies, which devote more of their time to correcting defects. Thus, these figures have less significance than do those based on the number of hours actually worked.

Note carefully that the strong association between productivity and quality is not explained by differences in technology or capital intensity, for most of the plants employed similar manufacturing techniques. This was especially true of final assembly, where manual operations, such as hand brazing and the insertion of color-coded wires, were the norm. Japanese plants did use some automated transfer lines and packaging equipment, but only in compressor manufacturing and case welding was the difference in automation significant.

The association between cost and quality is equally strong. Reducing field failures means lower warranty costs, and reducing factory defects cuts expenditures on rework and scrap. As Exhibit III shows, the Japanese manufacturers incurred warranty costs averaging 0.6% of sales; at the best American companies, the figure was 1.8%; at the worst 5.2%.

Exhibit III Quality and costs

In theory, low warranty costs might be offset by high expenditures on defect prevention: a company could spend enough on product pretesting or on inspecting assembled units before shipment to wipe out any gains from improved warranty costs. Figures on the total costs of quality, however, which include expenditures on prevention and inspection as well as the usual failure costs of rework, scrap, and warranties, lead to the opposite conclusion. In fact, the total costs of quality incurred by Japanese producers were less than one-half the failure costs incurred by the best U.S. companies.

The reason is clear: failures are much more expensive to fix after a unit has been assembled than before. The cost of the extra hours spent pretesting a design is cheap compared with the cost of a product recall; similarly, field service costs are much higher than those of incoming inspection. Among manufacturers of room air conditioners, the Japanese—even with their strong commitment to design review, vendor selection and management, and in-process inspection—still have the lowest overall quality costs.

Nor are the opportunities for reduction in quality costs confined to this industry alone. A recent survey¹ of U.S. companies in ten manufacturing sectors found that total quality costs averaged 5.8% of sales—for a $1 billion corporation, some $58 million per year primarily in scrap, rework, and warranty expenses. Shaving even a tenth of a percentage point off these costs would result in an annual saving of $1 million.

Other studies, which use the PIMS data base, have demonstrated a further connection among quality, market share, and return on investment.² Not only does good quality yield a higher ROI for any given market share (among businesses with less than 12% of the market, those with inferior product quality averaged an ROI of 4.5%, those with average product quality an ROI of 10.4%, and those with superior product quality an ROI of 17.4%); it also leads directly to market share gains. Those businesses in the PIMS study that improved in quality during the 1970s increased their market share five to six times faster than those that declined—and three times faster than those whose quality remained unchanged.

The conclusion is inescapable: improving product quality is a profitable activity. For managers, therefore, the central question must be: What makes for successful quality management?

Sources of Quality

Evidence from the room air conditioning industry points directly to the practices that the quality leaders, both Japanese and American, have employed. Each of these areas of effort—quality programs, policies, and attitudes; information systems; product design; production and work force policies; and vendor management—has helped in some way to reduce defects and lower field failures.

Programs, policies & attitudes

The importance a company attaches to product quality often shows up in the standing of its quality department. At the poorest performing plants in the industry, the quality control (QC) manager invariably reported to the director of manufacturing or engineering. Access to top management came, if at all, through these go-betweens, who often had very different priorities from those of the QC manager. At the best U.S. companies, especially those with low service call rates, the quality department had more visibility. Several companies had vice presidents of quality; at the factory level each head of QC reported directly to the plant manager. Japanese QC managers also reported directly to their plant managers.

Of course, reporting relationships alone do not explain the observed differences in quality performance. They do indicate, however, the seriousness that management attaches to quality problems. It's one thing to say you believe in defect-free products, but quite another to take time from a busy schedule to act on that belief and stay informed. At the U.S. company with the lowest service call rate, the president met weekly with all corporate vice presidents to review the latest service call statistics. Nobody at that company needed to ask whether quality was a priority of upper management.

How often these meetings occurred was as important as their cast of characters. Mistakes do not fix themselves; they have to be identified, diagnosed, and then resolved through corrective action. The greater the frequency of meetings to review quality performance, the fewer undetected errors. The U.S. plants with the lowest assembly-line defect rates averaged ten such meetings per month; at all other U.S. plants, the average was four. The Japanese companies reviewed defect rates daily.

Meetings and corrective action programs will succeed, however, only if they are backed by genuine top-level commitment. In Japan, this commitment was deeply ingrained and clearly communicated. At four of the six companies surveyed, first-line supervisors believed product quality—not producing at low cost, meeting the production schedule, or improving worker productivity—was management's top manufacturing priority. At the other two, quality ranked a close second.

The depth of this commitment became evident in the Japanese practice of creating internal consumer review boards. Each of the Japanese producers had assembled a group of employees whose primary function was to act as typical consumers and test and evaluate products. Sometimes the products came randomly from the day's production; more frequently, they represented new designs. In each case, the group had final authority over product release. The message here was unmistakable: the customer—not the design staff, the marketing team, or the production group—had to be satisfied with a product's quality before it was considered acceptable for shipment.

By contrast, U.S. companies showed a much weaker commitment to product quality. At 9 of the 11 U.S. plants, first-line supervisors told me that their managers attached far more importance to meeting the production schedule than to any other manufacturing objective. Even the best performers showed no consistent relationship between failure rates and supervisors' perceptions of manufacturing priorities.

What commitment there was stemmed from the inclusion (or absence) of quality in systems of performance appraisal. Two of the three companies with the highest rates of assembly-line defects paid their workers on the basis of total output, not defect-free output. Is it any wonder these employees viewed defects as being of little consequence? Not surprisingly, domestic producers with low failure rates evaluated both supervisors and managers on the quality of output—supervisors, in terms of defect rates, scrap rates, and the amount of rework attributable to their operations; managers, in terms of service call rates and their plants' total costs of quality.

These distinctions make good sense. First-line supervisors play a pivotal role in managing the production process, which is responsible for internal failures, but have little control over product design, the quality of incoming materials, or other factors that affect field performance. These require the attention of higher level managers, who can legitimately be held responsible for coordinating the activities of design, purchasing, and other functional groups in pursuit of fewer service calls or reduced warranty expenses.

To obtain consistent improvement, a formal system of goal setting is necessary.³ Only three U.S. plants set annual targets for reducing field failures. Between 1978 and 1981, these three were the only ones to cut their service call rates by more than 25%; most of the other U.S. plants showed little or no change. All the Japanese companies, however, consistently improved their quality—in several cases, by as much as 50%—and all had elaborate companywide systems of goal setting.

From the corporate level at these companies came vague policy pronouncements ("this year, let the customer determine our quality"), which were further defined by division heads ("reduced service call rates are necessary if we are to lower costs") and by middle managers ("compressor failures are an especially serious problem that must be addressed"). Actual quantitative goals ("improve compressor reliability by 10%") were often set by foremen or workers operating through quality control circles. The collaborative nature of this goal-setting process helped these targets earn wide support.

At the final—or first—level of goal setting, specificity matters. Establishing an overall target for an assembly-line defect rate without specifying more detailed goals by problem category, such as leaks or electrical problems, is unlikely to produce much improvement. A number of U.S. plants have tried this approach and failed. Domestic producers with the lowest defect rates set their overall goals last. Each inspection point along the assembly line had a target of its own, which was agreed on by representatives of the quality and manufacturing departments. The sum of these individual targets established the overall goal for the assembly line. As a result, responsibility for quality became easier to assign and progress easier to monitor.

Information systems

Successful monitoring of quality assumes that the necessary data are available, which is not always true. Without specific and timely information on defects and field failures, improvements in quality are seldom possible. Not surprisingly, at the poorest U.S. companies information on defects and field failures was virtually nonexistent. Assembly-line defects and service call rates were seldom reported. "Epidemic" failures (problems that a large proportion of all units had in common) were widespread. Design flaws remained undetected. At one domestic producer, nearly a quarter of all 1979–1981 warranty expenses came from problems with a single type of compressor.

Other companies reported more extensive quality information—daily and weekly defect rates as well as quarterly and, occasionally, monthly service call rates. These variations in the level of reporting detail correlated closely with differences in quality performance. Among average U.S. performers, for example, quality reports were quite general. Data on assembly line defects gave no breakdowns by inspection point; data on field failures were for entire product lines, not for separate models. Without further refinement, such data cannot isolate quality problems.

A 10% failure rate for a product line can mean a number of things: that all models in the line fail to perform 10% of the time, with no single problem standing out; that several models have a 5% failure rate and one a 30% rate, which suggests a single problem of epidemic proportions; or anything in between. There is no way of distinguishing these cases on the basis of aggregate data alone. What is true of goal setting is equally true of reporting systems: success requires mastering the details.

The best U.S. companies reported defect rates for each inspection point on the assembly line and field failure rates by individual model. The Japanese not only collected information that their U.S. counterparts ignored, such as failure rates in the later years of a product's life; they also insisted on extreme precision in reporting. At one company, repairmen had to submit reports on every defective unit they fixed. In general, it was not unusual for Japanese managers to be able to identify the 30 different ways in which Switch X had failed on Model Y during the last several years. Nor did they have to wait for such information.

Service call statistics in the United States took anywhere from one month to one year to make the trip from the field to the factory; in Japan, the elapsed time averaged between one week and one month. Differences in attitude are part of the explanation. As the director of quality at one Japanese company observed, field information reached his company's U.S. subsidiaries much more slowly than it did operations in Japan—even though both employed the same system for collecting and reporting data.

Product design

Room air conditioners are relatively standardized products. Although basic designs in the United States have changed little in recent years, pressures to improve energy efficiency and to reduce costs have resulted in a stream of minor changes. On the whole, these changes have followed a common pattern: the initiative came from marketing; engineering determined the actual changes to be made and then pretested the new design; quality control, manufacturing, purchasing, and other affected departments signed off; and, where necessary, prototypes and pilot production units were built.

What did differ among companies was the degree of design and production stability. As Exhibit IV indicates, the U.S. plants with the lowest failure rates made far fewer design changes than did their competitors.

Exhibit IV Quality and product stability

Exhibit IV conveys an important message. Variety, at least in America, is often the enemy of quality. Product proliferation and constant design change may keep the marketing department happy, but failure rates tend to rise as well. By contrast, a limited product line ensures that workers are more familiar with each model and less likely to make mistakes. Reducing the number of design changes allows workers to devote more attention to each one. Keeping production level means less reliance on a second shift staffed by inexperienced employees.

The Japanese, however, have achieved low failure rates even with relatively broad product lines and rapidly changing designs. In the room air conditioning industry, new designs account for nearly a third of all models offered each year, far more than in the United States. The secret: an emphasis on reliability engineering and on careful shakedowns of new designs before they are released.

Reliability engineering is nothing new; it has been practiced by the aerospace industry in this country for at least 20 years. In practice, it involves building up designs from their basic components, determining the failure probabilities of individual systems and subsystems, and then trying to strengthen the weak links in the chain by product redesign or by incorporating more reliable parts. Much of the effort is focused up front, when a product is still in blueprint or prototype form. Managers use statistical techniques to predict reliability over the product's life and subject preliminary designs to exhaustive stress and life testing to collect information on potential failure modes. These data form the basis for continual product improvement.

Only one American maker of room air conditioners practiced reliability engineering, and its failure rates were among the lowest observed. All of the Japanese companies, however, placed considerable emphasis on these techniques. Their designers were, for example, under tremendous pressure to reduce the number of parts per unit; for a basic principle of reliability engineering is that, everything else being equal, the fewer the parts, the lower the failure rate.

Japanese companies worked just as hard to increase reliability through changes in components. They were aided by the Industrial Engineering Bureau of Japan's Ministry of International Trade and Industry (MITI), which required that all electric and electronic components sold in the country be tested for reliability and have their ratings on file at the bureau. Because this information was publicly available, designers no longer needed to test components themselves in order to establish reliability ratings.

An emphasis on reliability engineering is also closely tied to a more thorough review of new designs before units reach production. American manufacturers usually built and tested a single prototype before moving to pilot production; the Japanese often repeated the process three or four times.

Moreover, all affected departments—quality control, purchasing, manufacturing, service, and design engineering—played an active role at each stage of the review process. American practice gave over the early stages of the design process almost entirely to engineering. By the time other groups got their say, the process had gained momentum, schedules had been established, and changes had become difficult to make. As a result, many a product that performed well in the laboratory created grave problems on the assembly line or in the field.

Production & work force policies

The key to defect-free production is a manufacturing process that is "in control"—machinery and equipment well maintained, workplaces clean and orderly, workers well trained, and inspection procedures suited to the rapid detection of deviations. In each of these areas, the Japanese were noticeably ahead of their American competitors.

Training of the labor force, for example, was extensive, even for employees engaged in simple jobs. At most of the Japanese companies, preparing new assembly-line workers took approximately six months, for they were first trained for all jobs on the line. American workers received far less instruction (from several hours to several days) and were usually trained for a single task. Not surprisingly, Japanese workers were much more adept at tracking down quality problems originating at other work stations and far better equipped to propose remedial action.

Instruction in statistical quality control techniques supplemented the other training offered Japanese workers. Every Japanese plant relied heavily on these techniques for controlling its production process. Process control charts, showing the acceptable quality standards of various fabrication and assembly-line operations, were everywhere in general use. Only one U.S. plant—the one with the lowest defect rate—had made a comparable effort to determine the capabilities of its production process and to chart the results.

Still, deviations will occur, and thorough and timely inspection is necessary to ferret them out quickly. Japanese companies therefore employed an inspector for every 7.1 assembly-line workers (in the United States the ratio was 1:9.5). The primary role of these inspectors was to monitor the production process for stability; they were less "gatekeepers," weeding out defective units before shipment, than providers of information. Their tasks were considered especially important where manual operations were common and where inspection required sophisticated testing of a unit's operating characteristics.

On balance, then, the Japanese advantage in production came less from revolutionary technology than from close attention to basic skills and to the reduction of all unwanted variations in the manufacturing process. In practice, this approach can produce dramatic results. Although new model introductions and assembly-line changeovers at American companies boosted defect rates, at least until workers became familiar with their new assignments, Japanese companies experienced no such problems.

Before every new model introduction, Japanese assembly-line workers were thoroughly trained in their new tasks. After-hours seminars explained the product to the work force, and trial runs were common. During changeovers, managers kept workers informed of the models slated for production each day, either through announcements at early morning meetings or by sending assembled versions of the new model down the line 30 minutes before the change was to take place, together with a big sign saying "this model comes next." American workers generally received much less information about changeovers. At the plant with the highest defect rate in the industry, communication about changeovers was limited to a single small chalkboard, listing the models to be produced each day, placed at one end of the assembly line.

The Japanese system of permanent employment also helped to improve quality. Before they are fully trained, new workers often commit unintentional errors. Several American companies observed that their workers' inexperience and lack of familiarity with the product line contributed to their high defect rates. The Japanese, with low absenteeism and turnover, faced fewer problems of this sort. Japanese plants had a median turnover of 3.1%; the comparable figure for U.S. plants was two times higher. Even more startling were the figures on absenteeism: a median of 3.1 % for American companies and zero for the Japanese.

In addition, because several of the U.S. plants were part of larger manufacturing complexes linked by a single union, they suffered greatly from "bumping." A layoff in one part of the complex would result in multiple job changes as workers shifted among plants to protect seniority rights. Employees whose previous experience was with another product would suddenly find themselves assembling room air conditioners. Sharp increases in defects were the inevitable result.

Vendor management

Without acceptable components and materials, no manufacturer can produce high-quality products. As computer experts have long recognized, "garbage in" means "garbage out." Careful selection and monitoring of vendors is therefore a necessary first step toward ensuring reliable and defect-free production.

At the better U.S. companies, the quality department played a major role in vendor selection by tempering the views of the engineering ("do their samples meet our technical specifications") and purchasing ("is that the best we can do on price") departments. At the very best companies, however, purchasing departments independently ranked quality as their primary objective. Buyers received instruction in the concepts of quality control; at least one person had special responsibility for vendor quality management; goals were set for the quality of incoming components and materials; and vendors' shipments were carefully monitored.

Purchasing departments at the worst U.S. companies viewed their mission more narrowly: to obtain the lowest possible price for technically acceptable components. Site visits to new vendors were rarely made, and members of the purchasing department seldom got involved in the design review process. Because incoming inspection was grossly understaffed (at one plant, two workers were responsible for reviewing 14,000 incoming shipments per year), production pressures often allowed entire lots to come to the assembly line uninspected. Identification of defective components came, if at all, only after they had been incorporated into completed units. Inevitably, scrap and rework costs soared.

In several Japanese companies incoming materials arrived directly at the assembly line without inspection. New vendors, however, first had to pass rigorous tests: their products initially received 100% inspection. Once all problems were corrected, sampling inspection became the norm. Only after an extended period without a rejection were vendors allowed to send their products directly to the assembly line. At the first sign of deterioration in vendor performance, more intensive inspection resumed.

In this environment, inspection was less an end in itself than a means to an end. Receiving inspectors acted less as policemen than as quality consultants to the vendor. Site visits, for example, were mandatory when managers were assessing potential suppliers and continued for as long as the companies did business together. Even more revealing, the selection of vendors depended as much on management philosophy, manufacturing capability, and depth of commitment to quality as on price and delivery performance.

Closing the Gap

What, then, is to be done? Are American companies hopelessly behind in the battle for superior quality? Or is an effective counterattack possible?

Although the evidence is still fragmentary there are a number of encouraging signs. In 1980, when Hewlett-Packard tested 300,000 semiconductors from three U.S. and three Japanese suppliers, the Japanese chips had a failure rate one-sixth that of the American chips. When the test was repeated two years later, the U.S. companies had virtually closed the gap. Similar progress is evident in automobiles. Ford's Ranger trucks, built in Louisville, Kentucky, offer an especially dramatic example. In just three years, the number of "concerns" registered by the Louisville plant (the automaker's measure of quality deficiencies as recorded at monthly audits) dropped to less than one-third its previous high. Today, the Ranger's quality is nearly equal that of Toyota's SR5, its chief Japanese rival.

But in these industries, as with room air conditioners, quality improvement takes time. The "quick fix" provides few lasting gains. What is needed is a long-term commitment to the fundamentals—working with vendors to improve their performance, educating and training the work force, developing an accurate and responsive quality information system, setting targets for quality improvement, and demonstrating interest and commitment at the very highest levels of management. With their companies' futures on the line, managers can do no less.

1. "Quality Cost Survey" Quality, June 1977, p. 20.

2. Sidney Scoeffler, Robert D. Buzzell, and Donald F. Heany, "Impact of Strategic Planning on Profit Performance," HBR March–April 1974, p. 137; and Robert D. Buzzell and Frederik D. Wiersema, "Successful Share-Building Strategies," HBR January–February 1981, p. 135.

3. For a summary of evidence on this point, see Edwin A. Locke et al., "Goal Setting and Task Performance: 1969–1980," Psychological Bulletin, vol. 90, no. 1, p. 125

A version of this article appeared in the September 1983 issue of Harvard Business Review.

What Is a Single Line of Poetry Called

Source: https://hbr.org/1983/09/quality-on-the-line

Share :