Defective units and defects are different counts
A defect is one failure to meet a requirement: a scratch, a short seam, a loose leg. A defective unit is a unit that has one or more defects. One chair with a scratch and a wobbling leg is two defects but one defective unit. That is why the same batch gives two different rates:
| Measure | Formula | ISO 2859-1 term |
|---|---|---|
| Percent defective | Defective units ÷ units inspected × 100 | Percent nonconforming (items) |
| Defects per hundred units (DHU) | Total defects ÷ units inspected × 100 | Nonconformities per 100 items |
Percent defective can never exceed 100. DHU can, because one unit can carry many defects. In ISO 2859-1, AQL values above 10 are used only with nonconformities per 100 items, for that reason.
Worked example
An illustration, not a real report: an inspector samples 125 cushions. Six cushions have faults. Four have one defect each, one has two and one has three, so there are nine defects in total.
- Percent defective: 6 ÷ 125 × 100 = 4.8%.
- DHU: 9 ÷ 125 × 100 = 7.2 defects per hundred units.
With a sample of 125, the major-defect limits at AQL 2.5 are accept 7, reject 8. Counted as defective units (6), the lot passes. Counted as defects (9), it fails. Neither count is wrong. The counting rule has to be agreed before the inspection, as the Ac / Re page explains.
AQL is not the defect rate of your lot
Buyers often read AQL 2.5 as "2.5% defects allowed in my order". It is not. The AQL is the worst process average that the sampling scheme tolerates over a series of lots. A single lot with more defects than that can still pass, and a lot with fewer can fail, because the decision rests on a sample. The defect rate found in the sample is an estimate of the lot, with sampling error, not a measurement of every unit.
Defect rates are also counted per class. A sample can show a low overall rate and still fail on one critical defect, because critical defects usually have an acceptance number of zero.
How Clear To Load reports it
Our reports give the count of critical, major and minor findings in the sample against the accept and reject numbers for each class, with the counting rule written into the checklist before the visit. Where a fault applies across a whole style, such as a colour variation, it can be recorded once for that style, as in our Sample 05 report. If you track supplier performance, ask for both rates on each report so trends are comparable across orders. To see the limits for any lot size, use the AQL calculator.