It is the last hour of a shift, and a batch of forty brackets fails final inspection over a hole that drifted out of position by six thousandths. Somebody has to decide, right then, whether the parts go to scrap, get reworked, or get dispositioned as use-as-is with an engineering concession. That decision, made under time pressure by whoever is standing closest to the bin, is where the cost of poor quality actually gets set. Everything that happens afterward — the paperwork, the root cause investigation, the conversation with the customer if the lot already shipped — is downstream of a call that took about ninety seconds. Shops that treat that moment as routine, rather than as a data point worth capturing consistently, tend to be the ones still asking why scrap rates never seem to move. A QMS software for manufacturing that survives an auditor's questions earns that description partly by making moments like this one impossible to lose track of.
What Cost of Poor Quality Actually Includes
Cost of poor quality is usually filed under four buckets: prevention, appraisal, internal failure, and external failure. Prevention and appraisal are the costs a shop chooses to spend — training, calibration, in-process inspection. Internal and external failure are the costs a shop is forced to spend after something has already gone wrong, and the gap between them is where most of the real money sits. A part scrapped at the saw costs a few dollars in material. The same defect discovered after five machining operations costs labor, tooling wear, and machine time that can never be recovered. Caught by a customer, it costs a corrective action response, a possible containment action across other lots, and whatever the relationship absorbs in trust. None of that shows up on a single line item, which is exactly why it is so easy to underestimate.
Why Scrap Numbers Lie When They're Tracked by Hand
Most shops that track scrap on paper or in a spreadsheet are tracking units, not causes. A tally of "12 parts scrapped this week" tells a supervisor almost nothing about whether the twelve parts failed for the same reason or twelve different ones. Without a consistent way to categorize nonconformances at the point of disposition — tool wear, fixture slip, operator error, incoming material defect — the data collapses into a single undifferentiated number that never points anywhere useful. Quality managers end up reviewing scrap totals in a management review meeting and reaching the same conclusion every quarter: it's roughly the same as last quarter. Nobody can say why, because nobody captured why in a form that could be aggregated.
The Feedback Loop That Never Closes
A defect found on the shop floor is only valuable if it changes what happens next time that part runs. That requires a loop: capture the nonconformance, link it to the operation and the equipment involved, route it to whoever owns the process, and confirm the corrective action actually happened before the next production run. In a lot of shops, that loop exists informally — a verbal conversation between the operator and the shift lead — and informal loops break under any kind of turnover or schedule pressure. The nonconformance tag gets filled out, filed, and never looked at again. Six months later the same fixture produces the same drift, and nobody remembers it happened before because there was no record to remember it by.
Turning Rework Into a Signal, Not Just a Loss
Rework has a strange reputation on most shop floors: it is treated as a nuisance to be absorbed quietly rather than a signal worth reading. But a part that needs rework is telling you something specific about a process capability gap, and if rework hours are logged against the operation and root cause rather than just against a general "quality hold" bucket, that signal becomes visible. Shops that get disciplined about this find that a small number of operations account for a disproportionate share of total rework hours — which is useful information, because it tells engineering exactly where a fixture redesign or a tooling change would pay off fastest. Without that granularity, rework just looks like background noise, an unavoidable cost of doing business rather than a prioritized list of process improvements waiting to happen.
Building the Discipline Without Adding Bureaucracy
None of this requires more inspection or more paperwork. It requires the same disposition decisions that already happen at the bin, captured in a structured way instead of a verbal one, with enough consistency that the data can be rolled up later. That's a cultural shift as much as a system one — operators and inspectors have to believe the extra thirty seconds of categorizing a defect is worth their time, which usually only happens once they see the data actually used to fix something. The first time a recurring tool-wear pattern gets caught and addressed because someone pulled a report instead of relying on memory, the habit tends to stick on its own. Cost of poor quality never disappears entirely — some scrap and some rework is the price of running real production — but it stops being an unexamined line item and starts being a list of specific, addressable problems, which is a very different thing to manage against.