Cars Per Employee And Productivity At Volkswagen Versus Toyota

Seen this morning in a Lean consultant’s blog:

“Two decades later, VW has topped Toyota as the world’s number one automaker, but Toyota generally is considered to be […] far more productive. In 2015, VW employs 600,000 people to produce 10 million cars while Toyota employs 340,000 to produce just under 9 million cars…”

Is it really that simple? VW produces 10 million/600,000 = 16.67 cars/employee/year, and Toyota 9 million/340,000 = 26.47 cars/employee/year. Ergo, Toyota is 60% more productive than VW — that is, if you accept cars/employee/year as an appropriate metric of productivity.  Unfortunately, it is a bad metric that can easily be gamed by outsourcing.

Continue reading

Using videos to improve operations | Part 7 – Detailed review of process segments

Asenta 2011-03 Roberto Cortés

Roberto Cortés

Asenta Juan Ortega head shot

Juan Ortega

This post was co-written with Asenta’s Roberto Cortés and Juan Ortega, based on a joint project in Spain in October, 2013. A detailed analysis of the video recordings on two operations was key to generating improvement ideas that the plant has implemented since. The company had shot some videos of operations before, but not used them this way before, and it was a learn-by-doing experience for the participants. 

The demand for the company’s products is growing, and it is struggling to keep up. Its core technology is a fabrication process, and engineering has focused its attention on it to increase capacity. After fabrication, however, the product needs several assembly operations. From direct observation, it was clear that the operators were working at a pace that could not be sustained for a whole shift. The manager confirmed that the pace slackened and the quality dropped towards the end of the shift.

The challenge was therefore to change the assembly process so that the operator could complete the tasks within the takt time of about 60 seconds, at a steady, sustainable pace, without running ragged or getting exhausted. While on site, we focused on two operations, shot videos as recommended in earlier posts — from an elevated position and focusing on the operator’s hands — and coached the plant team on reviewing the videos, with the goal of enabling them to do it on their own for the other operations.

1. Preparation

The detailed review breaks the operation down into its smallest identifiable steps  to discover improvement opportunities for each. If you are going to do this on a regular basis, you should probably invest in software to collect timestamps from videos, categorize the steps, and record improvement ideas, like Timer Pro or Dartfish.  Timer Pro was developed specifically for Manufacturing; Dartfish, for sports, but it has also been used in Manufacturing.

For the first time, it is best to do it on short operations, and you can make do with an Excel spreadsheet on which you manually record the timestamps. It needs the following columns:

  • Step number
  • End time
  • Step duration, calculated from the timestamps.
  • Cumulative time, aggregated from step durations.
  • Operation Description
  • Operation Category
  • Improvement Ideas

Sufficient time has to be allowed for the detailed review. It is customary to allow between 3 and 5 times the length of the recording and even more if the recording is very short. It is recommended to have a sample of the product and components at hand where the review is being held.

Asenta----Product-sample-in-conference-room

Product sample in conference room

2. Review

The video is analysed and the spreadsheet completed step by step. For short steps, you can play the video in slow motion  to give time to observe details. Because you are going to be adding times, you need record the timestamps at a higher precision than you are really interested in. For example, to analyze time in second, you need to record the timestamps to one tenth of a second. The video and the form are shown on the screen at the same time.

Asenta----Video-and-analysis-form-on-same-screen

Video and analysis form on the same screen

While conducting the analysis, do the following:

  • Describe each step with an action verb and a single object. If you find you can’t, break it down further until you can.
  • Do not criticize ideas. Write them down for later evaluation.
  • Aim to eliminate unnecessary steps (muda), reduce the variability in how the steps are carried out (mura) and their inconvenience (muri).
  • Assign a category to each step so that you can aggregate times by category.

You can generate your own categories as you go along and standardize them as you reach conclusions. There must not be too many (5 better than 10) and they are usually of the following type:

  • Pick up/put down
  • Walk
  • Assemble
  • Inspect or test
  • Wait
  • Adjust
  • Rework
  • ….

If there are large differences in how different operators perform the operation, several videos can be screened at the same time, with the same task carried out by different operators. It is essential to carry out this detailed review with the operators in the videos. They know things that nobody else knows, and have ideas that you want to use.

Asenta -- Operators participating in analysis

Operators participating in the analysis of their own work

3. Conclusions

When you analyze operations for the first time, it is common to discover that about 40% of the time is spent on activities other than assembly or test. This is due to a combination of wrong sequencing, redundant steps, multiple handling, inadequate fixtures, inconveniently located tools or parts, etc.

Of course, not all of these can be eliminated easily. Some can be, by redesigning or retrofitting the work station; others can be taken out of the assembly flow and performed in parallel so that, for example, the operator does not have to prepare a part while the product waits. The net productivity increase that can usually be accomplished is on the order of 30%, without overburdening the operator. In our client’s case, this means making the assembly jobs sustainable while absorbing a higher demand.

Once the summary of times by category has shown the “gold in the mine” — that is, the improvement potential, the team fleshes out the ideas generated during the review of the video, tries them out as much as possible immediately, and turns them into proposals. The following pictures shows the flip chart with sketches of the proposals generated in our sessions, and a snapshot of try-storming.

The team then turns the  improvement proposals into a detailed action plan for the short, medium-, and long term.

Once the improvements are implemented, the team shoots another video of the operation, for the following purposes:

  • Validating the improvements.
  • Standardizing the sequence of operations
  • Training other operators

David Meier on Productivity Metrics | The Lean Edge

See on Scoop.itlean manufacturing

davidmeier.thumbnail

The first point I want to make is that any measure has flaws and will not completely reflect reality. They should be considered indicators and in some way all refer to some sort of “standard” or desired condition. This is the basis for problem identification, which is the main purpose.

Any measure is a “snapshot” of conditions during a specific time period and reflects many variables that are occurring. Some measures such as productivity are based on assumptions such as standard hours. The notion of standard hours is flawed in many ways that I won’t get into, but this measure can be used effectively (if used carefully). The mistake that is often made is to evaluate performance based on the measure and to drive inappropriate behaviors like overproduction.

 

Michel Baudin‘s insight:

Insights on productivity metrics used at Toyota.

See on theleanedge.org

Comparative advantage, free trade, and productivity

In a recent post in Evolving Excellence, Bill Waddell pointed out that the economic theory of comparative advantage was wrongfully used to justify outsourcing to low wage countries. The way Bill puts it: “…the theory is based on productivity – not hourly wages.  It is driven by the idea that goods should be made wherever the collective combination of hours results in the overall minimum consumption of human effort…” (http://bit.ly/Uy8yfg)

In an earlier post, Bill cited the economic report of the president in 2010 as defining comparative advantage as the idea, “that nations specialize in producing the goods that they can produce cheaply relative to other goods.” (http://bit.ly/UsT0aA)

David Ricardo

Adam Smith

The comments by readers of  Bill’s post seem to confuse Ricardo’s comparative advantage with the  Adam Smith’s division of labor.

They also tend to confuse low labor costs with low wages, when increasing productivity achieves both low labor costs and high wages. In the 1910s, when Ford started paying workers $5/day, twice the going rate in Detroit, it still had lower labor costs than competitors because its workers on assembly lines were four times more productive.

Ricardo’s own words are easy to check, because the Kindle edition of his Principles of Political Economy and Taxation costs $0.00 and the discussion of comparative advantage starts on p. 92. It is more a case study than a theory, in which two countries benefit from free trade in two products, even though one is more productive at making both than the other.

British cloth

Ricardo’s case is cloth and wine made either in Portugal or England. Both needed less labor to make in Portugal but wine required much less while cloth only slightly less. In this case England was said to have a comparative advantage  on cloth even though Portugal has an absolute advantage on both wine and cloth. The practical consequence is that more wine and cloth are produced overall if England focuses on cloth and Portugal on wine.

Port wine from Portugal

It sounds like a hypothetical example, contrived for the purpose of illustration, and that is what I first assumed it was. It sounded particularly far-fetched, writing in 1817, that the production of cloth should take more labor in industrial England than in Portugal. But it is not a made-up case. It is a real one that unfolded in the century before Ricardo wrote, before the industrial revolution. If we believe the following excerpt from the Wikipedia article about the history of Portugal, it happened as follows:

“The 1703 Methuen Treaty between England and Portugal had both direct and indirect effects on the Portuguese wine industry. The treaty not only stipulated that the amount of duties on Portuguese wines was to never be more than two-thirds that of which was levied on French wines, it also allowed English woolen cloth to be admitted into Portugal free of duty. This second stipulation ended up having a devastating effect on the Portuguese textile industry, leading to huge numbers of shepherds and weavers becoming unemployed. In and around the Douro region, this segment of labor turned to the wine industry and encouraged a boom in vineyard planting.”

Three centuries later, while British textiles are not what they used to be, port wine from Portugal can still be found in liquor stores worldwide. It owes its unique taste to its fortification for sea travel to England, and it is sold under English brand names like Croft or Sandeman, from the British trading houses that used to ship it.

It is a great story, but hardly enough to prove that free trade always benefits all participating countries. It is not easily generalized from two countries trading two products to many countries trading many products, and it is not obvious that it can be used to justify outsourcing. On the other hand, I have applied it in production to the allocation of work among machines with overlapping capabilities.

Ricardo talks about “the labor of 100 Englishmen” as the main measure of the resources consumed in making a product. In the 1700s, labor was the main resource used in production, but, in 2012, labor is only one of many, the others including equipment, tooling, energy, and data, and a discussion focused exclusively on labor would not be appropriate.

So, what metric would you use to represent the total amount of resources used to make a product? The metrics used for this purpose are usually called “cost.” Their calculations involve debatable allocations that are not consistently made even within one country, and equally debatable exchange rates to make international comparisons. You would also use different metrics depending on whether you are deciding where to locate a new plant or how to allocate work among existing plants.

In using comparative advantage to allocate work among machines, the metric was the amount of machine time used per unit of product, which was relevant because we were trying to maximize production for a given mix of products. To minimize WIP instead, you would use a different metric, so as to penalize a machine for processing WIP-generating batches instead of single pieces.

When I first heard of comparative advantage, it was described to me as one of the few economic theories that is neither trivial nor false. And indeed it is, and I believe it makes Ricardo a precursor to today’s Operations Research and Game Theory, but we should be wary of jumping to overly broad conclusions about 21st-century globalization based on the trade of wine and cloth between England and Portugal 300 years ago.

Deming’s Point 8 of 14 – Drive out fear

(Photo by Lewis Wickes Hine, New York Public Library)

Deming’s complete statement of Point 8 is as follows:

“Drive out fear.”

This is a prescription that Doug Hiatt, a quality assurance manager at Boeing, found bewildering. First, he couldn’t see how fear could be “driven out,” and, second, where dangers are real, he didn’t feel that fear was something to be avoided. Deming is not arguing, however, that external threats, like competitors, should be hidden from employees to make them feel secure. In the 1980s, I worked for a software company whose managers were invariably friendly and courteous to subordinates, and where management communication was mostly “happy talk” that made especially the younger employees feel comfortable. Then, overnight, one third of them were laid off. Their sense of security was false.

Deming is advocating giving employees a genuine sense of security, which is both difficult to create and easy to shatter. Nothing can create such as sense quickly, but we can think of all sorts of human resource policies that can have this effect if carried out consistently over many years. Deming does not give us any pointer, but, in the US in 2012, few companies even try, particularly in environments like Silicon Valley.

Deming feels that fear always leads to “impaired performance and padded figures.” While the fictional Darth Vader can scare a crew into building a fully operational death star faster, the record in the real world is mixed. There, the ultimate manager by fear was probably Joseph Stalin, as shown in his January, 1940 telegram to a plant manager telling him that, unless results were produced within a tight deadline, his management team would be shot. The performance of Soviet industry supports both of Deming’s assertions.

MachiavelliBut even in the US, managers like Jack Welch, who introduced Rank-and-Yank  at GE, clearly feel that there is nothing wrong with making employees fear losing their jobs. Others like to quote Machiavelli’s “It is better to be feared than loved, if you cannot be both.” But Machiavelli’s world in 15th century Italy was more like the Game of Thrones than a contemporary manufacturing company. His prince is concerned exclusively with stabilizing his power, fending off rivals, and conquering more territory. Machiavelli’s advice is of limited value in areas like product development, marketing, manufacturing, or customer relationship management.

Intel’s Andy Grove was so famous for saying “Only the paranoid survive” that he wrote a book by this title, but the book is about business strategy, not about the way you treat employees. I had an extended project with Intel when Grove was its CEO; the Intel employees I worked with spoke of him with awe and respect, but never with fear. They trusted his steady hand steering the company and were not worried about being treated unfairly. Outside Intel, the company was perceived as secretive and aggressive, bordering on ruthless.

Does fear always impair performance? Stage fright can paralyze public speakers, stage actors or singers, but its complete absence is a sign of indifference to the audience that it doesn’t miss. The best performers are those who feel stage fright but are galvanized by it. Conversely, does the absence of fear always enhance performance? Academic tenure is the ultimate in job security. But do professors perform better once they are tenured than when they are on a tenure track pursuing it? Non-academics may be too quick to assume that they don’t. There is no valid general answer to that question. Some do and some don’t.

Deming sees a “widespread resistance to knowledge.” From the details he gives, what he means for individual contributors is that they are afraid new methods or new technology will make their hard-earned skills obsolete and threaten their positions; for managers, it is the worry that the investment in acquiring knowledge will never be recouped. These are two separate concerns.

The first fear is readily observed in organizations that hire people based on the immediate need for skills, as opposed to recruiting them for a career. If you know you are employed because you are the only one to know how to run a milling machine of a particular model, or navigate the user interface of a legacy information system, then you are naturally less than enthusiastic about the introduction of a way of working that requires you to train others to do your current job, or of new machines or systems that do not need your current skills. If company behavior over decades has built a foundation for this fear, you will not drive it out easily. It will require the establishment of new human resource policies, their communication to the work force, and their sustained practice over a long-enough period to build credibility with the work force

In operations, the managers’ primary responsibility is the output to customers, and employees do learn in the process of producing it, particularly if they rotate between stations. But even this form of knowledge acquisition is not free. It takes management attention to organize and monitor, each job an operator rotates into requires a learning period during which performance degrades, and there is always the risk that your most knowledgeable employees will leave. Other forms of knowledge acquisition include participation in improvement projects and experiments, technology watch, and formal training, in house and at public venues. All are investments in money and time, with  uncertain outcomes. Let us look at each in more detail:

  1. Improvement projects. They should always have the dual purpose of improving performance in their target area and learning by the work force. Participation in successful improvement projects develops both technical and managerial skills, in a way that pays for itself through the performance enhancements.
  2. Experiments. While experimentation is a normal part of product development, most managers do not make room for it in operations. A Lean Manufacturing plant, on the other hand,  sets aside space for it and encourage engineers or technicians to experiment with concepts, tools, machines, or systems that are  not  immediately applied in production. This is how they learn to be savvy buyers of technology, customize off-the-shelf equipment, or build from scratch machines that are not commercially available. You cannot write a discounted cash flow analysis to justify such an engineering sandbox, but you can see its impact in the proliferation of clever devices that enhance production performance on the shop floor.
  3. Technology watch. This is keeping up with new developments in one’s current specialty, by reading the trade press, attending conferences, visiting trade shows, and going on plant tours. These are activities that a manager may find difficult to justify, on the grounds that they are not anything a customer would be willing to pay for. Yet, not doing them is a sure path to technical obsolescence.
  4. Training. We discussed training issues in the review of Deming’s Point 6.

How do you “drive out” the fear of making the wrong decisions in this area?  This is particularly challenging when you break down functional silos and distribute technical specialists among the processes they serve, whose management owners rarely appreciate the need for them to stay current. If you are an extrusion engineer working for the production manager in a shop that makes extruded rubber parts for cars, you may be dedicated to making the lines perform well, but you will be isolated from professional peers. That is why some organizations either retain the functional silo structure while trying to make it work better using tools like A3 reports for better communication, or they adopt a matrix organization, in which the specialists maintain a “dotted line” reporting relationship to a technical manager whose job is to manage the maintenance and development of their skills.  A common strategy for IT in manufacturing companies is to outsource the technical work to a system integrator who is responsible for the technical skills of the contractors he sends.

Deming also describes as a loss from fear the inability to serve the best interest of the company because of rules or production quotas. It conjures up the image of Captain Queeg telling his officers how every rule in the book is there for a reason and has to be followed to the letter. Deming gives the example of a supervisor afraid to stop a machine for needed repairs because he might not fulfill his production quota. Of course, the machine breaks down and he can’t fulfill his quota anyway. But it is a dilemma. On the one hand, you want employees to use their judgement and break rules that are counterproductive. But, on the other hand, you don’t want them to think that shop operating standards and production plans are only guidelines. Finding the right balance is not easy between blind obedience to imperfect rules and absolute control by each individual of what to do and how much to produce. Here are a few pointers on how to do it:

  1. The rules have to be few in number and clearly stated. The following signs show the rules governing the use of a public park in Paris and a swimming pool in Palo Alto, CA. Every visitor to the Luxembourg gardens is expected to know nine articles of small print; the Palo Alto swimmers, seven bullet points.

  2. The purpose of the rules must be communicated, whether it  is regulatory compliance, safety, quality, etc. If no one can explain the purpose a rule serves, then it is a candidate for elimination.
  3. A process must exist to modify or cancel rules that are obsolete, ineffective, or counterproductive. The Accidental Office Lady is a memoir of Laura Kriska’s years at Honda in Japan. As a young American college grad, one rule she found particularly objectionable was the discriminatory requirement for women to wear a uniform at work. She recounts how she used Honda’s NH Circle system to organize a group of co-workers and make a case for the elimination of uniforms as an improvement in office work, and got it approved by Honda management.

How do you recognize the presence of fear in an organization? Deming lists 14 different types of statements that he has heard from employees and considers to be expressions of fear. Following is a summary of his list:

  1. The company may go out of business.
  2. Supportive superior may leave.
  3. Putting forth an idea may be perceived as treason.
  4. May not have a raise at next review.
  5. Long-term benefit may require short-term performance drop on daily report.
  6. May not be able to answer boss’s question.
  7. Credit for contribution may go to someone else.
  8. Admitting a mistake may have adverse consequences.
  9. Boss believes in fear; management is punitive.
  10. System will not allow expansion of abilities.
  11. Company procedures are not understood; employees don’t dare ask questions.
  12. Management is mistrusted, and perceived to have a hidden agenda.
  13. Inability to fulfill production quota (Operator or Plant Manager).
  14. No time to take a careful look at the work (Engineer)

Deming’s Point 7 of 14 – Institute Leadership

Deming’s complete statement of Point 7 is as follows:

“Institute leadership. The aim of supervision should be to help people and machines and gadgets to do a better job. Supervision of management is in need of overhaul, as well as supervision of production workers”

There are the following three parts to this statement:

  1. Institute leadership.
  2. The aim of supervision should be to help people and machines and gadgets to do a better job.
  3. Supervision of management is in need of overhaul, as well as supervision of production workers.

1. Part 1: Institute leadership

What does Deming mean by leadership, and how to you institute it? From Plato and Sun Tsu to Peter Drucker, many have written about leadership as a quality a person may possess, without ever precisely defining what it is. You see its effect in others’ willingness to follow, but no one has ever written a credible spec of what makes a leader. There is in particular no agreement on whether this quality is innate to a few individuals or can be nurtured in many. In The Practice of Management (pp. 158-160), for example, Drucker argues that leadership is innate, and that the best management can do in an organization is create conditions for the natural leaders in its ranks to emerge.

When Deming argues that the job of management is not supervision but leadership, he appears to be using leadership in the sense of this rare quality that escapes a precise definition. At the opening of Chapter 8, however, he writes:

“The aim of leadership should be to improve the performance of man and machine, to improve quality, to increase output, and simultaneously to bring pride of workmanship to people.”

In this context, he is referring to a role rather than a quality, and “management” could be substituted for “leadership” without changing the meaning. When you say “Xi is the new leader of China,” it describes Mr. Xi’s formal role, not his qualities.

2. Part 2: Supervision should help people, machines and gadgets do a better job.

In this light, Parts 2 of Deming’s Point 7 can be interpreted primarily as a recommendation on the role of first-line management in a factory. The name for this position has changed multiple times, from the 19th century Gang Boss, through the less brutal sounding 20th century Foreman, to the gender-neutral Supervisor, and now to the vague and fuzzy Group Coordinator or Area Coordinator. These repeated name changes for the same position suggest that it is an uncomfortable one, between the management hammer and the shop floor anvil. Top management and engineering titles have a longer shelf life.

Deming’s explanations  focused on what is wrong with supervising people whose work you don’t know, treating every problem as a special case, and managing by numbers. If being able to do the work of the people you manage is a requirement, then production supervisors will be exclusively drawn from the ranks of operators,  and this position will not be given to college grads on their first jobs. My experience, however, is that first-line supervision in production works best when the supervisory team includes members of both types, combining the book smarts of the college-educated rookies with the shop smarts of veteran operators. There may be tensions between the two but, if well managed, they can achieve together results that neither group could without the other.

Treating every problem as a special case is an easy trap to fall into, and it is what most managers do. Each problem they see is a line-item on their to-do list, they find a countermeasure, check it off and move on. The special cause is that we received a defective part last week, or the operator was new, or the cutting tool broke. But leaders should not be satisfied with such answers and dig deeper to consider whether the problem is a symptom of a problem with the process itself. In the Soviet Union, all problems had to be blamed on human error. Someone had to be made a scapegoat and punished. The idea that there might have been something wrong with the system was not allowed to be contemplated. Deming’s point here is that leaders must do exactly that.

In Deming’s view, it is because they don’t understand the work that supervisors fall back on managing by numbers. Even if you have no clue about the work of an operator, you can still count parts and, if your management only cares about the numbers, you end up doing nothing else. Deming’s perspective on managing by numbers in explained in Deming versus Drucker.

Underlying this discussion, but not said by Deming in son many words, is an underestimation of first-line management. In my experience, when backed by their superiors, production supervisors are the most powerful agents of change on the factory floor. Because they are part of management, support groups like Maintenance or Quality listen to them. They can work directly with operators as no one else in management can, and they are processes owners.

This combination of factors makes them uniquely effective as improvement project leaders. Deming  puts in their job description, which is necessary but not sufficient. Their area of responsibility must be small enough for them to have time to work on improvement. In the late Toyota-run NUMMI plant, a group leader in assembly had an average of 17 operators. Many other companies boast about having a “lean” management structure with one supervisor for 100 employees, who is too busy to do anything but minding daily numbers. Meanwhile, improvement is the purview of a specialized engineering department that has neither the resources needed to undertake all the necessary projects nor the rapport with the shop floor that is needed to make changes take root.

What does Deming mean by gadgets? We can assume that, when Deming says machines, he implicitly includes fixtures, jigs, and tooling under that term. Gadget is not a technical term, and Deming does not define it, but, except in Point 7, every use of it in his book is clearly derogatory:

“Lag in American productivity has been attributed in editorials and in letters in the newspapers to failure to install new machinery, gadgets, and the latest types of automation such as robots. Such suggestions make interesting reading and still more interesting writing for people that do not understand problems of
production.” p. 13

Among Obstacles, on p. 127: “The supposition that solving problems, automation, gadgets, and new machinery will transform industry. No one should sneer at savings of $800,000 per year, or even $1000 per year. A group of workers took pride in changes that saved $500 a year. Every net contribution to efficiency is important, however small.”

Gadgets and servomechanisms that by mechanical or electronic circuits guarantee zero defects will destroy the advantage of a beautiful narrow distribution of dimensions.” p. 141

In other words, whatever gadgets are, they are up to no good, so why would supervision worry about making them do a better job? Why not just get rid of them? One reason is that first-line managers usually do not choose the resources they have to work with, whether human or technical. They don’t choose to buy a particular gadget; their task is to use it as best they can.

3. Part 3: Supervision of management is in need of overhaul

This is clearly about the higher levels of management, but Deming’s elaboration on Point 7 says nothing on this matter. In higher-level positions, it is often impossible to find candidates with personal experience of the work of all their subordinates.  To take an extreme example, former presidents of the US are unanimous in saying that nothing could prepare you for that job.

We know that someone is a good leader by the readiness, willingness and enthusiasm of others to follow. Anyone can observe the behaviors of leaders and try to emulate them, but rarely to the same effect. Deming does not offer a theory or even a definition of what he means by leadership, but we know that he didn’t see much of it in American managers.

Deming’s Point 5 of 14 – Improve Constantly and Forever the System of Production and Service

Deming’s complete statement of Point 5 is as follows:

“Improve constantly and forever the system of production and service, to improve quality and productivity, and thus constantly decrease costs.”

At first sight, this point sounds exactly like the first one, which is about constantly improving products and services. What is the difference? Point 1 was about output; Point 5, about internal processes and systems. It says that improvement is an activity that must always be part of the life of any business organization. On this, Deming is on the same wavelength as Imai in Kaizen, which was published almost at the same time as Out of the Crisis.

“Constantly and forever” means that improvement in a plant starts on its opening day and ends only if it closes. Point 5 assumes that improvement is always possible, and should always be pursued. Imai had quoted a Japanese executive saying that he had found a US plant unchanged on his second visit after 30 years. Looking for examples of what this visitor might have seen, I found the following two pictures of coke ovens at the Ford River Rouge plant:

Figure 1. Identical operation 30 years apart

By contrast, a factory that practices improvement looks slightly different if you revisit it after six months and is unrecognizable after two years.

But the idea that you need to constantly improve a factory contradicts the conventional wisdom that it produces diminishing returns. One area of human endeavor where we might expect such diminishing returns is the 100m race, with athletes training ever harder to nibble ever smaller improvements in their times. It is what common sense tells us, but not what the data tells us. Figure 2 plots the world records in 100m racing from 1900 to 2010.

Figure 2. World records in 100m racing from 1900 to 2010

The linear trend is for the record to drop by an average 0.01 sec/year for 110 years, with no sign of a slowdown in improvement. Still, in manufacturing quality, you could claim that there are diminishing returns when the same amount of effort takes you for 30% to 3% defective, then from 3% to 0.3%, and then from 0.3% to 15ppm. In a competitive environment, however, the consequences of making or not making these improvements do not diminish. If you don’t make them, somebody else will and use them to take markets away from you.

In the 50s and 60s, some American appliance makers were not only failing to improve quality but were deliberately degrading it. Even though I had read reports of this, I still thought it so egregious that I didn’t believe it, until, in the 1980s, I met an engineer who had been personally involved. He had been part of a “reliability” department charged with redesigning products to fail as soon as possible after the warranty expired. This was a version of planned obsolescence that opened the door to competitors. Planned obsolescence still exists, but it is now about making customers want to buy a new product because it has new and better features, not because the old one broke down.

Deming does not mention the training value of improvement work. Improving the production system may become ever more challenging, but the work force that has taken it this far has learned lessons and grown skills that enable it to take on the next challenge. The problems may be harder, but the problem-solvers are stronger.

In his comments, Deming does not limit the size of the improvement actions. His recommendation isn’t just about what we now call continuous improvement or Kaizen. It is not just about small changes made to work methods by those who do the work. It can also be large-scale, radical changes. On the other hand, the only target of improvement he seems to have in mind is quality. As he describes it, if quality improve, so does productivity, because you eliminate the friction in the process caused by defects. Bringing processes under statistical control is front and center.

Lean Manufacturing goes much further. First, you cannot have flow lines with processes that are not under statistical control, and, if you have such processes, not much else matters besides bringing them under control. But there are many plants, in mature industries, where it is no longer an issue,  and machines, right out of the box, can hold tighter tolerances than required. In this case, Deming’s logic is turned on its head, and it is quality improvement that becomes a by-product of work on productivity.

For example, when you convert a batch production line to a one-piece-flow cell, the immediate effect is that you may see is that double productivity while reducing cycle time and WIP by 90%. Then, as you start operating the cell, you notice that it produces three times fewer defectives per shift than the old line did, essentially because, instead of burying defects in WIP, you detect them immediately. A part coming out of an operation is immediately loaded into the next one, which brings to light any defect it may have. This is a scenario that I have observed many times, but it is not part of Deming’s world view.

Today, you would never hear a manager openly oppose Deming on this. It has become part of the standard talking points but, if you listen closely, you hear different messages that contradict it, such as: “We’ve optimized production, and our big opportunity is now in the supply chain.” If you want to follow Deming’s advice, you should ban the word “optimal” from your vocabulary, because, by definition, if anything you do it optimized, you can no longer improve it. The completion of one improvement action sets the stage for the next one, forever; optimization, on the other hand, leads to a full stop. When you see the shop floor after hearing such a statement, you see plenty of opportunities that have been left on the table and are not being pursued.

There are still very few companies that genuinely pursue improvement “constantly and forever” at all levels of the organization, through all means available, including  individual suggestions, circle activities, Kaizen events, and large-scale innovation projects. We usually consider them showcases of Lean.