Pre Shipment Inspection China: AQL for Acrylic

What a pre-shipment inspection actually is

A pre-shipment inspection (PSI) is a structured check of finished, packed goods against your approved specification — drawn as a random sample under a published sampling standard — carried out before the balance payment releases the shipment. It is the pre shipment inspection China sourcing playbooks treat as the last gate: the point in a custom acrylic order where a defect is still the factory’s problem to fix, not yours to absorb.

The timing is the whole point. Custom acrylic orders from China typically run on a 30% deposit with the balance due before shipment, which means there is a built-in moment — production finished, cartons packed, money not yet sent — where you hold real leverage and real information can reach you. A PSI, whether performed by a third-party agency, your own sourcing agent, or the factory’s QC team on camera, is how that moment gets used. The convention across the inspection industry is to schedule it when production is 100% complete and at least 80% packed, so the inspector samples from sealed export cartons rather than from a curated table of best pieces.

Quality audits and buyer qualification calls keep showing me the same asymmetry: buyers negotiate price for weeks and then release a five-figure balance payment on a WhatsApp photo of one nice-looking riser. This guide is the corrective — the sampling math, the acrylic-specific checkpoints, and the remote-evidence checklist that turn “looks fine” into a decision you can defend. It covers checking the goods; vetting the factory itself is a different exercise, covered in our supplier audit guide and the 22 vetting questions.

---

AQL 2.5 does not mean 2.5% of your order can be defective

AQL 2.5 is the most misread number in importing. It does not grant the factory a 2.5% defect allowance, and it does not promise you a shipment that is 97.5% perfect. AQL — acceptable quality limit — is the worst average process quality that the sampling scheme is designed to keep accepting over many lots; each individual lot simply passes or fails on the defect count found in its sample.

The operating standard is ANSI/ASQ Z1.4 (harmonized with ISO 2859-1), the sampling system we and every serious inspection agency work from.¹ The order sets an inspection level (General Level II is the default), the lot size maps to a sample size code letter, and the code letter plus the chosen AQL gives two numbers: accept and reject. Find no more than the accept number of defects in the random sample and the lot passes; hit the reject number and it fails. The near-universal consumer-goods convention stacks three AQLs at once: 0 for critical defects (anything unsafe — for acrylic, think a razor-sharp unpolished edge on a retail counter piece), 2.5 for majors (a defect a customer would reject: a scratch across the display face, a crooked engraving, a cracked corner), and 4.0 for minors (visible flaws that would not stop use — a faint mark on a non-show face).

The classification argument matters more than the arithmetic. A scratch is not “a scratch”: a 20mm hairline on the show face of a retail riser is a major; the same mark on the underside of a base is a minor. Before any inspection — third-party or ours — the defect classification list should be agreed in writing per product face. When I brief my inspection team on a new order, the first document on the bench is not the sampling table; it is the approved sample and the face map that says which surfaces are show faces.

---

The sampling math: what 50, 200, and 500 pieces actually draw

Under Z1.4 General Inspection Level II, the lot size sets everything. The table below shows the single-sampling, normal-inspection plans for common acrylic order sizes at the standard AQL stack — this is the math your inspector runs before quoting you a man-day.¹

Lot size	Code letter	Sample size	AQL 2.5 (majors) accept / reject	AQL 4.0 (minors) accept / reject
50 pcs (our MOQ)	D	8	table redirects — see below	table redirects — see below
151-280 pcs (e.g. 200)	G	32	2 / 3	3 / 4
281-500 pcs (e.g. 500)	H	50	3 / 4	5 / 6
501-1,200 pcs (e.g. 1,000)	J	80	5 / 6	7 / 8

Read the 500-piece row the way an inspector does: 50 pieces pulled at random across cartons, and the whole lot’s fate rides on whether the majors found number 3 or 4.² That is a sharp, useful gate for volume production — and statistics, not perfectionism, is why it works: a factory whose process really runs at 2.5% defective will pass this plan most of the time, while a process running at 10% defective will fail it fast.

The table’s honest limit is the top row. A 50-piece lot draws a code-D sample of just 8 pieces, and at AQL 2.5 the master table does not even print an accept number there — it prints an arrow redirecting you to a different plan, the standard’s own way of saying the sample is too small to discriminate at that AQL. For a 50-piece custom run, sampling is close to theater. That is why our standard for MOQ-sized orders is not a sampling plan at all: every piece crosses the inspection bench individually — the same 100% piece-by-piece check our quality process applies at every quantity, with sampling statistics layered on top for volume lots rather than substituted for the full check.

---

Acrylic-specific checkpoints: what generic inspectors miss

A generic consumer-goods checklist catches broken parts and wrong quantities. Acrylic fails in quieter ways, and a PSI on acrylic goods needs six checkpoints written into the brief — whether the inspector is a third-party agent or my own team on camera.

Scratch inspection under defined conditions. “No scratches” is not an inspectable spec. Ours reads: show faces inspected under raking light at roughly 60cm viewing distance, with the protective film lifted at the inspection station — because film hides hairlines, and a scratch discovered at unboxing in Ohio is a warranty claim, not a QC note. Face maps mark which surfaces are show faces (major if marked) and which are concealed (minor).

Engraving and print registration. Engraved names, logos, and UV-printed graphics get checked against the approved artwork file for position, depth, and edge crispness — not against the inspector’s taste. Bottom-printed and multi-color work adds registration between layers; our tray print-quality guide shows what registration drift looks like before you learn to spot it in a report photo.

Bond lines, backlit. Solvent-bonded joints are inspected against a light source: bubbles, dry patches, and squeeze-out all silhouette cleanly when backlit. A bond line that looks acceptable face-on and shows a bubble chain backlit is a joint that will whiten under load.

Dimensions with the caliper in frame. Sampled pieces measured against drawing dimensions and tolerances, photographed with the caliper reading visible. For slotted and fitted products — card cases, dividers, inserts — a fit check against the actual mating piece beats any measurement.

Protective film and packing integrity. Film present on specified faces, cartons to the packing spec, and a packed-carton drop check consistent with transit-simulation practice — ISTA’s published test procedures are the reference frame serious packaging is validated against.³ Acrylic that survives the truck is a packaging outcome, not a material one.

Quantity and assortment. Boring, and the single most common discrepancy in inspection reports across industries: carton count, per-SKU count, spares included as agreed.

---

The remote PSI: a photo and video checklist that replaces your eyes

Most buyers at 50-500 piece order sizes will not fly to Shenzhen and will not always book an agency, and the remote inspection is only as good as the checklist behind it. We built ours across hundreds of custom orders, and I still tighten it whenever a new failure mode reaches my bench. Evidence you can demand from any factory, for any acrylic order, before the balance invoice:

#	Evidence item	What it proves
1	Every SKU face-on, raking light, film lifted	Scratch state of real show faces
2	Edge and corner close-ups, per SKU	Polish quality, chips, sharp edges
3	Bond lines shot against a backlight	Bubbles, dry joints, squeeze-out
4	Engraving/print beside approved artwork on screen	Registration and artwork match
5	Caliper on critical dimensions, reading in frame	Dimensional conformity
6	Random unboxing video: sealed carton opened on camera	Sample wasn’t curated
7	Packed carton: inner packing, then sealed and labeled	Transit readiness, correct labeling
8	Count sheet: cartons x per-carton quantity, per SKU	Quantity and assortment

Item 6 is the one that changes behavior. A factory choosing which pieces to photograph is running marketing; a factory opening the carton you point at, on video, is running QC. We send this evidence set as standard on custom orders — it is the remote version of the same 4-stage inspection my team runs on the bench, and it costs a buyer nothing but the request. When a print-lab client scaled their photo block program with us, the per-batch evidence set was what let their procurement team approve reorders without a site visit.

---

Third-party vs factory self-inspection: use both, in the right order

The framing “third-party inspection versus factory QC” is wrong — they answer different questions. Factory QC asks “does every piece meet the spec we agreed?” and runs during and after production. A third-party PSI asks “can the buyer trust the factory’s answer?” and runs once, at the end, on a random sample. A healthy order can carry both; a factory that resists the second is answering the first question for you.

When to book an agency, from the QC side of the table: first orders with a new supplier, orders where a retailer or enterprise customer downstream requires an independent report, high-value or compliance-sensitive programs, and any order where the relationship is too new for a video to carry a five-figure balance decision. Agencies quote by the man-day; for a typical one-day, single-location pre shipment inspection in China on an acrylic order, expect a few hundred US dollars — small insurance against a container of rework. For repeat orders of a stable product from a supplier whose reports have matched reality, most buyers rationally taper to the factory’s evidence set and spot-audit occasionally.

Our position is simple and we state it in writing on request: any inspection agency you choose is welcome, we schedule the visit at the 80%-packed point without shipping-date games, and our own 100% piece-by-piece inspection runs regardless — the third party checks our work, not the other way around. If a supplier you’re vetting hedges on exactly this point, weigh it the way our manufacturer vetting checklist suggests: cooperation with independent inspection is one of the lowest-effort honesty signals in the industry.

---

Building the PSI into your order from day one

The inspection that runs at day 20 is decided at day 0 — in what the order documents pin down. We ask buyers to put four things in writing with any custom acrylic order, ours included: the approved sample or sealed spec as the inspection reference, a defect classification list per product face, the sampling plan (Z1.4 level and AQLs — or 100% inspection for MOQ-sized lots), and the sequence “PSI report before balance payment” stated as a term. None of this is adversarial; it is the boring paperwork that makes disputes impossible instead of merely unlikely.

On our side it slots into the standard flow without moving the schedule: start your order with drawings or references, approve a physical sample (3-5 days, fee credited to your first production order), production runs 15-20 days with the 30% deposit paid up front, then the evidence set or your third-party inspection — and only then the balance and the FOB Shenzhen handover. Every order ships through the same 4-stage quality process, whatever the sampling plan says, because the sampling table is a gate for lots — the bench check is a gate for pieces.

Footnotes

1. ANSI/ASQ Z1.4 & Z1.9 sampling standards — American Society for Quality — the publisher’s page for the Z1.4 attribute sampling standard cited throughout: inspection levels, sample size code letters, and the accept/reject framework used in pre-shipment inspection. ↩ ↩²

2. Anatomy of the ANSI ASQ Z1.4 industry standard AQL table — InTouch Quality — walks the code-letter lookup used here (a 281-500 piece lot maps to code H and a 50-piece sample at Level II) and the arrow mechanics that redirect small-lot plans. ↩

3. ISTA test procedures — International Safe Transit Association — the published transit-simulation procedures referenced for packed-carton drop and transit checks in a pre-shipment inspection. ↩