Legal folks define warranty returns as consequential damages. If you are making CD jewel cases, what is the potential damage of a defect? At worst, the case scratches the CD, making it inoperable. The cost is replacement plus associated overhead through the retail chain. So a CD case that costs $0.03 might have a maximum consequential damage cost of $400-$500. While a huge cost, the profit you made multiplied by the millions produced, factored by the probability this damage will happen, might mean each unit’s profit is reduced by $.00000X that goes into a kitty to pay for the occasional claim.

Let’s look at the medical market. Suppose you’re making hypodermic syringes. Total volume is measured in millions per day worldwide. Let’s further suppose a small chunk of the elastomeric plunger chips off and is injected into someone. A chunk of rubber wandering around inside your body can block your heart, damage your kidneys, or lodge in your brain.

This manufacturing mistake, depending on who is to blame and many other factors, could have consequential damage measured in millions of dollars. Even if your quality standard is to make a 99.999999% “perfect” product, that pesky 0.000001% is a problem. Getting into the real world, this means (statistically) that one part in a million could cost you $3 million. Ouch!
Companies that understand those consequential damages generally have folks who statistically figure the average failures, the average cost of the failure, and its subsequent erosion of profit, and compare the results with designing against the failure and lowering the odds of the warranty claims.

This is why, when you buy a car, the salesman says, “Bring it back and we’ll fix anything that’s wrong.” Part of the purchase price of the car is a few hundred “actuarial dollars” devoted to warranty claims. It’s cheaper to have you bring the car back if something isn’t right than for the automaker to go on a snipe hunt through the thousands of parts that make up your car to avoid it.

The companies that don’t have actuarials, warranty analysts, and statisticians use a slightly cruder, but equally effective method. Everyone acknowledges that manufacturing variances exist in all parts and all processes. Everyone acknowledges that they’ll get the occasional warranty return. But when too many come in, or the consequential damages are too high (although they already have product in the field), the profit erosion causes people to sit up and pay attention. Since there are way too many lawyers in this country, it’s becoming an interesting trick to pay the warranty costs, and then sue the subcontractor who made the problem part for the costs.

In the commercial market, all the hooey about scientific molding, PPAP, 12 Sigma, and all the other similar stuff is really designed around making functional parts. The ability to change the specifications and designs is relatively easy. If the part isn’t to-print but works, there’s a little bit of silliness and a change is processed. What’s important here is that in most commercial applications, a process is established and once it’s finalized, parts are shipped and everyone goes home happy, knowing how many life cycles they have to go through until they can no longer expect to be held accountable. (For those of us who buy tires, this is called a limited warranty. A 50,000-mile tire that has run 50,001 miles is no longer eligible for any claim against the manufacturer—unless you can make a really good argument.)

In the medical sector there’s an analogous mentality about functional and cosmetic parts. But because consequential damages can be so high and the unit cost is low, the R&D people first and the molders second are required to challenge the robustness of the process and the design. You don’t just dial in the process. Once a robust process is established, you then go through a host of DOE procedures to find out how far one or all of the variables can drift from the initial setting before bad product is manufactured.

Then you have to be able to demonstrate that you’ve not only done your homework, but also put procedures in place so that what you learned is passed along to anyone or anything that may be involved in the process. The result is that they, too, know all the limitations.

That’s how we get process documentation, employee training, maintenance training, machine capability studies, gauge training, and gauge R&R as part of our normal way of doing business. And all this is done in the name of moving that one defect in a million to one defect in a hundred million and then hoping you never produce enough product to make the one in a hundred million a realistic probability.

I had a medical client once who freely admitted it was normal to have about 15 wrongful death suits a year in court (we’re not counting out-of-court settlements) for the company’s product line. It sold millions of units per week to a high-risk population. For that population a problem could be tragic, yet for the company, it was a cost of doing business.

While there will always be room for the slam-and-squirt molder, the demands for process control have made everyone better. In the medical business, you not only have to be better, you also must be able to prove you are better by at least an order of magnitude beyond your customer’s expectations.

If not, it’s the Lawyers’ & Experts’ Full Employment Act.

Consultant Bill Tobin  does expert witness work and also trains people to manufacture and assemble plastic molded parts that won’t get you into court.