AQL sampling plan for sample evaluation: what importers book before shipment: part 2
Approving a sample answers whether the design is right; sampling bulk production answers whether the factory reproduced it. Importers who book a sampling-based verification before shipment get a quantified accept-or-reject decision while corrections are still possible. Those who skip it learn about reproduction failures from customer complaints instead. The sampling plan is the bridge between an approved sample and shippable goods, and this article shows how to build it.
Key takeaways
- Verify bulk reproduction with a sampling plan agreed before production finishes, so the acceptance rule is settled while corrections are still possible.
- Base acceptance decisions on sampling theory in which a random sample represents the lot within quantified risks. [1]
- Follow factory acceptance logic: prove conformance where correction is still possible, before goods ship. [2]
- Tie every booking to the approved sample round and the written specification so inspectors judge correctness, not just consistency.
- Define defect classes and acceptance numbers in the booking so pass and fail are mechanical, not negotiated on the day.
Sample approval settles the design question, but bulk production introduces a new one: did the factory reproduce the approved configuration across thousands of units. Answering that by checking every unit is rarely economical, which is why importers verify bulk lots with statistically grounded sampling plans booked before shipment.
Sampling works because it converts a vague hope into a quantified decision rule. Instead of glancing at a few cartons, the buyer defines how many units to draw, how to classify defects, and how many defects trigger rejection. The factory knows the rule in advance, the inspector applies it mechanically, and both sides accept the outcome because the mathematics were agreed before the goods existed.
This article, the second part of the sampling series, explains the booking importers make before shipment: which sampling logic applies, how acceptance levels are chosen, and how the plan connects back to the approved sample and forward to the shipping decision.
What the pre-shipment sampling booking decides
A pre-shipment booking sends an inspector to draw units from the finished lot and judge them against the approved sample and specification. The booking states the lot size, the sample size, the defect classification, and the acceptance numbers, so the factory can prepare honestly and the buyer receives a decision, not an impression. Typical coverage includes quantity verification, workmanship, function on drawn units, labeling, packaging, and loading supervision where relevant.
Timing is part of the decision. Book the inspection when the lot is finished and packed enough to draw representatively, but early enough that a failed lot can be sorted or reworked before the vessel sails. Many buyers coordinate the schedule through UTS UTS so the booking, the reference sample round, and the acceptance criteria travel together in one instruction. A booking that arrives without the baseline wastes the visit.
The output is a report with measured findings and a clear recommendation against the agreed rule. Vague commentary helps nobody; classified defects with counts, photos matched to the reference sample, and a stated accept-or-reject outcome let the buyer act the same day. Ask for this structure explicitly when booking.
How acceptance sampling logic works
Acceptance sampling evaluates a lot by inspecting a random subset and applying a decision rule with known producer and consumer risks. A smaller sample costs less but leaves more uncertainty; a larger sample costs more and discriminates better between good and bad lots. The statistical foundation of drawing inferences about a lot from a properly selected sample is well established in quality engineering, where sampling plans balance inspection cost against decision risk. [1]
Randomness is non-negotiable. Units must be drawn from across the lot, across cartons, pallets, and production shifts, because factory staff naturally present their best cartons first. An inspector who accepts presented cartons inspects a showcase, not a lot. State random drawn in the booking and expect the report to describe how the draw was performed.
Defect classification drives the mathematics. Critical defects that threaten safety typically allow zero occurrences, major defects affecting function or appearance get tight acceptance numbers, and minor defects get looser ones. Classify before booking, with examples and photos, because arguing about whether a flaw is major or minor during the inspection poisons the decision.
Choosing sample size and acceptance numbers
| Situation | Plan choice | Reasoning |
|---|---|---|
| First order from a new factory | Larger sample, tight acceptance | Unknown process capability needs stronger evidence |
| Repeat order, stable history | Standard sample, normal acceptance | Proven capability justifies normal scrutiny |
| High-risk category or retailer penalties | Tightened inspection or full check of key traits | Failure cost exceeds inspection cost |
| Small lot below plan minimum | Full check of critical traits | Sampling mathematics needs minimum quantities |
| Failed lot after rework | Tightened re-inspection, new random draw | Reworked lots have higher defect probability |
| Mixed SKUs in one shipment | Separate plan per SKU family | Different baselines cannot share one decision rule |
Proving conformance before shipment
The logic of verifying conformance at the factory, before goods leave, mirrors the established practice of testing equipment at the maker's works before acceptance. In both cases the principle is identical: prove that what was built matches what was agreed while correction is still cheap, rather than discovering deviation after delivery when leverage is gone. [2] Pre-shipment sampling applies this principle to bulk consumer and industrial goods.
Make the proof complete by covering the unglamorous items: quantity against the order, carton count and markings, barcode scannability, manual and label versions, and packaging integrity. Many failed shipments pass every product test but fail on quantity or labeling, and these checks cost minutes during the visit while saving weeks of claims afterward.
Photograph findings against the reference sample rather than in isolation. A photo of a defect beside the approved unit communicates instantly across languages and time zones, while a lone close-up invites debate about severity. Require side-by-side evidence in the booking so the report arrives ready to forward to the factory.
From sampling result to shipping decision
Treat the acceptance number as binding. A lot that exceeds it fails, even if the factory promises the next shipment will be better, because waiving the rule teaches the factory that rules are decorative. Either reject, sort with re-inspection, or grant a documented conditional release with explicit limits, never a verbal wave-through that nobody can reconstruct later.
Failed lots need a corrective loop, not just sorting. Ask why the defect occurred, what process change prevents recurrence, and how the factory will demonstrate the fix on the next lot. Sorting removes bad units; only process correction reduces the next lot's defect rate. Buyers who skip this loop pay for the same failures order after order.
File the sampling report with the approval file, keyed to the same round number. Over time this archive reveals which suppliers reproduce reliably, which characteristics fail repeatedly, and where the sampling plan itself needs tightening. Data accumulated across shipments turns isolated inspections into supplier intelligence.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
Frequently asked questions
How do the approved sample and the sampling plan work together?
The approved sample plus the written specification define the acceptance criteria, and the sampling plan defines how many units get checked. Inspectors compare drawn units against the sample and specification, classify any defects found, and apply the acceptance numbers. Without the sample baseline, sampling only measures consistency, not correctness.
What should buyers do when a supplier repeatedly fails sampling?
Tighten the plan: use a larger sample, stricter acceptance numbers, or shift critical characteristics to full checks. A service that repeatedly fails sampling has a process problem, and repeating the same plan without addressing the cause wastes inspection fees while defective goods keep shipping.
Can sampling replace sample evaluation, or the other way round?
They answer different questions. Sample evaluation asks whether the design is right; pre-shipment sampling asks whether bulk production reproduced it. Skipping the first means sampling measures conformity to an unverified design, and skipping the second means trusting that production matched the sample without evidence.
What does a standard pre-shipment booking cover?
Quantity verification, workmanship and appearance against the sample, functional checks on drawn units, packaging and labeling verification, and container loading supervision where applicable. The exact scope follows the product risk, but every booking should state the reference sample round and the defect classification in writing.
Put this analysis on your district.
347 legislators already have. A 30-minute diagnostic with a Coalition principal is where it starts.