Quality improves when defects become structured evidence

Overall Equipment Effectiveness (OEE) incorporates quality as one of three multiplicative components alongside availability and performance efficiency to detect production losses.[S06] However, quality ratios become incomparable when teams divide by different populations or confuse nonconforming units with total defect counts.[S01][S0601] Statistical quality literature differentiates between a defect (an individual nonconformity at a specific point) and a defective (a nonconforming unit containing at least one nonconformity).[S0601] A rejected shirt with a crooked collar, two skipped buttonholes and an oil stain represents 4 illustrative defects in total but 1 nonconforming garment.[S0601]

If an analyst subtracts 4 defects from 100 inspected garments and reports a 96 percent yield, when all 4 faults occurred on the same garment and 99 garments passed without fault, the calculation mixes incompatible units.[S01][S0601] Conversely, if 4 separate garments each contain 1 fault out of 100 inspected garments, the defect count is 4, the Defects Per Hundred Units (DHU) is 4.0, and the proportion of nonconforming units is 4.0 percent, giving a 96.0 percent first-pass yield.[S0601][S0602]

6.4 Four yield denominators serve distinct operational decisions

Even when numerator and denominator are restricted to physical garments, a cut, make, trim and finishing plant operates across sequential stages, and a yield ratio changes meaning depending on where its boundary is drawn.[S01][S02] GFT requires every quality metric contract to specify which of four denominators governs the calculation:[S01][S06]

1. Gate or Operation First-Pass Yield: The ratio of units passing a specific inspection gate on the first attempt without correction to total units inspected for the first time at that gate, owned by the supervisor of the preceding operations.[S02][S10][S0602]
2. End-of-Line Sewing First-Pass Yield: The ratio of garments accepted on initial inspection at the end of the sewing line to total garments reaching the end of the line for the first time during the shift, feeding the quality component of sewing-line OEE.[S06][S10]
3. Rolled Throughput Yield (RTY): The product of gate-level first-pass yields across all sequential gates from cutting through finishing, measuring the probability that a cut bundle traverses the factory without any repair or recutting stop.[S06][S10]
4. Order Cut-to-Shipped Yield: The ratio of total first-quality garments packed for shipment (including recovered pieces) to total garment equivalents cut at order release, capturing net material utilisation while hiding in-process rework loops.[S01][S02][S10]

Exhibit 27 applies these conservation rules to a derived 1,000-piece garment batch moving across four sequential gates: cutting inspection, in-line sub-assembly inspection, end-of-line sewing inspection and finishing inspection.[S01][S06][S10][S0601]

As Exhibit 27 demonstrates, each gate in this illustrative lot appears to perform well in isolation, with gate first-pass yields between 92.0 percent and 99.0 percent, and 971 first-quality garments packed out of 1,000 cut pieces (97.1 percent shipped yield after 29 pieces are scrapped).[S06][S10] Yet multiplying the four gate first-pass yields (0.990 times 0.950 times 0.920 times 0.979) reveals a Rolled Throughput Yield of only 84.7 percent.[S06][S10] More than 15 percent of the initial illustrative lot required manual repair or scrapping along the routing, generating 131 repair transactions. While Exhibit 27 uses assumed piece counts to illustrate the accounting structure rather than an observed GFT factory result, it proves why reading only the 97.1 percent shipped yield conceals the 110 repairs occurring inside sewing stations.[S02][S07]

A second denominator failure occurs when factories combine records from 100 percent in-line or end-of-line inspection with statistical lot acceptance sampling at pre-shipment audit.[S0603] NIST guidance emphasises that acceptance sampling decides whether a lot is acceptable for disposition, not to estimate the quality of the lot.[S0603] When an auditor inspects an illustrative random sample of 80 garments out of a 1,200-piece packed lot under an Acceptable Quality Limit (AQL) plan and finds 2 defective garments (2.5 percent), that result governs lot release and must never be pooled with the sewing line's 100 percent inspection log.[S01][S0603]