AQL Inspection Software for Apparel Importers

A worked example: 3,200 garments at AQL 2.5 / 4.0

This is the calculation QualityIris runs the moment an inspector opens a final audit. Nothing is typed into a sampling table by hand, and nothing is looked up in a printed standard on the factory floor.

Severity is not fixed — switching rules change the numbers

An AQL scheme only delivers the protection its number implies when the ISO 2859-1 switching rules are applied. Inspecting every garment lot on normal severity forever quietly gives the buyer far less cover than the stated AQL suggests, which is why clause 9.3 of the standard makes switching mandatory rather than optional.

On the same 3,200-garment lot above, tightened severity keeps the 125-garment sample but cuts major acceptance from 7 to 5 and minor acceptance from 10 to 8. Reduced severity moves the other way, dropping the sample to 50 garments once the apparel factory has earned it through a sustained clean run.

• Normal to tightened Two rejected garment lots within the last five move that factory to tightened. QualityIris tracks the state per tenant, apparel factory, buyer and AQL pair.
• Tightened back to normal Five consecutive accepted lots restore normal. If tightened is not cleared within five lots, the standard calls for inspection to be discontinued until quality improves — and the app says so.
• Advisory, never silent QualityIris recommends the severity and shows the lot history behind it. The inspector can override, but must record a written reason that prints on the buyer-facing audit PDF.

Isolated lots: ISO 2859-2 LQ sampling

AQL sampling assumes a continuing series of lots from the same apparel factory. A one-off order, a first production run from a new garment supplier, or a single lot pulled from a warehouse has no series to switch against, so an AQL plan is the wrong instrument.

For those cases QualityIris offers ISO 2859-2 Procedure A, which is indexed by limiting quality (LQ) rather than AQL. LQ is defined from the consumer's side: a lot at the LQ percentage has only about a 10% chance of being accepted. Pick the LQ your buyer is willing to tolerate and the app returns the plan — switching is deliberately hidden in LQ mode, because an isolated lot has no history to switch on.

What to look for when comparing AQL inspection software

• Does it run the audit, or only the arithmetic? A sampling calculator returns a number and stops. Check that defect capture, garment photos, the accept/reject verdict, signatures and the buyer PDF all come out of the same record.
• Does it implement the switching rules? Most tools ship the normal-severity table only. Without tightened and reduced plans and the state tracking behind them, the AQL scheme is incomplete.
• Does it work with no signal? Garment factory floors are concrete, steel and dead zones. If sampling, defect capture and photos need connectivity, inspectors will fall back to paper and retype it later.
• Can the plan differ per buyer? Apparel importers rarely share one AQL. Per-buyer major/minor limits, inspection levels and defect classes should be configured once and applied to every PO automatically.
• Do you keep the raw data? Signed PDFs are the deliverable, but the defect-level data is the asset. Confirm you can export it as CSV without asking the vendor.