Mistakes made early in the process, if not corrected, are compounded by work that comes later, because subsequent work also is incorrect, hides the mistake, and prevents access to the mistake for correction. (One simple example is not realizing until after you arrive at work that you’ve put on two different colors of socks. If you’d noticed before you put your shoes on, let alone left the house, the mistake would have been much easier to fix.) Mistakes made late in the process — e.g., incorrectly lacing the shoes — may be easier to fix (if they don’t ruin or negate all the correct work that went before), but nonetheless involve rework.
The sooner you find a mistake and fix it, the less harm the mistake will create and the easier it will be to remedy. Mistakes are easiest to fix if they are identified as closely as possible in time and space to where they occurred. If you fix your mistakes soon enough, your work will have zero defects, and waste will be kept to a minimum. So the idea is to design your systems to find and fix mistakes before they can turn into defects.
The power of source inspection
When this concept is combined with inspections, it doesn’t take long to see the problems with inspections at the end of the process. Looking more closely, Shingo proposed that inspection can take place inside or outside a process, and that inspections outside the process — at the end of a production line, for instance — were simply too late. By then, it’s more difficult to tell where or why an error occurred, the costs of subsequent process steps have already been incurred, and the cost of rework can be considerable, if rework is possible at all.
Many businesses consider pre-delivery inspection to be a hallowed tradition, whether it is performed by employees or third-party inspectors such as governmental agencies and regulators. The folly of this approach from a quality and efficiency standpoint may not be immediately apparent in some industries, but imagine what would happen if hospitals routinely X-rayed patients after operations are complete to check for metal instruments inadvertently left inside the body. And we all know that software, which typically is tested thoroughly before it’s released, still hits the market with plenty of bugs.
Even worse, some businesses essentially allow the customer to do the inspections, either after the goods have changed hands or, less frequently, at the point of production. Boeing is just one example, along with many other companies that produce large-scale customized products. But when customers perform the inspection, even if problems they find are fixed before final delivery, they still create costly issues of rework as well as loss of customer confidence. Moreover, when the goods have already changed hands, customer inspections too often result in dissatisfaction, returns, loss of reputation, and so on.
Inspection inside a process, on the other hand, can take place immediately before or after the process step itself, or it can take place within the step as part of the process. Inspection within a process is called source inspection. An early example took place at Yamada Electric, a subcontractor for Matsushita Electric. In 1961, Yamada Electric was making consoles that included on and off buttons. The trouble was that workers needed to put a small spring on each of the on and off buttons, and sometimes one or the other spring was left off.
To improve this process, Yamada Electric began presenting the console to the workers with exactly two springs in a small dish. If the dish was empty at the end of the process (and assuming nobody simply dropped one and neglected to pick it up), clearly both springs had been placed on the console. This simple concept not only eliminated the problem of missing springs, it provided a visual indicator to workers, or anyone walking by, how far along in the process the worker had gotten and, once the dish was empty, that the process had been completed correctly. The workers could perform the inspection themselves with a glance at the dish.
Source inspection by definition permits faster, easier correction of mistakes because the inspection happens where the mistake has been made. Shingo distinguished two kinds of source inspection. In self-check, which is what was put into place at Yamada Electric, the check occurs within or just after the process step, and mistakes are fixed before the goods move onward in their flow. In successive check, a worker at the next process checks the materials just received, prior to performing the work, and if flaws are found, the worker sends the materials back to be fixed.
Clearly self-check, which makes every employee an inspector, is preferable in terms of wasted time and transportation. It’s also more reliable than successive check. One reason is that successive check requires the workers at both processes to know exactly what should be done at Process A and what constitutes a mistake. Another is that, depending on how much the mistake impacts Process B, Worker B may or may not be motivated to perform the check or to send mistakes back to A.
Of course, even self-check can have drawbacks. If the check is performed by a human, discipline to perform the step can still be an issue, and mistakes can occur. The best self-check is when the process or machine itself performs the check, automatically, and work cannot proceed if a mistake is detected. A good example is provided by the various gas hoses used by the anesthesiologist in a hospital operating room. Various hoses supply half a dozen or more gasses, and an error in connecting a hose to other equipment or to the patient could be deadly. To prevent errors, each hose is not only color-coded but tipped with a connector of unique design, so it is literally impossible to hook up to the wrong gas supply. If a connector doesn’t fit the device the anesthesiologist is trying to attach, the process stops. This is self-check at its highest level.
By 1977, Shingo and Matsushita Electric had demonstrated that by using self-check source inspection, zero-defect manufacturing was indeed possible.
Pokayoke or mistake-proofing
Shingo coined a term for moving the source check so close to the point of any mistake that the process can stop itself and the problem can be fixed immediately with negligible rework or wasted time. This term is pokayoke. Literally, pokayoke means mistake-proofing. Originally the term used was bakayoke, or fool-proofing. In Japanese, however, baka is rather stronger than “fool,” with a more derogatory meaning closer to idiot or imbecile. Use of this term so insulted an employee of a firm where Shingo was consulting that she wouldn’t return to her job until after he changed the name to pokayoke instead.
Pokayoke is usually built into the task or the function of a machine, guarding against human error, forgetfulness, or inattention, and it is therefore both the lowest-cost and the fastest route to 100 percent inspection. Pokayoke provides immediate, objective feedback and allows for corrective action if a mistake occurs.
When pokayoke isn’t possible, self-check permits the next fastest and most efficient source inspection, but remember that self-check may be contingent upon not only subjectivity but compliance. Successive checking is thus the last-ditch opportunity to find mistakes and correct them in a still-timely and efficient manner.
With these source inspection complexities in mind, we’ll look in Chapter 4 at employee involvement as a necessary first step to lean, world-class production.
Lesson 4: Employee involvement is the foundation. Without it, you can’t build a lean, world-class production system.
The question of how to involve the work force — an absolutely essential foundation of a world-class operation — continues to challenge companies worldwide. My own experience with this knotty problem began during my career in the military, but took a quantum leap in 1982 while I was working for Boeing. That year, Ernie Fenn, vice president and general manager of Boeing’s 757 program, asked me, “What can we do to make the 757 program into a more participative culture?” He wanted to give employees a greater voice in improving how the work got done. I told him we could not change how employees behaved without changing how managers behaved.
Employee