Pre-Shipment Inspection & AQL Sampling: What to Check at Origin Before You Pay

A pre-shipment inspection (PSI) checks a finished, packed order at the factory — before it ships and before the balance is paid — by drawing a random sample under a statistical plan and judging the lot against agreed accept/reject limits. The standard behind that sample is ISO 2859-1 (the international equivalent of ANSI/ASQ Z1.4), and the limit you set is the Acceptance Quality Limit, or AQL. Get the plan right and a few hundred units tell you whether tens of thousands are good. Get it wrong — or skip it — and the first time you see the goods is at an Egyptian port, with the money already gone. This guide covers the sampling maths, the defect classes, the lab tests, and the on-site checklist.

Why sample instead of checking everything

Inspecting 100% of an order is rarely practical: it’s slow, expensive, and tired inspectors miss defects. A 100% check usually yields little more information than a properly drawn statistical sample (QualityInspection.org, “AQL Inspection Levels”). So instead of checking every unit, a PSI pulls a random, statistically representative sample and uses it to estimate the quality of the whole lot.

The framework that governs this is ISO 2859-1: Sampling procedures for inspection by attributes — Part 1: Sampling schemes indexed by acceptance quality limit (AQL) for lot-by-lot inspection (ISO 2859-1). Its American sibling, ANSI/ASQ Z1.4, derives from the old MIL-STD-105E military standard; the two are close but not identical — the 1999 ISO revision changed some accept/reject pairs, so name the standard you’re using in your inspection brief (ECQA, “ISO 2859-1 vs ANSI Z1.4”). In more than 95% of attribute inspections worldwide, the inspector is following ISO 2859-1 (QualityInspection.org).

Two concepts do the work, and they are separate:

• Inspection level decides how many units to sample.
• AQL decides how many defects are tolerable in that sample before the lot is rejected.

Confusing the two is the most common AQL mistake. Changing the level changes sample size; it does not change your AQL (QualityInspection.org).

How AQL sampling actually works

The inspector uses two master tables from ISO 2859-1:

1. Table 1 (code-letter table) maps your lot size and inspection level to a single code letter (A–R).
2. Table 2 (single-sampling table) maps that code letter to a sample size, and — for your chosen AQL — to an accept number (Ac) and a reject number (Re).

The rule is then mechanical: if defects found in the sample are ≤ Ac, the lot passes; if they reach Re, it fails (Eurofins, “Explaining Acceptance Quality Limit (AQL)”).

Inspection levels

ISO 2859-1 offers three general levels (I, II, III) and four special levels (S-1 to S-4). General Level II is the default for routine visual inspection (QualityInspection.org):

• Level I — smaller sample, less discrimination. Used for trusted suppliers or to save time; risky as a default.
• Level II — the standard for general consumer goods, used by default.
• Level III — larger sample, more discrimination. Used after recent quality problems or for high-value goods.
• Special levels S-1 to S-4 — very small samples, higher sampling risk. Used for destructive or time-consuming tests, or quick container-loading checks.

The effect on sample size is large. For a 5,000-unit lot, the sample drawn is roughly 80 units at Level I, 200 at Level II, 315 at Level III, against 5 / 8 / 20 / 32 units at S-1 to S-4 respectively. For a 40,000-unit lot it scales to about 200 / 500 / 800 at the general levels (QualityInspection.org). More samples means more chances to catch bad lots — and more inspector days.

A worked example

Order 5,000 units, General Level II. Table 1 gives code letter L, and Table 2 sets a sample size of 200 units. At AQL 2.5, the accept number is 10 and the reject number is 11: if 10 or fewer sampled units carry that defect class, the lot passes; at 11, it fails (QualityInspection.org, “What is the AQL”). Run the same 200-unit sample at AQL 4.0 and the accept/reject numbers loosen; at AQL 1.0 they tighten. Same sample, different tolerance — which is exactly why you set AQL by defect severity, not by guesswork.

Approximate sample sizes per ISO 2859-1 master tables, illustrating the level effect. Exact code letters and accept/reject numbers come from the standard’s tables for your specific lot size and AQL.

Critical, major, minor: setting AQL by defect class

You don’t run one AQL across the whole inspection. Defects are graded, and each grade gets its own tolerance (Eurofins; QIMA, “Acceptable Quality Limit”):

• Critical defects — safety hazards or breaches of mandatory regulation (sharp edges, electrical/fire risk, toxic materials). Almost always AQL 0 — zero tolerance; any critical defect found rejects the lot.
• Major defects — faults that affect function or saleability (a motor that won’t start, a zip that won’t close, wrong labelling). Industry standard AQL 2.5.
• Minor defects — cosmetic issues that don’t affect function (slight colour variation, minor scratches, loose threads). Industry standard AQL 4.0.

A typical consumer-goods brief therefore reads 0 / 2.5 / 4.0 for critical/major/minor. Tighten the numbers for safety-sensitive or premium goods; the supplier should agree the grading and the defect catalogue before production, so there’s no argument over what counts as “major” on inspection day.

A lower AQL number = stricter. AQL is a process-average tolerance, not a guarantee that the accepted lot is defect-free; it’s the maximum defective rate you’ll treat as acceptable over the long run (QIMA).

When PSI happens — and the inspections around it

Pre-shipment inspection is one point on a QC timeline, and it’s usually done when the order is 100% produced and at least ~80% packed, so the inspector sees the goods as they’ll actually ship. But final inspection alone is late — by then a bad lot is a bad lot. A fuller QC programme stacks checks:

For an importer placing a first order with a new supplier, DUPRO + PSI together are far stronger than a single end-of-line check.

Lab tests: what samples can’t tell you by eye

Visual AQL sampling judges appearance, workmanship and basic function. It does not confirm chemistry, safety or performance — those need a laboratory. Pull samples during the PSI and send them to an accredited lab. For credibility (and for Egyptian conformity), the lab should hold ISO/IEC 17025 accreditation; results from an unaccredited lab may not be accepted (US Dept of Commerce / trade.gov, “Egypt – Standards for Trade”).

Common test families:

• Composition / material verification — confirm the grade, polymer or alloy is what the PO specifies, not a cheaper substitute.
• Safety / restricted substances — heavy metals, migration limits, restricted chemicals, depending on product and destination rules.
• Performance — strength, durability, electrical safety, function under load.
• For food and ingredients — microbiological, contaminant and compositional testing against the relevant Egyptian (EOS) standard or, where none exists, an international standard such as Codex. The document trail starts with the Certificate of Analysis (COA); how to read and verify one is covered in how to evaluate a flavour COA.

This testing also underpins Egyptian customs clearance. Many regulated categories require conformity inspection at origin — physical inspection plus accredited-lab sampling — before the goods ship, with a Certificate of Conformity issued per consignment. That’s where PSI and the regulatory side meet; the registration and conformity mechanics are in GOEIC inspection and registration.

What to check at origin: the practical checklist

A competent PSI works through, at minimum:

1. Quantity — units produced and packed against the PO.
2. Conformity to the approved sample — colour, dimensions, material, construction against the signed reference sample. No approved sample = arguments later.
3. Workmanship — graded against the agreed defect catalogue (critical/major/minor).
4. Function and on-site tests — switch-on, fit, assembly, basic performance; product-specific safety checks.
5. Measurements — key dimensions and weights against spec, with the inspector’s calibrated tools.
6. Labelling and markings — product, carton and shipping marks; barcodes; country of origin; any mandatory destination markings.
7. Packaging — retail and export packing, drop/carton integrity where applicable.
8. Sample retention — units pulled and sealed for the lab and as a reference.

The result is reported as PASS / FAIL / HOLD (pending) against the AQL plan — not a vague “looks fine.” A HOLD typically means a fixable issue (re-work, re-pack, re-inspect).

Traps that quietly break an AQL inspection

The maths is sound; the misuse is common. Watch for these:

• The “check 10%” myth. Inspecting a flat percentage of the order — 10%, say — has no statistical basis. A 10% sample of 1,000 units and of 100,000 units give wildly different confidence, yet the rule pretends they’re equivalent. ISO 2859-1 deliberately ties sample size to lot size non-linearly, which is the whole point of the code-letter table (QualityInspection.org, “AQL Inspection Levels”). Use the standard, not a percentage.
• Counting “sets” as one unit. If your product is a set (a 6-piece tool kit, a 12-pen pack), the inspector must define whether the sampling unit is the set or the individual piece. Treating a defective single piece as if it were a whole defective set — or vice versa — quietly changes your effective AQL. Settle the unit of sampling in the brief.
• No agreed defect catalogue. Without a signed list of what counts as critical, major and minor for your product, the accept/reject call becomes a negotiation on inspection day. The supplier argues a fault is “minor”; you say “major.” Agree it before production.
• Skipping the golden sample. Colour, finish and construction checks need a signed reference at the factory. “It looks like the photo” is not an inspection criterion.
• Treating a pass as a guarantee. AQL accepts a known, non-zero defect rate. If your customer or your category tolerates zero of a particular fault, that fault belongs in the critical class at AQL 0, or it needs a 100% sort — not a relaxed sampling plan.

Booking and running an inspection: the mechanics

A pre-shipment inspection is performed at the supplier’s factory or warehouse, usually completed in one full day on-site, with the report typically delivered within 24 hours of the visit (QIMA, “Guide to the Pre-Shipment Inspection”). Established third-party firms can place an inspector on-site quickly, but it’s sensible to book several days in advance to secure scheduling and avoid rush fees; urgent next-day bookings usually cost more.

The timing rule is the one that catches importers out: schedule the PSI when at least ~80% of the order is produced and export-packed. Earlier than that and the lot isn’t statistically representative; later and there’s no window left to fix anything before the shipping deadline (QIMA). Build the inspection date into the production schedule from the start, with a few days’ buffer before the cut-off for the booked vessel.

On the day, the inspector selects the sample using the ISO 2859-1 procedure, with the inspection level and AQL fixed in advance in the inspection brief; verifies quantity and carton/label integrity; runs visual workmanship checks against the defect catalogue; performs on-site function and, where applicable, carton drop tests; and pulls samples for the lab (QIMA). The output is a graded report — pass, conditional/hold, or fail — with photos and the defect tally against your plan.

Choosing the inspection company

For routine commercial quality, an independent third-party inspection firm is fine, provided the inspection brief (standard, level, AQL, defect catalogue, golden sample) is unambiguous. For Egyptian customs conformity, the choice narrows: the inspection body must be one GOEIC recognises, and any laboratory testing must run through an ISO/IEC 17025 accredited lab so the results are accepted at clearance. Where a single visit can satisfy both your commercial QC and the conformity inspection, plan it that way — one trip, one set of pulled samples, two outputs. Confirm an inspection company’s standing for your product category before you book; an inspection from a body the regulator doesn’t recognise is wasted money.

How Innovote sources this

We build inspection into the order rather than bolting it on at the end. For each line we:

• Agree the plan before production. We fix the inspection standard (ISO 2859-1), inspection level (Level II by default, III for new suppliers or sensitive goods), and the AQL by defect class (typically 0 / 2.5 / 4.0), and we sign off a defect catalogue and a golden sample so accept/reject isn’t subjective.
• Stack the checks. For first orders or higher-risk goods we add a during-production check, not just a final PSI, so problems surface while they’re still correctable.
• Test what the eye can’t see. We pull samples for an ISO/IEC 17025 accredited lab to verify composition, safety and performance against the cited Egyptian (EOS) or international standard — and we line this up with the conformity certificate the consignment needs.
• Tie inspection to payment. Where terms allow, we structure the balance to release against a passed inspection, so the supplier carries the incentive to get it right the first time.
• Report straight. PASS / FAIL / HOLD against the plan, with photos, measurements and the defect tally. We don’t dress up a marginal lot, and we don’t call goods “certified” beyond what a test report or certificate actually says.

Tell us the product, order quantity and your risk tolerance; we’ll propose an inspection level, AQL by defect class, the lab tests worth running, and how to sequence it all with shipping and Egyptian conformity.

Frequently asked questions

What AQL should I use?
For general consumer goods the common default is 0 / 2.5 / 4.0 — zero tolerance for critical defects, AQL 2.5 for major, AQL 4.0 for minor. Tighten these for safety-sensitive or premium products. The right number is a commercial decision about how much risk you’ll carry; set it with your supplier before production and write it into the PO.

Does a passed AQL inspection mean zero defects?
No. AQL is the maximum defective rate you’ll accept as a long-run process average, not a guarantee of perfection. A lot can pass at AQL 2.5 and still contain some minor defects within the accept number. If you need defect-free, that’s a 100% sort, not sampling.

ISO 2859-1 or ANSI/ASQ Z1.4 — which standard?
They share a common ancestry (MIL-STD-105E) and Level II as the default, but the 1999 ISO revision changed some accept/reject pairs, so they aren’t identical. ISO 2859-1 is the more internationally recognised; ANSI/ASQ Z1.4 is common for US trade. Pick one and name it in the inspection brief so the inspector applies the right table.

How big is the sample for my order?
It depends on lot size and inspection level. At General Level II, a 5,000-unit lot samples around 200 units and a 40,000-unit lot around 500 — the exact figure comes from the ISO 2859-1 code-letter and single-sampling tables. Bigger lots and higher levels mean larger samples and longer inspections.

Is pre-shipment inspection the same as the inspection GOEIC requires for Egypt?
Related but not identical. A commercial PSI checks quality against your AQL plan. Egyptian customs clearance for regulated goods needs conformity inspection — physical inspection plus accredited-lab testing at origin against the Egyptian/international standard, with a Certificate of Conformity per consignment. Run them together: one inspection visit can serve both quality and conformity if planned that way. See GOEIC inspection and registration.

Can the inspector check 100% of the order instead of sampling?
Yes, but it’s slow and expensive and rarely justified except for high-value goods or a supplier with a poor recent record. For most orders a properly drawn ISO 2859-1 sample gives you a statistically sound read at a fraction of the time and cost.

Related guides

• The Complete Guide to Importing into Egypt: NAFEZA, ACID, GOEIC, NFSA, Incoterms & QC — where inspection fits in the full import chain.
• GOEIC inspection and registration: getting goods cleared without surprises — the conformity certificate your consignment needs at the Egyptian port.
• How to evaluate a flavour COA and run an incoming-quality check — reading the lab document trail behind your samples.

Placing a first order and want to know it’s right before you pay? Tell us the product, order quantity and how much risk you’re willing to carry — we’ll propose an inspection level, AQL by defect class, the lab tests worth running, and come back with grade, MOQ, lead time and a landed-cost path.

Byline: Innovote Trade Desk

Compliance note: this article describes ISO 2859-1 / ANSI/ASQ Z1.4 sampling practice and Egyptian conformity requirements as of June 2026. Accept/reject numbers come from the current standard’s tables; regulatory requirements change. Verify the applicable standard and Egyptian rules with the standard body, an accredited lab, or a licensed customs broker before relying on them.