What is a pre-shipment inspection?
A pre-shipment inspection (PSI) is an on-site check of a random sample drawn from a nearly finished production batch, performed by an independent party at the factory or warehouse, before you pay the balance and before the goods leave. The purpose is simple: to see what you actually bought while changing it is still possible and still cheap.
A PSI normally covers six areas: quantity packed, conformity to the approved sample and specification, workmanship and visible defects, on-site tests where feasible such as function, pull, leak or dimensional checks, packaging, labels, barcodes and carton markings, and weights and dimensions to validate freight calculations.
When exactly should the inspection happen?
The working rule is: once at least eighty percent of the order is produced and a reasonable portion is packed in final export cartons. Inspect earlier and the sample does not represent the batch; inspect later, after the container is booked, and you have lost your negotiating position. Always anchor the inspection date to the production schedule, never to the shipping schedule.
PSI should also not be your only control point. The normal sequence is incoming material check, then during-production inspection (DUPRO) at roughly fifteen to thirty percent completion, where systematic errors can still be corrected, then PSI at completion, then container loading supervision so that what was inspected is what actually gets loaded.
What is AQL, and why not inspect every unit?
AQL stands for Acceptable Quality Limit: the highest proportion of defective units at which a batch can still be considered acceptable. One hundred percent inspection is theoretically possible but slow, costly, and for many products economically pointless. So the industry uses sampling inspection under ISO 2859-1 (the equivalent of MIL-STD-105E), which converts lot size and AQL into three concrete numbers: sample size, accept number (Ac) and reject number (Re).
One point deserves precision: AQL is not a quality target and not a promise that the batch contains that many defects. It is a statistical decision limit. Sampling gives a high probability of accepting good lots and rejecting bad ones; it does not give certainty.
How is the sample size determined?
- Establish the lot size, the total quantity presented for inspection.
- Choose the inspection level, normally General Level II, which is the default.
- Choose an AQL for each defect class. In consumer goods a common set is zero for critical, 2.5 for major and 4.0 for minor defects.
Lot size and inspection level give a code letter, the code letter gives the sample size, and the sample size crossed with the AQL gives the accept and reject numbers.
What does the AQL table look like at General Level II?
The table below is simplified for illustration. It shows single sampling plans for normal inspection following ISO 2859-1 logic at General Inspection Level II. Values are shown as accept/reject: the lot is accepted if defects found are at or below the accept number, and rejected once the reject number is reached. The binding reference is the standard itself.
| Lot size | Code letter | Sample size | AQL 1.5 | AQL 2.5 | AQL 4.0 |
|---|---|---|---|---|---|
| 51 – 90 | E | 13 | 0/1 | 1/2 | 1/2 |
| 91 – 150 | F | 20 | 1/2 | 1/2 | 2/3 |
| 151 – 280 | G | 32 | 1/2 | 2/3 | 3/4 |
| 281 – 500 | H | 50 | 2/3 | 3/4 | 5/6 |
| 501 – 1200 | J | 80 | 3/4 | 5/6 | 7/8 |
| 1201 – 3200 | K | 125 | 5/6 | 7/8 | 10/11 |
| 3201 – 10000 | L | 200 | 7/8 | 10/11 | 14/15 |
| 10001 – 35000 | M | 315 | 10/11 | 14/15 | 21/22 |
| 35001 – 150000 | N | 500 | 14/15 | 21/22 | 21/22 |
Technical note: in the official tables some cells at the extremes carry arrows, meaning you move to the sampling plan above or below and the sample size changes accordingly. The standard also defines tightened and reduced inspection plans applied according to a supplier's performance history. So write the level and the standard explicitly into your contract rather than the bare word "AQL".
What are the inspection levels and how do they differ?
| Level | Sample size | Typical use |
|---|---|---|
| S-1 to S-4 | Very small to small | Expensive or destructive tests, or narrow checks such as a single dimension |
| General Level I | Smaller than default | Proven supplier, low-risk product, or cost pressure |
| General Level II | Default | General use for most consumer goods |
| General Level III | Larger | New supplier, history of problems, sensitive or high-value product |
How do you choose the right AQL for your product?
- Critical defects: AQL zero is standard, meaning a single critical defect in the sample rejects the lot. This applies to safety hazards such as electric shock risk or small parts that a child could swallow.
- Major defects: 2.5 is common for general consumer goods, tightened to 1.5 for higher-value items or demanding retail channels.
- Minor defects: 4.0 is common, adjusted up or down according to how sensitive your market is to appearance.
- Lower AQL numbers mean stricter inspection, higher rejection probability and usually higher production cost. Choosing an AQL is a commercial decision, not purely a technical one.
What must be written into the contract before inspection?
- The inspection level, the AQL per defect class, and the reference standard (ISO 2859-1).
- A written definition of what counts as critical, major and minor for this specific product.
- The approved golden sample, signed by both parties, and where it is held.
- Which on-site tests are required, how many units are tested, and the pass criteria.
- The buyer's right to appoint an independent inspector and that inspector's right of access.
- Balance payment tied to the inspection result, and who pays for re-inspection after a failure.
What are the limits of pre-shipment inspection?
Sampling does not catch every defect, and it does not replace accredited laboratory testing where a market requires a certificate of conformity. It also sits at the end of the chain: if the raw material was wrong from the start, the outcome is a rejected lot after its full cost has already been spent. That is why during-production inspection remains the cheapest possible point of intervention.
Practical takeaway
Define acceptance criteria before production, not after: an inspection level, an AQL per defect class, a signed golden sample, and payment tied to the result. Then spread control across stages instead of concentrating it in one moment. At ALSHUMUL we build inspection plans on exactly that structure: DUPRO during production, PSI under ISO 2859-1, then loading supervision, with a photographic report delivered before the payment decision.
