Defects as waste: counting the full bill for work done twice

Scrap value is the smallest line on the bill

The material written off is easy to count and rarely the main loss. Add the capacity consumed making the part, which matters enormously if it was consumed at the resource limiting the plant's output, since that hour cannot be recovered. Add the disruption of resequencing to remake it, the expediting of a replacement, the handling and storage of the quarantined stock, the time supervisors and engineers spent on it, and any inspection introduced afterwards that is still running. Assembling that picture once, for a representative defect, changes how a management team talks about quality far more than a scrap report ever does.

Where it is found determines what it costs

The same fault detected at the machine that created it costs a part and a few minutes. Found at final test, it costs everything added since, plus disassembly. Found at the customer, it costs a return, an investigation, possible sorting of their stock, credit notes, and a position on a supplier scorecard that takes a long time to recover. This escalation is the practical argument for checking close to the point of creation rather than concentrating inspection at the end, and it is worth working through with real examples from your own plant rather than as a general principle.

Rework is the version that vanishes from the numbers

Once a recurring correction has a bench, a person and an allowance in the standard hours, it stops being counted as a defect at all. The plant has absorbed the fault into its normal cost of doing business, and the improvement pressure disappears with it. Look for dedicated rework areas, operations described as touch-up or fettling, and routings containing a step whose only purpose is to correct the previous one. Anything that has been normalised in this way should be brought back into the defect figures, because otherwise the process appears capable when it is not.

Choosing which defects to attack first

Frequency alone points at the small irritations; value alone points at rare expensive failures; disruption points at whatever most recently caused an argument. A workable ranking combines how often it occurs with what it costs to put right and whether it consumes constrained capacity. All of that depends on defect coding being consistent between shifts and areas, and in most plants it is not: the same fault appears under several descriptions and the largest category is a general miscellaneous one. Cleaning up the coding is unglamorous and usually the precondition for any sensible prioritisation.

The handover to the quality process

Improvement work identifies, quantifies and prioritises. What happens next, namely establishing cause, containing affected stock, agreeing corrective action and verifying that it worked, belongs to the quality function with its own methods and records. Blurring the boundary produces two familiar failures: an improvement team conducting an amateur investigation and settling on the first plausible explanation, or a quality department pursuing a defect that costs the business very little while a frequent, disruptive one is ignored. Agree who hands what to whom, and make sure the resulting actions live on one list rather than two.

Frequently asked questions

Should operators be allowed to stop the line when they find a defect?

Yes, provided the plant has decided in advance what happens next. Stopping is only useful if someone responds quickly, if the reason is recorded, and if nobody is criticised for calling a halt that turns out to be unnecessary. Where response is slow or the reaction is hostile, operators learn to pass the problem on, and the plant loses the earliest and least costly point of detection. The authority to stop and the obligation to respond have to be introduced together.

How do we account for defects found at a customer?

Capture the direct costs of return, replacement and investigation, then record the effects that never reach the ledger: the sorting your people did on the customer's site, the management attention consumed, the additional inspection imposed on subsequent shipments, and any commercial position lost. Keep those as a written record attached to the event rather than an estimate in a report. When the same customer raises a second issue, that record is what justifies serious investment in prevention.

Is inspection itself a waste?

Checking adds no value to the product, so in strict terms yes, but it is often necessary and sometimes contractually required. The useful distinction is between checks that catch a problem the process cannot yet avoid and checks that continue out of habit long after the cause was fixed. Give every added inspection an owner and a point at which it will be reviewed, otherwise containment measures accumulate permanently and the plant slowly pays for faults it no longer has.