2014: Advent Quality: Widget Testing

by on December 11, 2014

I’ve mentioned testing several times, and not gone into detail; let’s dig into that.

Testing is one of the most obvious things that a quality department does. They take samples away, they come back with numbers, and sometimes you’re allowed to use the widgets and sometimes you’re not. From the outside, it all looks a bit mysterious, but testing is one of the most pragmatic parts of quality.

When you’re writing an SOP, you can in principle do it perfectly. You can write down every step of every process and describe them correctly. The SOP can be exact.

Testing cannot.

When you test a material or a product or a process, you have to accept that it cannot be done perfectly. You cannot have absolute certainty that something is correct. You will always be approximating. This is risk management, and it’s very like other fields of risk management – you are always asking yourself how likely it is that something will go wrong, and how bad the consequences could be if it does.

So why not test every widget, and be absolutely certain that the widgets are correct? Two main reasons: because it would be prohibitively expensive, and because most testing is destructive. If I test the breaking point of every widget I make, I now have a pile of broken widgets, which were all strong enough, and which are now all broken. I have no widgets left to sell. So I have to test some of my widgets, and sell the rest, in the faith that if some of them pass, they all would.

How much faith should you have?

Leave a Reply