I don't think one will find many people who really object to the idea of doing things properly the first time. Alas - it's easier said than done due to the inherent fallibility of human thought, planning, and effort. A key problem with the zero defect approach as I have seen it applied is that it is often treated as a quasi-religious dogma, rather than a quality objective or methodology. Some processes (I suspect the VAST majority) are incapable of producing zero defects. They may produce VERY low levels of defective product, but not zero. Often times the cost involved in reaching zero is so high that it vastly outweighs the potential benefits.

I have had personal experience with a process that produced and leaked a rejectable part once every 4 to 6 months. That defect had no affect on safety or on the reliability of the final assembly, and was virtually invisible to the final user of the product. The part cost something like 25 cents. The cost to reliably detect each and every one of those would be tens of thousands of dollars, and a proper process redesign to get (maybe) to zero would cost 6 figures. There is no economic justification for pushing to zero defects in that case or the many similar cases out in the world at this time. Yet the customer for that product is baffled that anyone would be reluctant to dump adequate resources into the effort to go from very few bad parts to zero bad parts. After all, bad parts are morally repugnant. Defects = sin, and they make the angels cry. It's an absurd, legalistic mindset divorced from the reality of the economical production of goods.

Continual improvement efforts, focused on the biggest cost drivers, are the key to profitable allocation of scarce resources. Good engineers, technicians, maintenance men, machine operators, etc. are increasingly hard to find, and usually overworked. Zero-defect dogmatism threatens to disrupt the logical process of assigning priorities by giving it a moral dimension and the associated emotional preoccuption with things that are "wrong."

I fear we're doomed to live in a world that has a few defects now and then.

I used to manage semiconductor packaging design and support engineering and was accosted one day in the 1980s by a young quality engineer full of Zero Defects zeal. He was concerned that the yield of the prototype runs of a brand new package construction and design was less than 100% (it was actually ~93%) and he was 'threatening' to shut down the production line start-up as a result. I politely told him that I would allow that if he would shut down the whole wafer fab. operation first.
He didn't, at first, understand my response so I pointed out that the semiconductor yields from silicon wafers had never been 100% and were never expected to be. Frequently the first wafers yielded zero yield and only the incredible 'microsurgery' techniques of a very gifted colleague managed to verify most of the problems and their resolution without costly and time-consuming complete new wafer run. The new wafer run would ratchet up yields to perhaps 10 - 20% if we were lucky, but that would be sufficient to initiate production runs. What was important was that, despite very much less than perfect yields from the current state-of-the art wafer yields, the business could be commercially viable, either immediately, or in the near future. Waiting until perfection had been achieved was untenable and ludicrous. The packaging technology was instrumental in enabling at least a two years advantage over the competition, on a breakthrough product line, because it resolved a series of problems that the mainstream industry packaging approach could not handle.

What do we learn from this anecdote?

I believe it is that, for relatively simple, deterministic, mature process systems (e.g. machined parts, software etc.), the zero defects target is a reasonable target that can be closely approached or achieved. However, for complex, less mature, non-deterministic processes (e.g. silicon wafer fab., or complex new productline production shakedown etc.) the ultimate 'target' should be 100%, but initially, realistic, intermediate, lower targets are commercially rational and essential.

Zero defects can and is achieved in many ways throughout the manufacturing industry and service organizations. Granted there are some processes that yield less than 100% but zero defect performance is related to what is required by the customer. Most customers that I have communicated with do not want to pay for any defective products. They do not want to sort for defects or add any no-value added steps to their processes to deal with potential defects. Would you fly on a plane if 1 out of 10,000 landing gear assemblies contained a defective $0.25 part that may fail during landing? I think not. How many babies are doctors and nurses allowed to drop at birth? I would think zero. The point is that you set a performance standard and then look for cost effective ways to achieve results 100% of the time. Mistakes are made but you don't just say it is too costly to fix. If you don't do it, your competitors will.

I've worked in healthcare quality for nearly two decades (so including the TQM drive in the early 1990's). In hospital care, I think we can set a goal for zero process step defects, but we can never totally eliminate such defects as death, infections, certain complications, and the like.

As an example, a patient can get perfect care as defined by the current evidence based medicine with regard to thromboembolism prevention (e.g. the proper blood thinner such as lovenox, have surgical stockings, etc.) and still get a Deep Vein Thrombosus (DVT) or even worse a pulmonary embolism (PE) resulting in major morbidity if not death. Similar "perfect care processes" can occur in care for heart attack, pneumonia, stroke, etc., and there can still be negative outcomes including death. Even infections are often more related to patient condition (e.g. smoking, obese, diabetic patients have much higher infection risk) than to the processes of care.

In healthcare at least, to set a goal of zero output defects, becomes an empty exhortation by management (one of Deming's sins). On the other hand, we should set the bar at zero defects for process metrics (like hand washing, patient identification, following evidence based practice, etc.)

Steve Osborn, CSSBB, CPHQ
VP, Quality
Saint Vincent Health Center
Erie, PA