Build a Culture That Stops to Fix Problems

Table of Contents

WE ARE FREQUENTLY asked, “What is it about the way Toyota does things that makes their quality consistently better than other car companies?” While there is no one simple answer to this question, it can be said that a large part is due to the principle of building in quality and the decision to stop and fix problems as they occur rather than pushing them down the line to be resolved later. On the surface, this idea seems logical. If you have a problem, it‘s better to stop and take care of it. Correct it, prevent its recurrence, and make things better in the long run. In reality, when people are faced with the demand to “make the numbers,” the primary focus becomes short-term results—hitting the production target every day at any cost. The focus in mass manufacturing is in getting the mass. In lean the focus is on eliminating waste.

Not that Toyota wants the line to stop. Excessive line stoppage would severely reduce production and profitability, so if you stop the line often the results will be terrible. If you commit to the concept of stopping the line, you will either continue to have terrible results or you’ll work diligently to eliminate the problems. This commitment takes real courage and understanding of the long-term objective. Over the years, Toyota has built a system that provides the long-term benefits of stopping the line, and minimizes the negative effect by building a support structure to quickly identify, respond, and correct problems.

Developing the Culture

The demand for better quality has been on the increase for the past two decades or more. Everyone understands that survival in the marketplace depends on the ability to deliver consistently high quality. The interest in Six Sigma—and the allure of a quality level of only three defects per million products—has at least helped to create the awareness that delivering a quality product is a must in today’s world. The question now is not whether the ability to deliver exceptional quality is necessary, it’s: “What do we need to do to get there?” The best place to start is with yourself. If you want to learn a few lessons from Toyota, first you must develop a clear understanding of how and why Toyota accomplishes what they do.

The diagram in Figure 8-1 is similar to the models we have used in previous chapters, but it’s applied to stopping the line. The traditional method of reducing per unit cost creates a mind-set of never stopping the line because higher production numbers theoretically equal lower cost per unit. Any problems that arise can be corrected later, according to this approach, and quality is controlled by additional inspection and containment. This erroneous thinking creates an attitude among the workforce that identifying problems and possible solutions is not important. People may have good ideas to solve the problems, but they “won’t bother” because they’ve been told: “Don’t worry, someone down the line will take care of that. You just worry about your job.” In this environment, quality control must accept responsibility for catching problems (and for catching the offending person), and resentment often develops since the inspectors are seen as “police” or “narcs” if they report a problem. In the long run this system does not invite cooperation and mutual respect—two important ingredients of the Toyota Way.

Figure 8-2 depicts the Toyota core philosophy of eliminating waste. Notice that all the examples begin with the same core philosophy. If the thinking begins with a focus on waste elimination (in this case waste of correction), the natural extension of that philosophy is to develop a system that emphasizes getting quality right the first time. Toyota has developed an extensive support system to provide people the tools and resources to identify problems and solve them. Of course, the pressure of stopping an entire line creates a sense of urgency, and everyone must make a concerted effort to resolve problems permanently, or else the line stoppage would be excessive. People know they will be supported when a problem occurs, and with the fear of retribution eliminated, they can develop a cooperative attitude toward improving performance.

Based on the phenomenal financial performance of Toyota and the legendary quality levels they have achieved, there is no question that the “stop and fix” process works. For some reason the idea that it is better to keep the line going at all costs still pervades many organizations today. Often the “culture” seems to be the scapegoat. How do you change a culture? How do you change the habits that have been developed since the inception of your company?

Figure 8-1. Traditional method of correcting problems

Changing a culture is a challenge. Before you run out and start creating a culture, understand that cultures don’t just happen. Cultures are created over time. They arise out of need, in response to the system that exists to support them; or if there is no support structure, the culture that develops is one of selfsufficiency, “Every man for himself.”

Figure 8-2. Toyota method of stopping to fix problems

Case Example: Stopping the Line Is More Than a Cord with Lights and Bells

By David Meier

The following situation occurred in an automotive assembly plant that I coached in the lean methods, which was run by one of the U.S. automakers. This situation was extremely difficult for someone conditioned in the Toyota Way, and I felt severe anxiety. The Toyota Way conditioning is similar to that of Pavlov’s famous dogs. I have been conditioned to search out potential problems and to respond when called. Even though I was not in a Toyota Way facility and was not responsible for this situation, I responded the Toyota Way.

I realized I had this condition sometime after leaving Toyota. As I tour other work areas, I am constantly aware of potential problems and the need to take corrective action even before a “problem” surfaces. I notice also, with dismay, that the leadership in these facilities seems “blissfully oblivious” to the conditions around them. We walk past current problems and potential problems as if they weren’t there. My mind is screaming, “Wait, here is a problem. It must be dealt with or the consequence will be large.” Then I realize that the problem is “covered” and is lost in the “clouds.” There is no immediate urgency to correct it. I also realize there are so many problems that I would be overwhelmed. Then I recognize that leadership is in fact overwhelmed, and that there is no support structure. Problems happen all around and people do their best to get through the day. It is then that I truly understood the value of the Toyota system.

In the course of my consulting and while observing the final vehicle assembly line (sometimes referred to as the “money line”) in a Big Three plant, I noticed what appeared to be a tear in the carpet on the driver’s side floor. I happened to be standing with the supervisor at the time. My first reaction was to look for a way to stop the line. Of course there was no “cord,” as there is in Toyota, so I pointed out the tear to the supervisor and waited for a response. He looked and confirmed that in fact the carpet was torn, and did nothing! I was panicked and confused. I asked him what we should do, and he told me that the problem would be fixed in the repair area. I asked if we should look for the source of the problem in case it was repetitive and received a shrug. “They probably already know” was the reply.

This was my first experience with this kind of situation, and I did not know how to react externally, but internally I was very anxious. This was a potentially serious problem. At the very least the line should be stopped and this vehicle should not be completed, because all work done on the interior after this operation would need to be “undone” at the repair area. This included removal of the seats and much of the interior trim work. I know that this type of major repair, in addition to being costly, almost assuredly results in a product that is inferior to the original work. Reworking and replacing trim and seats is a significant cause of “squeaks and rattles” after time, and those issues are very annoying to customers.

In the end we completely walked away from the problem. We did not go to the end of the line to make sure the defect was identified and the repair done (preventing the escape to the customer), nor did we go to find the source to prevent further occurrence. We simply left!

I came to understand many other underlying issues later. For example, if a supervisor (or other person) finds a problem and points it out to the worker, the worker could file a complaint with the union that they were being “badgered.” While the claim may be unsubstantiated, the hassle of dealing with it is greater than the the hassle of fixing the problem later. The antagonistic environment between management and the workforce that has been honed for decades prevents cooperation (although I did find out that there are exceptional people who truly wanted to make things better). This is part of the culture that would need to change if the “stop the line” strategy was going to work here.

You don’t just announce to people, “Starting today, things will be different!” and suddenly the culture is changed. How do you change the supervisor who for the past 30 years has learned to survive within the old system? How do you change the mind-set about how people’s performance is measured? If people are measured on output, how will they respond? There is more to this than just deciding that from now on, we will stop to fix problems.

The following list includes many of the things you will need to do in order to be able to effectively create a “stop the line” culture and system. Note that when we refer to “stop the line,” we are also referring to stopping a machine or stopping the work process. It means that the work is halted when a problem is discovered.

1. Understand your current culture and why it developed.
2. Create a clear vision for change.
3. Pay attention to the respect and dignity of the people.
4. Establish a reasonable degree of stability in processes.
5. Have a method to stop the line.
6. The process must provide an audible and visual indication of the exact point of the problem. (Forget about paging systems!)
7. Have people designated to respond when the line stops.
8. Define the roles and procedure for response to problems.
9. Change the measurement process from just quantity to built-in quality.
10. Teach people to solve problems.
11. Increase the urgency, and make it necessary to fix problems.

The Role of Jidoka: Self-Monitoring Machines

Jidoka is roughly translated to mean “intelligent machines,” and specifically refers to the machine’s ability to detect a problem and to stop itself. It is an effort to have the machine work without continuous direct human monitoring, and it will sound an alert when there is a problem. Fortunately, many machinery manufacturers today are building self-checking capabilities into machines. As with many of the Toyota concepts, there is more to the concept of jidoka than self-stopping machines.

At the center of the Toyota philosophy is a respect for people and the value they provide. Only people can think and solve problems. Machinery is used to relieve human burden but is not a master to the person. Self-stopping machines relieve the person from the burden of constantly supervising a machine, and allow them to use their talents for more beneficial things (like adding value).

A legendary story at the Georgetown plant tells of a reporter who was doing a story on Toyota and the plant. When the reporter observed the door assembly and reattachment line, he commented about the lack of robots, which he had seen in competitors’ plants. Didn’t this reduce the efficiency of the plant, he wondered? The president of TMMK patiently explained that robots had limitations. They were not able to think, and they could not feel. It was important in the door installation process for a worker to sense what the customer wanted and to complete the task with the customer desire in mind. How should the door feel when it’s closed? How should it sound? A robot could not be trained for these things. While the cost may be higher for labor, the total benefit gained from having this sensing ability of a human was of greater benefit.

Utilizing jidoka is a matter of understanding where waste is in any process. Do you currently have machines that need constant attention? Does this create waiting time for the operator? You may have to observe closely to understand the true condition. Long ago people realized that having an operator standing around waiting for a machine was not desirable, so in many cases the operator filled the time with “busy work.” You may not see any actual waiting, so you need to look at the activity being performed while the machine is running. Is it value-added?

Notice whether you have machines idle because they need service but there is no recognition of this. We often see machines that automatically feed material and the feed gets jammed, or the material supply runs out, and the machine is waiting. This is waste also. Machines should be equipped with sensing devices and andons that sound an audible alarm and convey a visual signal to notify operators when they need service (preferably before they run out of material).

TRAP: Stack Lights May Not Be Effective Andons

When the concepts of andon and jidoka made their way into the general public, we started to see “stack lights” appear on machines. These are usually a small set of lights with three or four different colored lights stacked in one unit. This was to serve the purpose of an andon. There are a few problems. First we see that there is a general disregard of the lights. We can look across the work area and see many lights lit with different colors. What we don’t see is any specific response to the lights. The proliferation of lights has desensitized people. Also, the lights generally have no indication as to what they mean. When we ask people to explain the meaning, we get a variety of answers. Finally, there is no audible aspect to the lights. It’s relatively easy to ignore a light, but more difficult to ignore a buzzer. (By the way, Toyota’s andons have a different tune for the different conditions indicated. The supervisor call, or line stop, is Beethoven’s classic melody “Für Elis,”1 for example.)

Sadly, this is a classic case of the application of a lean tool with-out a deep understanding of the purpose, and without hansei to reflect on the shortcomings. People falsely believe that because they have the lights, they have andon, or jidoka. You must evaluate to determine whether the tools you implement are serving the function for which they were intended.

The Problem-Resolution Cycle

Before you try to build a system, it is important to understand the full cycle of the problem, from recognition through resolution and prevention. Figure 8-3 depicts the problem-resolution cycle visually. This cycle is typical within Toyota. This entire cycle is repeated many times throughout the day. Problems are constantly being surfaced and corrected, with minimal interruption to the production flow (“fixed position stop” is discussed in the section on “Minimizing Line Stop Time,” and Figure 8-4 illustrates the fixed position stop system). Think of these steps as a “chain of events” with each event triggering the successive event as necessary. This process is coordinated and orchestrated as well as any basketball team executing a certain play.

1. The first step of the process is the recognition that an abnormal situation exists. Recognition is possible because there are established visible standards (see Chapter 6) that are easily distinguishable by everyone. Let’s say, for example, that an operator is performing a task and realizes that he or she will not be able to finish it in the prescribed amount of time (takt time). This will be clear to the operator because the standardized work is synchronized with the line movement and demarcations on the floor indicate the step in the job. If the operator passes the line before the step is complete, he is falling behind and must request assistance.

Let’s look at another example, in this case quality. First of all, if an employee is to recognize that a defect had occurred, he will need a point of comparison. You guessed it, a standard. Refer to Chapter 6 for the discussion on the importance of quality boundary samples. In addition, the operator may have some latitude to correct minor problems on his own provided he does not exceed the takt time. These “rules” are also part of the standard. The rules are an important element of the next step, which involves a decision on when the operator must make a request for assistance. This must be defined!

2. If the condition exceeds the defined span of control of the employee she must elevate the problem and request support. This is done by “pulling the cord” or another means to signal the need for assistance. The andon device is used by Toyota to quickly indicate to the designated support people (team and group leaders) exactly where the problem is (by workstation). It typically includes an audible alarm to signal and a visual light to pinpoint location.

Figure 8-3. Toyota problem-resolution cycle

In many companies that try to implement an andon system, workers have a difficult time admitting they need support. They are concerned they will be held accountable. The leaders develop perceptions of workers and their abilities based on how often they need support (the “good ones” don’t stop the line so often). This is a critical juncture in culture development. Leadership must develop the attitude that their role is to support and ultimately find better methods so everyone can perform the work effectively. If resentment develops by workers or by leaders, the andon will become ineffective.

When the leader responds to the request for assistance he or she must take over responsibility for the problem from the operator. The operator explains the condition, and after the leader understands, the operator will return to his or her regular regular duties. From here on out ownership of the problem belongs to the leader for containment (permanent correction of the problem cause may be a joint activity with the team).

3. When taking over responsibility, the first thing the leader must do is evaluate the condition. Is this an isolated problem or a major problem? If the problem is contained or easily controlled (such as when an operator fell behind) and the leader assumes responsibility, the first consideration is to restart the line or perhaps intervene before the line actually stops. If the problem appears to be large or the source is unknown (such as a quality problem that originates somewhere else), the line will likely stop and stay stopped until the condition can be eliminated.

If the responding leader is unable to restart the line immediately, the situation must be elevated further. Of course, you can see by now that this repeated elevation cycle is based on predefined standards. For example, the team leader will have so many minutes to try to identify and correct the problem before he or she must elevate the situation to the group leader. When the group leader responds, there is a time limit after which she must notify the manager if the problem is not corrected. As the magnitude of a problem increases, the level of elevation must also increase. This ensures that larger problems receive the proper amount of attention, and also that upper management is not called upon to deal with smaller issues that can be handled by the appropriate leader. The role of management is to ensure that resources are available to correct problems quickly, and that corrective action to prevent recurrence is taken.

4. The first consideration is to keep the problem within station and to ensure that the problem will not reach the customer. The leader would typically walk downstream on the line to verify that the problem has not escaped to the customer. Stopping the line effectively controls the spread of the problem. This is the key point of the system—stop the line until the problem is effectively controlled or corrected. Stopping the line is a major decision and doing so will bring immediate attention to the problem. This is exactly the point. In traditional environments stopping the line will cause negative attention and is avoided, or if possible is done without bringing attention to the situation. The Toyota Way, in a sense, “celebrates” the fact that the problem has been forced to the surface, and everyone is encouraged because it can now be corrected. This is not to say that Toyota is happy when people make mistakes, but that when the cause of the mistake is found and eliminated, everyone understands that over the long term the process will be more robust.

5. The leader must identify the source of the problem so it can be contained. In the case of quality problems, the leader would begin to systematically walk the line to attempt to identify the source of the problem. Familiarity with the process aids in this procedure. For example, if a certain part is improperly installed, the leader can go directly to the operator who installs the part to find the origin. If the problem is random or sporadic, the leader must trace back at each operation until the source is located. If a defect occurs randomly, a decision may be made to restart the line while the search for the source continues. This decision is generally made by the group leader or above, and would be based on the severity of the problem.

Another purpose of the containment effort is to identify the parameters of the problem. When and/or where did it start, and where is the end? This is important for finding the source, but also for assuring that all defective parts have been corrected. When a more serious problem is identified, several leaders respond and each assumes responsibility for a portion of the containment effort.

6. After the problem has been controlled and contained and production has resumed, the focus shifts to prevention. In some cases preventive measures are short term in nature, meaning they are temporary measures until more effective permanent (long-term) measures can be implemented. The team leader implements these short-term countermeasures immediately to prevent further occurrence of the problem. If a longterm, more permanent solution is necessary, the responsibility for a solution may be returned to the entire team. All members are responsible for the development of effective countermeasures. The problem-solving process is used to find root causes and to determine effective, permanent solutions.

TIP: Develop Stability and Support Before Attempting to Implement Stop the Line

As we will see Toyota lines do not immediately stop when the andon cord is pulled. There is a very short time window (maybe 5 to 30 seconds) after the cord is pulled and before the line stops so that the team leader can respond and override the line stop if warranted. This system is very sensitive and requires a responsive support structure with high capability. Clearly, Toyota did not reach this level of capability overnight. A high level of stability was reached first so that the line does not stop continually and investments were made in developing a team leader structure to almost instantly respond to the andon calls. Move forward with your own implementation of andon in such a way as to balance the urgency to respond with the level of problems within your processes. If your processes are not reasonably stable, you will overload your support system and things will quickly unravel.

Minimizing Line Stop Time

Toyota has developed a system that allows problems to be identified and elevated without necessarily stopping the line. When a problem is identified and the cord is pulled, the alarm sounds and a yellow light are turned on. The line will continue to move until the end of the work zone—the “fixed position stop” point (Figure 8-4). The fixed position stop is especially useful for reducing actual line stops in the case of a worker who is behind in the work sequence. Markings are placed on the floor throughout the work zone that indicate the corresponding step of the standardized work. If the leader responds quickly and can reset the line by pulling the cord again before the fixed position stop is reached, the line will continue without interruption. Failure to reach the line in time or if the leader determines the problem warrants it, the line will stop when the fixed position is reached and the andon will turn red.

Many smaller problems can be corrected in this way without the annoyance of repeated starts and stops of the line. Also note that anytime the line is stopped, the problem is compounded because everyone on the line has to get resynchronized with the line when it restarts. Toyota also uses an audible alarm to signal all operators that the line is restarting.

Toyota assembly lines are generally very long and snake around corners. This can be viewed as a series of straight line segments connected by “U” shaped corners. A line stop is really a line-segment stop. Each segment of the line can stop briefly without stopping the others (Figure 8-5). As we noted, small intermittent line stoppages are disruptive to the flow. Imagine traffic stopped at a red light. When the light turns green, what happens? Do the cars all move simultaneously? No, they begin to move one by one until eventually all are moving.

This same phenomenon occurs when attempting to initiate flow. Small, defined buffers at the corners are used to absorb minor interruptions—no more than 10 minutes total buffer. If an operator activates the andon and the leader responds before the car has moved to the next work zone, the line never actually stops. If the problem is more significant and cannot be corrected before the end of the work zone, one segment of the line will stop (e.g., Line A), but the other lines (B and C) will continue, provided there is product in the buffer. If the stop time exceeds the capacity of the buffer, the following line will then stop due to shortage. Of course, these buffers are very small so that larger problems are not hidden by inventory.

Pull Cord



Visual indicators of work progress

Figure 8-4. Fixed position stop system

Fixed Position Stop

Response time before line stops

Small defined buffer

Line A

Small defined buffer

Figure 8-5. Fixed position stop and small buffering

General Motors has perhaps had the greatest opportunity outside of the Toyota group to learn the Toyota Production System (TPS). They co-own NUMMI, a joint venture and Toyota’s first application of TPS to an assembly plant outside of Japan. General Motors has had free access to NUMMI, including sending many employees to work there for one year or more. When they first started learning about TPS, they merely copied what they saw at NUMMI. They quickly discovered that it did not work. The andon system, for instance, did not work like it did at Toyota. General Motors had invested in some of the most sophisticated technology of fixed position, line-stopping systems to no avail. Workers did not use it to stop the line and surface and solve problems.

In the Hamtramack, Michigan, plant that built Cadillacs, GM got smart. They had an andon system set up with a fixed position stop system. But they did not turn on that automatic line-stopping capability. Instead they worked on stability of the process and on teaching the various lean methods in the plant, like standardized work and disciplined use of the kanban system. They set up work teams. Then they used an assessment process to assess each individual team in the plant. It was a broad assessment of the team’s discipline in using kanban, standard work, and responding to problems on the line. Only when the team achieved a strong score on the assessment could they have the automatic linestopping turned on. This was made very visible and was celebrated by each team that achieved this milestone. Teams worked hard to achieve this honor. And the line-stopping process began to work as intended.

Case Example: Making Line Stopping a Reward for Lean Maturity

General Motors has perhaps had the greatest opportunity outside of the Toyota group to learn the Toyota Production System (TPS). They co-own NUMMI, a joint venture and Toyota’s first application of TPS to an assembly plant outside of Japan. General Motors has had free access to NUMMI, including sending many employees to work there for one year or more. When they first started learning about TPS, they merely copied what they saw at NUMMI. They quickly discovered that it did not work. The andon system, for instance, did not work like it did at Toyota. General Motors had invested in some of the most sophisticated technology of fixed position, line-stopping systems to no avail. Workers did not use it to stop the line and surface and solve problems.

In the Hamtramack, Michigan, plant that built Cadillacs, GM got smart. They had an andon system set up with a fixed position stop system. But they did not turn on that automatic line-stopping capability. Instead they worked on stability of the process and on teaching the various lean methods in the plant, like standardized work and disciplined use of the kanban system. They set up work teams. Then they used an assessment process to assess each individual team in the plant. It was a broad assessment of the team’s discipline in using kanban, standard work, and responding to problems on the line. Only when the team achieved a strong score on the assessment could they have the automatic linestopping turned on. This was made very visible and was celebrated by each team that achieved this milestone. Teams worked hard to achieve this honor. And the line-stopping process began to work as intended.

Build Quality Inspections into Every Job

This is an interesting paradox. Inspection is not a value-adding activity, but Toyota stresses the inclusion of self-inspection in all standardized work. On the surface this seems contradictory. A deeper look will reveal that this is somewhat of a trade-off. Inspection does not add value, but it prevents greater waste from occurring. It is important to realize that Toyota does not take the addition of any waste lightly. There is always an effort to minimize the wasteful activity. There are several methods of incorporating inspection into the work in such a way as to minimize the waste and maximize the value.

Every operator has three responsibilities regarding quality:

  1. Check the incoming work to ensure that it is free of defects.
  2. Verify that his or her work is free of defects.
  3. Never knowingly pass defective product to the following operation.

The first item, to check the incoming work, can be accomplished while the part or the operator is in motion. For example, when the operator is completing the work cycle and returning to the next vehicle, she can visually check specific items as she walks. When a part is picked up, it is not picked up absent-mindedly, it is picked up with intent—the intent to verify that it is the correct part and that it is defect free. This inspection is an expectation for the correct performance of the job. These are not typical thorough inspections; they are very quick and specific. Thorough 100% inspections are done at the end of any subassembly or major process like body welding, body paint, or chassis production.

Targeting specific areas with a history of problems will increase the effectiveness of this checking process (use the data!). Other inspections can be performed as a part is being installed or removed. Make a point of training people to look at a specific location when performing the work. Chapter 11 has more details on job instruction training, the method Toyota uses to train employees.

In a similar manner, each person can check his or her work. Work is checked as it is removed from the machine or handed to the following operation. Specific quality “key points” are identified, and operators check them. For critical operations, or tasks that have had a history of missed steps, a yoshi is used. (Pronounced “yosh,” it is similar to a pilot calling “Check” while going over the preflight checks.) The standardized work dictates that upon completion of the task, the operator will point (yes literally!) at the part and call out “Yoshi,” signifying “I have checked this item.”

The pointing provides a visual cue to leaders that the check step is actually being performed (aiding in the auditing of standardized work). If this inspection were only visual, it is not possible to see if a person actually looked as instructed. Also, the physical act of pointing requires intention, and the intentional act causes the brain to engage. A step is less likely to be omitted if a yoshi is used. A similar process on parts where color marking does not matter is to use color markers and physically mark each place where a check has been done. The act of making the physical mark helps avoid missing checks.

Of course, one of the primary purposes of stopping the line is to prevent passing defects to following operations. Even with this extensive system and support available, it is one of the more difficult ideas to instill. People seem to have an aversion to admitting failure or incapability. One of the great benefits of small batch production is that if a defect is missed at one station and subsequent operators are checking incoming work, there will be a very short feedback loop from the time when the problem is created to the time when it’s discovered at a downstream operation. It might be a matter of one hour or less, for example, between when a welding operation is performed on a Toyota car body and when someone trying to install parts notices a bad hole position. In a traditional large batch operation the feedback loop time could be a week or more.

TIP: Don’t Give People Rules They’re Unable to Follow

This tip applies in many ways, but in this case it relates to the rule of never knowingly passing defects to following processes. This involves more than just telling people not to do it. What do they do if they find a defect? Who do they call? Where do they put it? If these issues are not defined, the people will be confused and conflicted. They want to do the right thing and follow instructions, but if it isn’t possible to follow the rule and get the job done effectively, they will choose to get the job done and violate the rule. Watch to see what happens. Try the task yourself to get firsthand experience. Don’t assume that people break the rules because they don’t care. Maybe there is not a good system for helping people to follow the rules.

Poka Yoke

Workers are assisted with the prevention of mistakes by the utilization of poka yoke methods or devices. This term is generally translated into English as “mistake proofing” or “error proofing.” Error proofing is not so much a lean “tool” as it is a way of thinking and evaluating problems. It is based on the philosophy that people do not intentionally make mistakes or perform the work incorrectly, but for various reasons mistakes can and do occur.

There is a significant difference in the Toyota Way of thinking about the causes of mistakes and the thought process used within other companies. In our work with other companies, everyone unanimously agrees, “People make mistakes.” It is also unanimously agreed, “If people paid attention they would not make as many errors.” Conventional thinking tends toward identifying the cause of mistakes as “human error,” while the Toyota Way always starts with the assumption that an error is a failure of the system and methods that are used to perform the work. Quite simply, errors occur because the current method allows them!

The difference in thinking shifts the responsibility for errors from the people to the method, which also shifts the blame for mistakes from people to systems. When people are released from blame, they are free to focus on creating more effective systems and actually solving problems, rather than defending themselves. It is common within Toyota for a manager to apologize to a worker when the worker makes an error, because management bears the responsibility for creating effective systems that prevent mistakes. When was the last time someone in your company apologized to a worker when the worker made a mistake?

The following case example typifies the thinking within most organizations.

Case Example: Errors in Faxing Orders

During an activity to improve order processing throughput time in an office setting, it was discovered that order approvals were often delayed by several days because of errors made during faxing to dealers. Orders were to be returned to the dealer for review and approval prior to submitting for production. The normal procedure was to return the preapproval proposal via fax to the dealer for final approval. The required response time from dealers was two workdays. If the proposal was inadvertently sent to the wrong dealer, two or three days would pass before follow-up was initiated to see if the dealer approved the proposal. During the follow-up it was discovered that the fax was never received. Investigation of the fax transmittal records showed that faxes were in fact inadvertently being sent to the wrong dealers.

Further investigation also revealed that certain employees made a higher number of mistakes, and the conclusion was that they were “more careless.” As a possible “solution,” instructions to employees were posted by the fax machines telling them to “Pay attention” and to “Be careful” and “Verify that faxes were sent correctly.” Of course, this did not solve the problem, and the conclusion was that certain employees would always “be problems” and that more checking was needed.

When the Toyota Way of thinking was suggested, the response was, “Human error is a reality. You will never eliminate human error.” Here is the issue. A person who develops a system generally understands it well and assumes that others should also. The developer (or even a person that has used it for years) believes that the system is simple and understandable. They have difficulty recognizing that others may have a different experience with the system and that people have different capabilities. Anyone who does not perform as well is considered to be incapable. Rarely does anyone consider the system. Let’s look at the method in this example to find the causes of errors in the system.

Due to the large number of incoming and outgoing faxes, four machines were used. Each machine could store 100 fax numbers and use a “shortcut” code to automatically dial the phone number. A list of all dealers was posted behind the fax machines so employees could locate the dealer and the correct code (Figure 8-6). The list was separated into three sheets that were each very large (20 by 20 inches). They were also posted on the wall behind the fax machines, some distance from the employee.

When we evaluated the method in which work was performed and tried to understand why the errors could occur, we saw that while all information necessary to perform this task was available, it was not laid out in a manner conducive to the specific task. Let’s evaluate the steps of the task.

Dealer Name

Dealer Region

Dealer Address

Dealer Representative

Phone Number

Fax Number

Pre-Programmed Code

A Plus Cabinets

South West

111 Short Street Anytown, AK

John Smith



Fax Machine 3#49

Astounding Cabinets


555 West Main, Yourtown, MS

George Jones



Fax Machine 2#32

Figure 8-6. Sample dealer list

  1. Look at order and identify dealer
  2. Locate the dealer on the list
  3. Scan across the page to identify the fax machine and code (remember this information)
  4. Find the fax machine
  5. Enter correct code and send fax

Evaluating the method, we discover that errors could occur during each step. It’s possible to identify the dealer on the order and then incorrectly identify the dealer on the list. When scanning from the dealer to the fax code, it’s possible to cross over into another line and identify the wrong dealer (remember, the sheet was behind the fax machines, where it was not possible to track across the lines with a finger). Moving from the sheet to the fax machine, the employee had to remember the correct fax machine and the code on the machine. It was possible to get the correct machine and the incorrect code or the correct code on the wrong machine or forget the information that had been looked up.

Management concluded that employees should be careful when scanning the list and that remembering the machines and codes should be easy. When they looked at the system, they concluded that is was “easy” largely because they only had to do the task one time error free to prove the point that the system was fine. If they had to do the process hundreds or thousands of times, and sometimes were in a hurry, they would discover that they too would make numerous errors. This is a common failure of thinking by management. It may be easy to do a task one time without errors. Doing it hundreds of times without errors is another matter entirely.

To simplify this task to minimize errors, let’s look at the potential causes of errors.

Error: Incorrect matching of dealer with fax machine and code. Why does this error occur?

  1. The two pieces of information used are at opposite ends of the long sheet (almost 20 inches apart).
  2. There’s no visual delineation between lines on the sheet, making it easy to cross into the next
  3. The sheets are behind the machine, where the employee can’t use a finger to trace across the

Solution: Reformat the sheet so the dealer name and the code are side by side. This minimizes the possibility of crossing lines. Also, shade every other line so it’s easier to stay within the correct line, as shown in Figure 8-7.

Dealer Name

Pre-Programmed Code

Dealer Region

Dealer Address

Dealer Representative

Phone Number

Fax Number

A Plus Cabinets

Fax Machine 2#49

South West

111 Short Street Anytown, AK

John Smith



Astounding Cabinets

Fax Machine 2#32


555 West Main, Yourtown, MS

George Jones



Figure 8-7. Reformatted dealer list

Error: Employee does not use the correct fax machine identified. Why does this error occur?

The sheets are spread out behind the fax machines and the employee has to remember the correct machine identified and locate the correct machine. Figure 8-8 shows this condition.

Figure 8-8. Arrangement of dealer lists by fax machine

Figure 8-9. Revised layout of dealer lists by fax machine

Solution: Separate the dealer sheets by designated fax machine and place the sheets in front of the machines so employees can trace the line with a finger. The new layout is shown in Figure 8-9.

Implementing these changes reduced the errors significantly. It was also discovered that errors were made in the initial programming of codes. Even if the employee correctly identified the code, the fax was sent incorrectly.

You may also ask, “Did you consider a verification process to ensure that the fax was received by the dealer?” This was in fact initiated, but bear in mind that this “solution” is similar to an inspection of product after it has been produced. This extra step does not address the root causes of the error, and it adds cost. It should only be used for critical processes, and only after root causes have been corrected.

The key to developing effective mistake proofing lies in understanding how or why the mistake occurred. Do you understand the circumstances that led to the error? Was this a random mistake or a repetitive one? Does the mistake happen with everyone or is there one person who has difficulty? If it is an individual, the answer may be to evaluate standardized work and make sure that no steps are omitted. If everyone has the problem, there may be a singular cause such as missing information or a step that is not clear. Don’t make the mistake of believing every error requires a “device” to prevent recurrence.

The following case example demonstrates that there are always multiple ways to solve any problem. Encourage your employees to be creative and to seek solutions that are highly effective and low in cost. There is great power in simplicity. Look around you to apply existing solutions from other problems to your problem. The idea for Velcro was developed after an observation of “stickers,” and the “problem” was converted to a solution to other situations.

Case Example: There’s Always More Than One Way to Prevent Errors

In the Toyota Georgetown plant the parts were painted in a leveled color mix and sequence, meaning that one part would be painted white, the next might be blue, then black, and then back to white. It was possible to have the same color back-to-back, but the mix varied according to several conditions. The paint system required continuous circulation of paint, and only one paint gun per painter was used. Each time the color was changed, the painter would disconnect the paint line from the QD (quick disconnect) and attach it to the paint gun. The painter would flush the line momentarily and then paint the next part. Because the color changed from part to part, the painter had to disconnect and reconnect the paint line for each part. One key element of error proofing lies in understanding that people will generally behave like electricity— they’ll seek the path of least resistance. In this case the painters wanted to avoid the continuous changing of paint lines.

Each paint booth had three painters. After the first painter applied color to the part, subsequent painters could visually see the colors of the parts (and the color mix) coming to them. On occasion the pattern was such that a white part was followed by a red and then another white, for example. When the painter could see that the white would be needed again, he or she would hold the white line and reconnect it after the red part (never reattaching the white line to the wall).

Sometimes the painter would deviate from the rule and have multiple lines disconnected at one time, which caused the operator to mistakenly reconnect one paint-line color to a QD of another paint color. Then both colors would be mixed throughout the entire system—a big problem! This happened several times each year, and the total lost labor, materials, and waste disposal amounted to over $80,000 per year, which did not include the cost to the customer (the assembly line).

The paint line would stop while the team leaders prepared hand-mixed paint in “paint pots” for each painter so the line could be restarted. The line stoppage often created a shortage of parts to the assembly line—now a very serious problem. Previous efforts at error proofing yielded the following “solutions”:

  1. Notify the painters that the standardized work specified only one paint hose be removed at any time and that standardized work must be followed. As might be expected, this level of error proofing— tell the employees the correct method—is rarely effective.
  2. Post a sign that stated “Only one paint hose can be removed at any time” at each This commonly applied attempt at error proofing—to post notification of the rule or method—is also rarely effective. Most people assume that a sign, clearly in view, will prevent errors. This seems logical. People don’t break the rules maliciously (most of the time), but they often rationalize, “I don’t think I’ll make the mistake, so it’s okay if I break the rule.”
  3. Label the paint lines. Overspray buildup quickly obliterated any labels and made them unreadable.
  4. A cover “flap” was placed over the QD for the white paint line, which required the painter to lift the cover to reconnect the hose. This countermeasure was based on the fact that the majority of incidences in the past involved white paint mixing with another Since 40 percent of all vehicles were white, the odds of having multiple white bumpers within the sequence simultaneously were higher. The cover over the white QD was intended to make the operator “think” before replacing the white paint line (similar to a yoshi). This “solution” also had no effect because it did not prevent connecting the wrong lines. It only made the work more difficult (lifting the cover for 40 percent of the jobs).

These four attempts at prevention represent the hierarchy of error proofing moving from telling or sharing information, to posting notices, to attempts at prevention by self-checking. The efforts may have prevented some occurrences, but they did not prevent them all.

After these attempts to eliminate the error failed, a solution was proposed to use a device known as a “peanut.” This allowed the paint to recirculate at the paint gun, and eliminated the need to disconnect the paint line. This was an effective preventive device, but the negative points were additional weight on the paint gun (nearly one pound), which was an ergonomic concern, and the high cost of the devices. An installation on the entire paint line would cost over $10,000—not a low-cost solution.

It was clear from observation that the operators’ inclination to deviate from the described method was not out of spite. They were deviating because of a natural desire to reduce their own burden, and they assumed that they would never make the mistake. In this case, an errorproofing method was needed that removed any need for a conscious act (following the rules). It was necessary to remove any options.

Watching the painters disconnect and reconnect the line, it looked like someone putting a key in a lock and opening a door. This was the seed for an idea. What if each paint line could somehow be made to fit only one QD? What if each QD and paint-line combination could be like a lock and a key? The QDs would need to be like locks, and the paint lines keys. A toolmaker was able to make a mock-up sample using a quarter-inch steel plate over the front of the QDs that had unique slots milled for each separate color (Figure 8-10). Then he made a sleeve that was attached to each paint line with a pin configuration that would fit the slot. Just like a lock and key! The prototype was installed at one station to test and to verify ease of use. After modifications, matching sets were made for each paint station (all were identical lock/key configurations).

Figure 8-10. Lock and key error-proofing device

Of course, this method is not completely mistake proof. In fact, it’s possible to defeat the method for any system that is developed (think about computer hackers breaking into “secure” systems). In this case, if the pins were broken or removed, the device would be ineffective. To counter this possibility the team leader added a check of the pins to the daily preshift inspection to verify that everything was in correct working order.

The cost of this simple solution was about $200 for material and labor, and the sleeves only added a few ounces of weight to the operator— not a significant ergonomic impact. The device effectively prevented the problem with zero recurrences.

There are some key points that support the effort of mistake proofing. The key determinant of your success is your mind-set. Error-proofing techniques and tools are simple and easy to apply. The greatest challenge is in discovering the root cause and using your imagination to effectively eliminate it.

There is a hierarchy when it comes to the mistake-proofing effort. The highest order is to prevent the occurrence of the error completely. But complete elimination is not always feasible or practical. Any system or device that is installed can be bypassed if desired. If a prevention device is installed that is problematic or cumbersome, people will tend to “work around” it. If you create a cure that is worse than the disease, people will sidestep your cure.

If it’s not possible to completely prevent the error (most of the time), then try to detect the error as it occurs. Detection devices or methods are more common (jidoka devices fit this category). The device may detect a broken tool or signal the operator that a component is missing. The tool is already broken and some parts may be damaged, but the problem is detected and corrected quickly.

TRAP: Don’t Go Overboard with Poka Yoke

Unfortunately, too much of a good thing can be bad. We’ve seen a trend toward higher and higher levels of mistake proofing. In many cases the devices are developed by engineers, and the actual workers have no input. The devices become extremely sophisticated and add layers of complexity. We have seen operations that take 15 seconds to perform the actual work, but an additional 25 seconds or more to engage the mistake-proofing device! In one example, a molded part has a few clips and slide rails added, then is placed in a mistake-proofing device. The part is locked in the fixture and then cycled to test for application of the correct parts. After the cycle is completed, it’s unclamped and removed from the fixture. The checking takes nearly twice as long as the actual work! The sad part is that after this elaborate testing process it’s possible for the parts to fall out, causing a defective part to reach the customer anyway! In addition, the sophistication of the devices creates problems with understanding how to operate them, reset after errors, and so forth. Try to find simple, effective mistake-proofing methods.

In any case, it’s important to prevent any defective items (or mistakes) from affecting the customer. This is true even if an effective error-proofing device is used. There must always be protective “gates” to ensure that the customer is never compromised. As mentioned earlier, each employee acts as a “gate” by self-inspecting and inspecting key areas of others’ work. More inspection points (gates) reduce the possibility of escape.

This list of possible causes of errors or omissions may not be all-inclusive, but it covers the primary causes:

  1. Deviation from defined work method (work must be standardized before attempting poka yoke).
    • Omitted steps
    • Steps out of sequence
  2. Missing parts (or components of the work)
  3. Improper part (watch out for interchangeable parts)
  4. Incorrect setup (wrong tools or settings)
  5. Errors in information or documentation
  6. Transposing type errors (watch for long number strings)
  7. Misinterpretation type errors (look for similar descriptions, numbers, and appearance)
  8. Recognizing the mistake but failing to segregate or correct it

Note that mistake proofing extends well beyond the prevention of defects. It applies to any work activity and to any mistake that creates a deviation from the defined standard. Perhaps the material handler forgets to pick up an item, or a designer forgets to put key information on the prints. Consider this example that we are all familiar with:

You pull into the drive-through of your favorite fast food restaurant to order lunch and notice that a screen displays your order as you call it in and the cashier enters it. The screen has a statement that says, “This screen is to ensure order accuracy.” So one possible point of error is in entering the order. How could an error occur? Perhaps the cashier could not hear clearly. Perhaps the customer ordered the wrong thing or the wrong quantity (not that that would ever happen in your business!). The cashier may press the incorrect key (note that the cash registers have preprogrammed buttons for all items—an errorproofing and time-saving example), or enter an incorrect quantity.

Well, there are several possibilities for mistakes to occur, and we’re only at step one! Remember in the “old days” when the order would be called back to the food preparation area? This is another opportunity for errors. Again, what if it were called out incorrectly, or heard incorrectly, or either party forgot part of the order? Many restaurants have installed monitors that display the order to the kitchen as it’s keyed in. No chance to misunderstand or forget. Then comes the actual food preparation. This may be the area of greatest opportunity for error. The error rate is most likely tied to orders that “deviate” from standard (even though they promise that you can “have it your way”). Did you ask to hold the mayo or to add extra pickles? Errors might occur in the actual preparation of the nonstandard item, or the order could be prepared correctly and then a different item placed in your bag. Are the “specials” separated from the standards? How are specials visually indicated to prevent errors? With all of the possible opportunities for errors, it’s a wonder the order comes out right most of the time.

Creating a Support Structure

In most traditional operations miscellaneous problems occur throughout the day, and leaders are seldom notified (based on the individual preference of the worker). We have observed machines that were not operating, defects that were piling up, and even operators who had left the work area for some reason, and there was no response. The number of problems often overwhelms the leaders because they are spread too thin.

One of the major differences between Toyota and other companies is the support structure and how it’s utilized to effectively control problems and keep the system operating. The roles and responsibilities of the team and group leaders, and suggestions for selecting the right people, will be explained in Chapter 10. Suffice to say here that a critical aspect for the supporting roles is the “span of ” It is not possible for a leader to respond to the needs of several dozen people if the line or operation will be stopped each time a problem occurs. Again this is an issue of compromise—adding the waste of extra indirect workers who do not directly add value in order to eliminate or avoid much bigger wastes. The Toyota Way is full of short-term investments that result in longterm payoffs many times over.

Reflect and Learn from the Process

  1. Take time to reflect on your organization’s culture regarding building in quality and getting the job done right the first time.
    1. What is the cultural view of people and mistakes? Is it that people make mistakes intentionally or are careless or that there are “bad” employees? Listen to conversations and make a mental note of the comments.
    2. Do you believe that some problems are due to carelessness?
    3. How will your thinking and actions need to change in order to influence the organization?
    4. Are people in the company expected to participate in the identification and elimination of problems they detect?
  2. During your “waste walks” pay particular attention to what happens when a problem occurs.
    1. How did you know there was a problem? Could you (visually) see a deviation from a standard?
    2. How did the person who detected the problem know it was a problem? Does he or she have a standard for comparison or “just know” from “experience”?
    3. How was the problem handled? Was the person able to elevate the problem from the work site, or did the person have to to find help?
    4. Was there a defined response to the elevation of the problem?
    5. Did the response include verification that the problem did not affect the customer, and if it did, was the problem contained to prevent spread to the customer?
    6. Did the response include verification of the cause of the problem and corrective action to prevent further occurrence?
    7. What is the total response time for the problem resolution cycle? What are the total losses from having an ineffective system?
    8. What actions need to be added to your implementation plan to improve your system?
  1. The foundation for building quality at the source is standardized Evaluate your standardized work process to answer the following questions:
    1. Is the standardized work clear and understandable?
    2. Is an incoming quality check included in every job? Have specific areas been identified for checks based on historical data (check known problem areas)?
    3. Have the key quality points been identified for each job, and are they verified prior to completion of the job?
    4. Does your system allow people to stop the process if they detect a problem?
    5. Does this system automatically elevate the problem to ensure that corrective action is taken?
  2. The next time a problem occurs that is caused by someone making a “mistake,” evaluate the corrective action response.
    1. Does the countermeasure go beyond reminders to employees, signs, and retraining? If not, this indicates inability to find true causes and identify effective solutions.
    2. Are suggestions for solutions solicited from employees?
    3. Evaluate the mistake for the true root cause (see problem solving section). What can be done to prevent anyone from making the same or a similar mistake?
    4. What is the level of your attempts at mistake proofing? Are you putting up signs, detecting problems that already occurred, or preventing the problem from occurring in the first place?
    5. Add to your implementation plan details for teaching your leadership how to do root-cause analysis and error proofing.