Toyota Kata : Managing People for Improvement, Adaptiveness and Superior Results Read online

  Figure 4-3. Cross section of Ford Highland Park six-story buildings, 1914

  The concept behind these unique buildings was that final assembly is on the ground floor, and subassembly and parts fabrication processes are on the upper floors. In those days, materials were brought to factories by rail, and as you can see in the elevation drawing, railroad tracks went down the center craneway of the building. In the craneway, material would be hoisted from the railcars onto balconies that opened to the appropriate floors.

  At this point I’ll let Mr. Edward Gray, Ford’s chief construction engineer at the time, describe the rest of the material flow in these buildings, which he designed:

  There are thousands of holes cut through the various floors of those buildings, so that the parts that started in the rough on the top floor gravitated down, possibly through chutes or possibly through conveyors or tubes, and finally became a finished article, well down on the ground floor; landed on the conveyor at the ground floor.3

  In doing some secondary research for this book, my colleague, Jim Huntzinger, and I became fascinated with Ford’s six-story buildings at Highland Park, and this statement in particular. Once we had read Edward Gray’s testimony in the Ford Tax Case files at the Detroit Public Library, it seemed that the assertion of Ford’s Model T era engineers striving for one contiguous flow could only be confirmed by seeing for ourselves the holes in the floors of the six-story buildings.

  Imagine how disappointed we were when we could not find even one hole in the floors as our Ford hosts kindly walked us through the now-unused six-story buildings. Fortunately, we had an astute University of Michigan Ph.D. student with us, Eduardo Lander, who suddenly realized, “These floors are 90 years old and have probably been resurfaced many times. We should be looking at the ceilings, not the floors.” And as we looked up, there they were, plain as day, lots of patched holes.

  Ford’s six-story building experiment was ultimately not a success and the concept did not spread. We can speculate that the two cranes in each craneway—for unloading materials from the railcars—would have been a serious flow bottleneck. Transferring parts through holes in a reinforced concrete floor must also have been quite inflexible, since changing a machine layout could mean having to patch one hole and jackhammering open a new one.

  There was also still plenty of work-in-process inventory in the Highland Park value streams; in all the different conveyors, chutes, slides, barrels, etc., transferring components between processes, and often between individual processing steps within one process too. Ford was still a long way from the ideal of a 1×1 flow from A to Z, but that misses this key point: whether consciously or not, by striving to continuously improve the production flow toward an ideal of one connected flow, the early Ford Motor Company was utilizing a vision and interim target conditions in a way that highlights critical obstacles and makes them something to be worked through rather than circumvented. This is surprisingly similar to how Toyota’s improvement kata utilizes a long-term vision and interim target conditions to manage people and move the organization forward (Figure 4-4). Ford’s story has been told many times, but from a management and organization behavior perspective, we have missed this point.

  Figure 4-4. Early Ford utilized a vision and target conditions similar to the way that Toyota does

  End of the Flow Experiments

  After the six-story buildings Ford made one more big attempt to connect all processes from raw material to finished product, at the integrated, horizontal layout River Rouge factory complex. The Model T completed its production run there from 1919 to 1927. But by the mid 1920s customers were less willing to keep buying the same Model T. The number of different product variations began increasing, while the lifespan of any one model decreased.

  These two new demands on the factories—higher variety and shorter product life span—made it much more difficult to try to synchronize the production flow compared to the one-product Model T days. Some processes in the value stream now had to produce different versions of an item and change over between them. Say, for example, a crankshaft machining process that had produced only one crankshaft for the Model T would now have to produce a few different crankshafts for a few different engine variants. Ideally this machining process would change over at the same time the engine assembly process changes over, in sync, but this is difficult because a machining area often feeds more than one assembly process and has significantly longer changeover times.

  In this situation there are two basic options. The challenging option is to continue pursuing the “one contiguous flow” vision. This requires tackling and working through the admittedly difficult obstacles to a connected, synchronized flow and developing new solutions. The easier and quicker option, on the other hand, is to move away from the synchronized flow ideal, decouple the processes in the value stream from one another and operate them as islands.

  Generally speaking, after the Model T, manufacturers increasingly chose the decoupling option. Besides the increasing product variety, another reason for the move away from pursuing the flow ideal may have been that, around 1924, the production capacity of the U.S. automobile companies finally began to match the demand level. Orders were no longer outpacing capacity, and this conceivably reduced the urgency to keep striving for further flow and productivity improvements.

  Another reason was that General Motors struck out in a new direction with its new management approach, and it, no longer Ford, became the company to emulate. As the Model T era came to a close, it seems that so did focused experiments to keep improving factory flow, and the associated improvement kata style behavior. Pursuit of the one contiguous flow ideal went dormant again, until Toyota took up the mantle in the 1950s.

  The General Motors Approach (1920s to Present)

  A New Direction in Management

  Early Ford put emphasis on and effort toward a vision that described a condition—the production flow ideal—but ultimately focused too little on product development and on organizing and managing the company in systematic ways. In contrast, General Motors (GM) put a lot of attention on developing systematic management and structuring the organization. Three concepts from GM’s then new management approach pertain in particular to our discussion here. They should look familiar to anyone who has worked in a medium- or large-sized company.

  Rate-of-Return for Decision Making

  The GM financial committee relied on a rate-of-return analysis (cost-benefit analysis or return-on-investment calculation) for decision making on investments. The predicted return determined the choices that were made, as opposed to early Ford’s idea to do what is necessary to pursue an ideal.

  In other words, make money became the guiding vision or overall direction for further development of the business or the factory. We were now not moving in a particular direction (aiming at successive target conditions on the way to a vision) but rather judging and selecting options independently based on their rate of return.

  No other financial principle with which I am acquainted serves better than rate of return as an objective aid to business judgment. . . .

  We are not in the business of making cars, we are in the business of making money.

  — Alfred P. Sloan, Jr., President of General Motors, 1923–37; Chief Executive, 1937–46; Chairman of the Board, 1937–564

  Maximizing the Output of Individual Processes

  Early GM seems to have concluded that low costs are achieved when large quantities are produced with high machine utilization. Management began to think of the production value stream in terms of separate segments or departments, viewing each as an island, and created incentives that led those departments to produce as much as possible as fast as possible in order to reduce cost according to managerial accounting calculations (pieces per man hour per department or segment of the value stream).

  Centralized Planning and Control Based on Managerial Accounting Data

  GM introduced a decentralized divisional operating organization
, but, increasingly, with centralized operational decision making and control. That control was based on setting quantitative targets for the divisions and reporting back performance metrics from the divisions. Decision making was based heavily on analysis of reported managerial accounting data.

  Of course, GM also introduced well-known practices to influence the consumption side of the equation. These included segmenting the consumer market and providing each segment with a product line, an annual model change, segment-specific marketing, and providing credit to consumers. Since this book is about organization management, I will concentrate on changes GM introduced inside the company, on the management side of the equation.

  Intended and Unintended Effects

  The results of General Motors’ new approach and practices were dramatic and positive. GM achieved phenomenal success, grew to be the world’s largest corporation, and greatly influenced the nature of business management. Over the following decades GM’s management approach was widely publicized and was adopted by countless other companies. By the 1950s it had become general practice at U.S. corporations and at companies around the world. Today it is so pervasive that it is essentially invisible. It is simply how things are done.

  I should add one qualification to the above paragraph however: GM’s managerial approach achieved great success in the market conditions that prevailed through the 1960s. In later years, under different conditions, the same management approach no longer worked as successfully.

  Let’s take a look at some of the effects that those three GM concepts had on how companies are managed. Again, the following should look familiar to anyone who has worked in a manufacturing company.

  Effect of Rate-of-Return for Decision Making

  GM’s formula-based rate-of-return decision-making approach is effective enough in a growing market when there are business opportunities from which to choose, but it becomes less so in the crowded or low-growth marketplaces we have today.

  GM’s approach involved, to a degree, selecting between options in the early days of the U.S. automobile industry, when there were multiple options from which to choose. But in a lower-growth market with many competitors, the immediately profitable opportunities— the low hanging fruit—will have been picked. In this situation, management’s task becomes more one of nurturing promising processes, products, and situations into profitability than selecting ones that would be directly profitable.

  The ROI approach of General Motors is more about making choices than about improving and adapting. For example, in the second half of the twentieth century, Detroit automakers opted repeatedly to not significantly enter the market for small cars, even as that market grew noteworthy, because from an ROI-selection perspective it was not profitable. The media has often criticized these decisions, but that denunciation is at least partially misplaced. Executives were making those decisions rationally and correctly, in accordance with the management system within which they worked.

  In contrast, Toyota’s approach is about getting people to work systematically and creatively at the detail level to do what is necessary to achieve ambitious target conditions, which at first pass may not make it through a rate-of-return calculation. As shown in the previous chapter, Toyota utilizes cost benefit analysis less as a means for determining direction or what to do, and more as a means for figuring out how to cost-effectively achieve a desired condition.

  If we go even further with our ROI thinking and use it to evaluate individual decisions or steps, then the result is likely to be suboptimization. According to systems theory, trying to maximize the individual parts of something reduces the effectiveness of the whole.

  As we make these comparisons between GM and Toyota, we should keep in mind that it is not a judgment. The two approaches represent reactions to different conditions at different points in time in the history of the automobile industry. What’s most important is that we understand their long-term effects on an organization.

  Effect of Maximizing the Output of Individual Processes

  Seeking to maximize individual process output—for example, by measuring each process separately with a pieces per man hour calculation—generates the following effects on a value stream:

  A process or department becomes even more decoupled from the next process as it strives to produce as much as possible as fast as possible.

  Since changeovers interrupt production, there is a natural tendency to avoid them and produce large lots.

  The next process in the value stream does not yet need all those parts that were produced too soon, so the parts must be stored as in-process inventory. (Inventory which is, by the way, counted as an asset by the managerial accounting system.)

  When the next process finally does use the parts, it will discover defects among them. However, it is impossible to trace the root causes of those defects because the parts were produced some time ago, and the conditions in the preceding process that caused the defects have long since changed.

  This situation repeats over and over all the way through the value stream and results in a total lead time through the factory that is measured in days or weeks, whereas the total value added time is actually only minutes. Interestingly, when we speed up a process to improve its pieces-per-man-hour numbers, we only reduce the minutes of value-adding time and do nothing to reduce the days and weeks of lead time. You can observe these effects in factories around the world.

  To keep inventory from swelling too much in this situation, we started placing limits on inventory buffers and set targets for inventory levels, without necessarily understanding the actual situation in the factory processes. The goal then became trying to schedule each individual segment of the value stream so accurately that items would be made not long before the next segment actually needs them. But this holy grail is not consistently attainable in the real world, even with sophisticated software, because process conditions up and down the value stream are constantly changing.

  It takes a certain amount of inventory to hold a value stream together, and the quantity of inventory required depends on the current performance characteristics of the processes in that value stream. If we reduce inventory targets to below this level, then shortages, expediting, and emergency freight will increase. Every day’s work in the factory then involves adjusting schedules and expediting. Such daily adjustments in turn cause even more volatility in the value streams, and soon everyone in the factory becomes almost completely occupied with trying to make the production quantities and shipments.

  People in an organization act rationally in a way that maximizes their success. Putting the emphasis on departmental output maximization, rather than on optimizing the overall flow for the customer, means that the natural interests of the departmental manager may come into conflict with the long-term survival interests of the company. In the long run, overall cost will be higher and the organization will become so involved in firefighting that it is standing still, even though the departmental manager is meeting and even exceeding his or her objectives.

  To put it briefly, systems theory tells us that we cannot optimize a system by trying to maximize its individual parts.

  Effect of Centralized Planning and Control Based on Managerial Accounting Data

  As the above description of everyday life in a factory illustrates, with centralized decision making from a distance based on accounting data, management tends to lose connection with, and understanding of, the actual situation on the work floor. Trying to manage from a distance through data abstractions often results in managers making incorrect assumptions and inappropriate decisions, and trying to make adjustments and adaptations too long after the fact. In addition, on-site managers naturally try to make the numbers upon which they are evaluated look good, which means that even less accurate information is reaching the decision makers in the levels above.

  Not only are the centrally controlled divisions unable to adapt autonomously and quickly, but the decision makers in the central office are basing t
heir decisions on inaccurate, after-the-fact quantitative abstractions.

  What Happened to Management By Objectives?

  The original thinking behind management by objectives (MBO), as outlined by Peter Drucker in his 1954 book The Practice of Management, is not too distant from how Toyota is managing. Drucker even mentions, in a short case example, how what he calls “some of the most effective managers I know” go beyond only deploying quantitative targets downward. He briefly describes how these managers engage in a two-way dialogue with the level below them in order to develop written plans for the activities that will be undertaken to reach the targets. In other words, paying attention to the means that are utilized to achieve the results.5

  It appears, however, that in subsequent actual business practice and education, MBO became something more like planning and control from above executed to a large degree by setting quantitative targets and assessing reports of metrics. Some call this “management by results.” Unfortunately, there are plenty of different ways to achieve a quantitative outcome target, many of which have nothing to do with making real process improvement and moving the pieces of an organization in a common direction.

  So why did a watered-down version of MBO work so well for us for so long? Here are some possible reasons:

  In the period of limited international competition and continued growth, which ran until the 1970s, occasional improvement was good enough. In such market conditions it is possible to make a good profit even if there is considerable waste in the system and we are not continually improving.