What is a Warranty Anyway?

One of the greatest parts about my job as a Quality Engineer is it puts me in the middle of some rather interesting arguments and debates. Much of my work involves research into the root cause of specific issues, then presenting a logical claim to a supplier or customer about the findings, then implementing a fix. Of course, nobody wants to be at fault over a given issue as nobody wants to pay the charges associated with either providing defective product, or misusing product.

Earlier this week I heard something from a customer that I here often, essentially, “We want warranty”, but it wasn’t the question by itself that got me, it was the way they asked it. To put it into perspective the piece of equipment in question was $30,000 new, and agreed to on both sides to have failed due to misuse after a lengthy analysis on our part. That aside, however, the factory warranty had expired anyway. In any case we fixed the unit with a $20 part, and I said (in a 4 page report detailing all our testing) that everything is now “form, fit, function” performing to the design intent. Great, wrap it up, get approval on the charges and ship it! Oh yeah, customer, don’t do that thing you were doing again!

Thinking I was ready to move onto the next project I suddenly get an email from the customers project managed which stated that they wanted full warranty on the unit. To be clear, they didn’t want me to cover the unit under warranty, they simply wanted full warranty coverage going forward. They argued, “Since we returned the unit back to the factory (you), and you have deemed the product good, you should give us full warranty. If not, you better come up with a pretty good reason or it’s not going to look good for you.” That’s pretty much a direct quote. My laughter, incidentally, was heard up and over the cube walls, and down to the other side of the building. He was serious.

His logic was simple enough – if the factory says something is good, then that should be a guarantee that it’s good within the given warranty period. If you’re not guaranteeing it’s operation for any amount of time, then what’s the point of even saying it’s good. (for the record, we do warrant the parts replaced, but that’s all) Now I could have taken the legal stance on this and said, “Look Mr. customer, we offer a two year warranty on the product (which is industry standard), once it’s up, it’s up.” However he said, “…you better come up with a pretty good reason…” And that answer is not a good reason.

So then, just what is a warranty anyway? Simply put, it’s an explicit guarantee that a given product will last a stated amount of time with no defects. If a defect should occur within that time, the factory agrees to repair or replace the product. The other part of that is, just how is a warranty derived? That can be a pretty complex question to answer, however in most cases it’s driven by whatever industry you’re in. If the industry is offering 10 years, 100,000 miles bumper to bumper, then you better get on board or start risking sales. Another driver is sales and customer service. How about let’s be the first to come to the market with a 15 year, 200,000 mile bumper to bumper.

Getting back to my customers question. He feels that the factory stating something is good is as good as a guarantee that it’s good for a particular amount of time (that’s a lot of goods). Now where as that may seem logical, it isn’t, and here’s the reason for that. One of the determining factors in warranty time (or the useful life of a product) is a statistical notion called MTTF, or “Mean Time to Failure”. Essentially this is the average time that a given product lasts in the field before it fails. Suppose the MTTF for product “X” is 10 years, with a standard deviation of 2. That means that every unit X in the population fails between 4 and 16 years – this is because statistically, the total population is contained within 3 standard deviations of the mean. Given these numbers, I can make the guarantee that a given unit X will last 4 years failure free (DISCLAIMER: that’s a statement of statistical certainty. We wouldn’t necessarily place our warranty at 4 years given those numbers). However once the time a unit spends in the field enters into the scope of the bell curve, there becomes a statistical probability that a unit will fail at any point. In the case of a failure at 4 years, there’s a statistical probability of failure of 2.5%. It’s important to note that, even if “the factory” states a given product is good (that it operates/functions to the design specification), the product still carries with it (so to speak) that statistical notion. i.e. if at four years we evaluate a product and call it good, there is nevertheless a 2.5% chance that upon plugging into the customers application, it blows up. So long as a given product contains all or most of its original parts, it will be subject to this Mean Time clock, therefore the factory can NEVER make a guarantee unless all parts are replaced (turning back the clock) or the product is replaced outright.

Incidentally I had started off by telling the Project Manager, “If what you’re saying follows logically, then we as consumers could send a given product back to the factory after it’s warranty has expired (considering it’s good), and once they’ve made the statement, “all is well”, that means we immediately get full warranty coverage again. That’s preposterous.” He then responded, “Yeah, but you’d have to play for the evaluation.” True enough, “But the cost of the evaluation won’t be enough to offset the cost of potential warranty repairs that might occur in the future.” I replied. My hope here was for the guy to figure it out on his own that his reasoning was cracked, but I never got though to him until I gave the MTTF example.