Key takeaways
Short answer: Defect and defective sound interchangeable but mean different things in quality, and confusing them corrupts your metrics. A defect is a single nonconformity — one flaw. A defective is an entire unit judged unacceptable, no matter how many defects it contains. A unit with three scratches has three defects but is one defective. This matters because some metrics count defects (how many flaws) and others count defectives (how many bad units), and they answer different questions. For the metrics this feeds, see yield vs scrap rate.
A defect is a single instance of nonconformity — one specific way a unit fails to meet a requirement. A scratch on a painted panel is a defect. A missing label is a defect. An out-of-tolerance dimension is a defect. The key is that a defect is counted per occurrence: a single unit can carry many defects, each one a separate nonconformity. Counting defects answers how many things are wrong, which is useful when you care about the total burden of nonconformities — for instance when several minor flaws per unit are common and you want to track and reduce their overall rate, not just whether a unit passed or failed.
A defective — or a defective unit — is a whole unit that fails to meet requirements and is therefore unacceptable, regardless of how many individual defects caused that judgement. A unit with one critical defect is one defective; a unit with five minor defects is still one defective. Counting defectives answers how many bad units did we produce, which is the question that maps directly to scrap, rework, and what the customer experiences. The unit is the thing you ship, sell, or bin, so for most outcome-focused quality metrics the defective — pass or fail at the unit level — is what matters, not the internal tally of flaws.
The difference is not pedantry; it changes which metric and which statistical tool you use. If you count defects, you are working with nonconformities per unit, and the right control charts are those for counts of defects (c-charts and u-charts in statistical process control). If you count defectives, you are working with the proportion of bad units, and the right charts are those for defectives (p-charts and np-charts). Pick the wrong frame and your data misleads: tracking only defectives hides that units are accumulating more and more minor flaws as long as they still pass, while tracking only defects can obscure how many actual units are unsellable. The question you are asking dictates which you count.
A shift produces 500 painted panels. Inspection finds 90 individual flaws — scratches, runs, thin spots — spread across the units. But those 90 defects are not evenly one-per-panel: 30 panels carry them, some with one flaw, several with three or four. So the numbers tell two stories. Defects: 90 nonconformities, a rate of 90 per 500 units, or 0.18 per panel. Defectives: 30 unacceptable panels, a defective rate of 6%. Report only the 6% defective and you miss that quality is degrading toward multi-flaw panels; report only the 90 defects and you obscure that 470 panels were perfectly fine. You need both numbers to see the whole picture, and you must not confuse one for the other.
Track defectives when the outcome is what matters most — scrap, rework, yield, customer-received quality — because the unit, pass or fail, is what gets shipped or binned. Track defects when you want to understand the burden and sources of nonconformity, especially where multiple minor flaws per unit are normal and you are driving them down before they tip units into being defective. In practice, mature quality systems watch both: defectives for the bottom-line result, defects for the early, granular signal of where problems are concentrating. The mistake is collapsing them into one loose number called quality and losing the ability to tell a flaw problem from a bad-unit problem.
The quality factor of OEE works at the level of defectives, not defects: it counts good units versus not-good units, because OEE is about how much sellable output you produced. A unit with three defects subtracts one unit from the good count, the same as a unit with one — OEE cares whether the unit is acceptable, not how many flaws it has. That makes the defective the right unit of account for the yield and scrap behind OEE. But defect-level data is still valuable underneath it: tracking defects is often the earliest warning that your defective rate, and therefore your OEE quality factor, is about to slip.
Fabrico counts good versus not-good units at the point of production — the defective-level view that drives the quality factor of live OEE and the scrap and yield numbers behind it. By trending that result over time and tying rejects to reason codes, it also helps surface the defect-level patterns building underneath — the recurring flaw that is pushing more units over the line into defective. Outcome and early signal in one place. Book a demo to see unit-level quality drive your OEE.
A defect is a single instance of nonconformity — one flaw on a unit. A defective is a whole unit that fails to meet requirements, regardless of how many defects it has. One unit can carry several defects but counts as one defective.
Yes. A single unit with, say, three scratches has three defects but is counted as one defective unit. The defect count is per nonconformity; the defective count is per unacceptable unit.
Because different charts apply. Counting defects (nonconformities) uses c-charts and u-charts; counting defectives (bad units) uses p-charts and np-charts. Using the wrong chart for what you are counting breaks the statistics.
Track defectives for outcome metrics like scrap, yield, and customer quality, since the unit is what ships. Track defects to understand the burden and sources of nonconformity. Mature systems watch both — defectives for the result, defects for the early signal.
OEE's quality factor counts defectives — good units versus not-good units — because it measures sellable output. A unit is either acceptable or not, regardless of how many defects it contains. Defect-level data is still useful as an early warning underneath.
Anmeldung
Blog
