you cannot inspect quality into a product

When defects start reaching a customer, the fastest looking fix is to add inspection. Put another set of eyes at the end of the line. Add a containment sort. Hire a few more inspectors. It feels like action, and it buys a little breathing room. It also almost never fixes the problem, because inspection does not change the process that made the bad parts. It only sorts them after the fact, and it does that imperfectly.

sorting is not improving

Inspection is a detection activity. At best it catches a defect before it ships, which is better than the alternative, but the defect was still made. The scrap was still produced, the machine time was still spent, the material was still consumed. All inspection does is decide which pile a part goes in. A process running at three percent defective and sorted to a clean shipment is still a process running at three percent defective. You have paid for the bad parts twice, once to make them and once to find them, and changed nothing about why they exist.

the math nobody likes

Inspection is also less reliable than people assume. Human visual inspection on a repetitive task rarely catches every defect. Detection rates in the seventy to ninety percent range are common, which means a meaningful fraction of what you are sorting for still slips through. When quality depends on catching defects at the end, you are trusting the least capable control in the whole system to be the last line of defense. Escapes are not a sign of careless inspectors. They are the predictable result of asking inspection to do a job it was never able to do well.

where quality is actually decided

Quality is designed and built long before the inspection station. It is decided in whether the process is capable, whether the measurement system can be trusted, whether the failure modes were understood and controlled, whether the operator has a clear standard and the means to hold it. By the time a part reaches final inspection, its quality is already fixed. The inspector is not adding quality. They are reporting on quality that was determined upstream. This is the whole reason prevention beats detection, and it has been understood for decades, yet the reflex under pressure is still to add more checking.

when inspection is the right call

None of this means inspection has no place. Containment is exactly the right response in the short term when a problem is reaching a customer. You sort to protect them while you find and fix the root cause. The failure is not using inspection. The failure is stopping there, letting the sort become permanent, and calling a containment action a solution. A temporary inspection that never gets removed is a quiet confession that the real problem was never fixed. The goal of every good containment is to make itself unnecessary.

the better questions

When defects appear, the reflex question is how do we catch them. The better questions are why is the process making them, and what would it take to make the process incapable of making them. That shift, from detection to prevention, from sorting to capability, is the difference between a quality function that firefights forever and one that actually reduces defects. A mistake-proof that makes the error impossible is worth more than any number of inspectors looking for it. A capable process with the measurement to prove it beats a sort every time, and it costs less over any horizon that matters.

Adding inspection will always feel like doing something about quality. Real quality work is the harder, slower job of fixing the process so there is less to inspect. The plants that understand the difference spend their energy upstream, where quality is actually made, and treat inspection as the safety net it was meant to be rather than the strategy it was never meant to become.