AQL sampling plan for final random inspection (fri): what importers book before shipment: part 5

Key takeaways

1. Acceptance and rejection numbers are derived from the lot size, the inspection level, and the agreed AQL values, never from negotiation on inspection day.
2. Critical defects conventionally carry zero tolerance, so one confirmed critical finding fails the lot unless the contract states otherwise.
3. When a critical finding is disputed, sending the failed sample to an accredited laboratory keeps the argument technical instead of commercial. ^[2]
4. Pre-shipment inspection works like other conformity procedures: it demonstrates that goods meet the applicable technical requirements before they move. ^[1]
5. Writing the level, the three AQL values, and the re-inspection rule into the purchase order lets every later booking inherit them automatically.

Most inspection disputes are not really about quality. They are about arithmetic that was never agreed. The buyer imagined one defective piece would fail the shipment, the factory imagined a handful would be tolerated, and the inspector arrived with a sampling table neither side had read. By the time the cartons are open, every borderline defect becomes a negotiation, and negotiations on inspection day favor whoever argues loudest, not whoever is right.

The fix costs nothing: agree the sampling plan in writing before the goods are finished. A complete plan has four elements, and each fits in one line of the purchase order. First, the inspection level, usually general level II for consumer goods. Second, the lot definition, meaning which goods the sample represents. Third, the AQL values for critical, major, and minor defects. Fourth, what happens on failure, including who pays for sorting and re-inspection.

This article, part 5 of the series on AQL sampling plans for final random inspection, walks through how acceptance numbers are derived from those four elements. It shows a worked table of sample sizes and accept-reject points, explains why critical defects normally carry a zero-tolerance clause, and gives a booking clause you can copy into your next order. Read it before the goods are finished, not after the inspector calls with bad news.

The acceptance number is the whole decision

An acceptance number answers a single question: how many defective pieces in the sample still allow the lot to pass. If the plan says Ac 5 Re 6 for major defects, finding five or fewer major defects passes the lot, and finding six fails it. There is no middle outcome, no partial pass, and no room for the factory to argue that the sixth defect was minor. That bluntness is the point. Both sides know the pass-fail line before a single carton is opened.

The number itself comes from published single-sampling tables that pair each sample size with accept-reject points per AQL. Larger lots draw larger samples, and larger samples get proportionally higher acceptance numbers, which keeps the statistical protection roughly constant across order sizes. A buyer who understands this relationship stops fearing large orders, because the plan automatically scales the scrutiny to match the quantity at risk.

The practical consequence is that the only real decision the buyer makes is upstream: which inspection level and which AQL values to write into the contract. Everything downstream is arithmetic. Importers who grasp this spend their energy on the clause rather than on the inspection-day phone call, and their shipments fail or pass for reasons they chose themselves.

Worked example: general level II with AQL 2.5 major and 4.0 minor

What each defect class means in practice

Classification decides the verdict before counting starts, because each class is judged against its own acceptance number. Two useful habits keep classification honest. First, attach a defect photo library to the specification so the inspector and the factory recognize the same boundary. Second, review any defect the factory disputes against the written definition, not against the shipment schedule.

Critical defects: the zero-tolerance class

A critical defect makes the product unsafe or unusable, or breaks a legal requirement: a live accessible wire, a sharp edge on a toy, a missing safety device. The conventional AQL for this class is effectively zero, meaning one confirmed critical finding fails the lot on its own. This severity exists because no sampling statistic can make an unsafe product acceptable, and buyers should never let a factory negotiate a critical finding down to major.

Major defects: the functionality and appearance class

A major defect impairs use or marketability without creating a safety hazard: a garment with a broken zipper, an electronic item that will not charge, a print run with the wrong logo. AQL 2.5 is the common contractual value for consumer goods, which at level II tolerates a small count that grows with the sample. The boundary between major and minor is where most inspection-day arguments happen, so define the frequent borderline cases with photos in the specification.

Minor defects: the workmanship class

A minor defect departs from the specification without affecting use: a slightly uneven stitch line, a faint scuff on inner packaging, a barely visible molding mark. AQL 4.0 is the usual value, and minor findings often dominate the defect tally without threatening the verdict. Track them anyway across shipments, because a rising minor-defect trend is frequently the earliest signal that a factory process is drifting.

Writing the sampling clause into the booking

1. Define the lot in one sentenceState the finished quantity and how the lot is defined, for example all cartons of one purchase order packed from one production batch. If several batches ship together, define each as its own lot so every batch earns its own sample and verdict.
2. Name the inspection levelName the general inspection level, normally level II, and add any special-level checks for label or measurement attributes only where the specification justifies them. One level for the whole visual inspection keeps the arithmetic clean.
3. Fix the three AQL numbersWrite the three AQL values explicitly, for example critical zero tolerance, major 2.5, minor 4.0, and reference the single-sampling normal inspection tables as the source of the accept-reject pairs. Explicit numbers beat phrases like standard AQL, which every party interprets differently.
4. Agree the failure path before failureDescribe the failure path: hold the lot, sort or rework the defective portion, and re-inspect the affected lot under a fresh booking. Naming who bears sorting and re-inspection costs in advance removes the most bitter post-failure argument.
5. Send the clause with the bookingMany buyers ask an independent pre-shipment inspection team to apply the clause exactly as written, because a referee both sides accept removes the argument about borderline defects. Send the clause with the booking request, attach the defect photo library, and confirm that the inspector will sample only from finished, packed cartons.

When the lot fails: what the numbers obligate

A rejection verdict obligates action, not despair. The standard path is containment first: the failed lot is held, the defective portion is sorted or reworked, and the corrected lot faces a fresh sample under a new booking. Each stage is documented with the defect counts that triggered it, so the shipment file shows why goods were held and what changed before release. This discipline matters commercially because the next order from the same factory starts from the recorded history, not from fading memories of an argument.

There is a wider logic behind this strictness. Conformity assessment procedures, including pre-shipment inspection, exist so that technical requirements are verified before goods cross borders rather than litigated after arrival. ^[1] Treating the acceptance number as a binding gate rather than an opening offer aligns a private purchase order with that public logic. And when a failure turns on a testable claim, such as a restricted substance or a safety parameter, routing the disputed sample to an accredited laboratory settles the question with data both sides can respect. ^[2]

Frequently asked questions