How to reduce the cost of poor quality (COPQ) in manufacturing
The cost of poor quality has two kinds of cause. There is the machine and process side: a worn tool, drifting parameters, bad material. Your MES and QMS already catch most of that, and they catch it well. Then there is the decision side: the call a supervisor makes in the moment to run, hold, adjust, or pass a part. That share rarely gets measured, so it rarely gets fixed, and that is where the fastest unclaimed gains sit.
COPQ is large and most of it is hidden. Across typical operations it runs 15 to 20 percent of sales (ASQ), mostly rework, scrap, and the time spent finding defects after the fact. On a 300 crore business that is 45 to 60 crore a year. Some of that is machine and process, and the right tools already work on it. Some of it traces back to decisions, and that part stays recoverable once you can see it.
Start with the machine and process side, because it is real
A genuine share of poor quality is mechanical. A tool wears past its limit. A parameter drifts out of band. A batch of material comes in off-spec. This is what statistical process control, your MES, and your QMS are built for, and they do the job. They flag the drift, they hold the lot, they keep the audit trail. If your machine-caused COPQ is still high, the answer is to use those systems harder, not to look elsewhere. Credit where it is due: that layer is solved by tools that are good at it.
The share your quality system was never built to see
Once the machine side is accounted for, a stubborn pile remains. It gets labelled "operator error" and closed. That label is a dead end, because it names a person instead of a decision. The real cause is the call that person made: carrying on when they should have stopped, adjusting when they should have held, passing a part they should have pulled.
That call is Decision-at-the-Edge, the call a frontline supervisor makes in the moment. It is the hidden driver of rework, downtime, and escalation. It is the point where a factory's day is actually won or lost, and the one most management systems never measure. Your QMS records that the part failed. It does not record that the failure started with a judgment call ten minutes earlier. The hub guide on why supervisors escalate defines this in full, and the guide on rework caused by wrong calls works the same idea through the rework log.
Why more inspection does not lower COPQ
The first instinct is to add an inspection step. It catches the defect later, after the time and material are already spent. It does not stop the wrong call that caused it. You pay twice: once to make the bad part, again to find it, and inspection is itself a cost-of-quality line. The only way to actually lower the number is to fix the decision upstream, on the line, before the part is made. First-time-right is a judgment problem for exactly this reason.
How to cut the decision-driven share of COPQ
- Go through your COPQ and rework log and split each item: a machine or material fault, or a call someone made. Be straight about the "operator error" pile, because that is where the decision cost hides.
- For the decision pile, name the specific calls behind the defects, one situation at a time. Run or hold. Adjust or wait. Pass or scrap.
- Tell apart the calls a person knows but botches under pressure from the calls they have never learned to make, then close each person's specific gap instead of training everyone on everything.
- Track the decision-tagged share of COPQ as it falls, measured on real line data, not on a course-completion certificate.
Steps three and four are the Diagnostic Gap Model: the gap between the recognition ceiling, what a person knows is the right action in a given scenario, and the behavior floor, what their real work data shows they actually do. The size and shape of that gap prescribes the practice. You do not train everyone on everything; you close each person's specific gap. The hub guide covers the model in detail.
How you will know it worked
The decision-caused share of your COPQ falls, and you can point to it specifically because you separated it out at the start. The machine side keeps being handled by the systems built for it. The decision side, the part nobody was measuring, finally moves. That is the number that shows up on your P&L.
Where this fits
Cutting the decision-driven share of COPQ is one piece of building decision quality on your floor, and it is where decision quality meets your quality cost directly. The full picture is in the hub guide on why supervisors escalate and how to fix it. PDGMS HCD is the system that builds the judgment on your real floor situations and proves it on the COPQ number.