Run the audit against one physical object — the sealed golden sample on the QC shelf — not against a photo, a Canton Fair booth unit, or a PDF. A sealed golden sample with approved material, colour, decoration, dimensions and photos is a practical production-QC reference, but it does not replace required market testing. The sample is the standard, the documents are the map, and your job is to prove they still describe the same lot.
This audit is a document-and-sample review you can run yourself. It is not laboratory testing, it is not a defect-rate guarantee, and it does not produce a pass certificate. What it produces is a defensible release-or-hold decision on building blocks wholesale orders — including sets with hundreds of small parts, where a colour shift or a warped plate shows up only after assembly.
What buyers should decide before opening the carton
- Hold the lot if any required test, marking, document or critical inspection is incomplete or has an unresolved failure — a shipment should be blocked on that basis, not on a general impression of quality.
- Get the sampling number from the agreed inspection plan on the PO, not from a catalogue sentence or a blog. Inspectors work from ISO 2859-1 or its US twin ANSI/ASQ Z1.4, which use identical sampling plans, and the default general inspection level for most consumer products is Level II.
- Tag every audit line as supplier statement, measured observation, or recommendation. An audit that mixes the three is an opinion, not a record.
- After any change that can affect safety or compliance — material, colourant, mould, decoration, supplier — document the retest decision even if the product name and SKU stay the same.
- Write the commercial terms that cause disputes: tooling fee, ownership, storage period and idle-tool release conditions, plus the Incoterms version and the named place as actually booked.
Step 2: Build the SKU and document pack
For every SKU in the order, assemble four things: the product specification, the packing specification, the instruction sheet, and the market/lot/file map. Missing any one of them means you cannot audit the lot — you can only look at it.
This is where building-block ranges get messy. A car-and-truck set, a marble-run dinosaur, a flower stacking set and a stick-building kit often ship on one PO with different cartons, different instruction sheets and different destination markets. The pack has to be built per SKU, not per container.
Ask the factory: for each SKU on this PO, which document proves the product spec, which proves the packing spec, and which instruction sheet version ships with it?
Step 3: Tag each line as statement, observation or recommendation
Every line in the audit gets one of three tags. A supplier statement is what the factory says. A measured observation is what you saw or counted. A recommendation is what you want done about it. Keep them in separate columns, because the moment they blur, the factory can answer a recommendation as if it were a statement and the audit stops being evidence.
This one habit prevents the most expensive confusion in toy sourcing: treating a supplier's compliance sentence as if it were a test result.
Step 4: Run the physical comparison against the sealed piece
Open the pack and compare the bulk against the sealed sample on colour, fit, flash and function. Fit matters more than cosmetics in construction sets: uneven cooling or unbalanced shrinkage can warp a moulded part and create problems that appear only at assembly, when a child or a line worker forces two plates that should click.
Mould texturing is a related trap. Texturing can give a repeatable surface appearance without a separate paint operation, but it also changes draft and makes defects more visible — so a texture change is a sample-change event, not a cosmetic detail.
Ask the factory: can you confirm the bulk matches the sealed piece on colour, fit, flash and function?
Step 5: Apply the PO inspection plan, and record nonconformances properly
Apply the defect classes and sampling plan written on the PO. If the plan names no classes, that is the finding: the buyer has no release rule. A nonconformance report should state the defect, the lot, the quantity, the containment, the disposition and who decided — not just a pass/fail note.
On the floor, a failed lot is isolated, the defect class is recorded, and release happens only after rework. Buyers who want a specific acceptance number get it from the agreed inspection plan on the PO. Nobody should be quoting you a defect rate from a catalogue.
Ask the factory: which inspection plan on the PO governs this lot, and what defect classes does it name?
Step 6: Match every paper to the market, the lot and the test
Certificates are not interchangeable. UKCA and a VOC file on an electric toy are different papers, and a buyer who tries to use one for the other has no paperwork at all. Map each market to its file, each file to its lot, and each file to the test it actually covers.
For US children's product lots, check the tracking label content: manufacturer, date and lot on the product or pack as required, with the CPC matched to the SKU rather than to the factory door. Tracking labels are not decoration — an unpopulated barcode is a documentation defect even when the bricks are fine.
Ask the factory: which certificate or file covers which market, lot and test for this SKU, and what manufacturer, date and lot appear on the tracking label?
Step 7: Check the commercial documents before you check the bricks
Two documents cause more post-shipment arguments than any mould. The first is the tooling policy: fee, ownership, storage period and what happens if the tool sits idle. A buyer who paid a tooling fee and never froze storage years or release conditions does not own a solution, they own a dispute.
The second is the Incoterms clause. FOB without a named place has started freight arguments that outlast production. Write the Incoterms version and the named place — Shantou or Shenzhen as actually booked.
Ask the factory: what are the tooling fee, ownership, storage period and idle-tool release conditions in writing, and which Incoterms version and named place are booked?
Step 8: Flag reorder candidates as a recommendation, not a fact
The last pass is commercial, not technical. Note which categories fit small storage space and move fast enough to justify a reorder signal. Small, cheap, high-turnover lines have a habit of outselling pricier categories and selling out ahead of schedule — the category a buyer treats as a filler can become the line customers reorder in advance.
Keep this in the recommendation column. Turnover is a market observation, not a measurement of the lot in front of you.
Worked example (illustrative, not a real shipment)
This scenario is invented to show where an audit bites. It is not a real order, factory, or inspection result.
A buyer books one lot of roughly 4,000 construction sets across four SKUs: a large multi-model car and truck set, a dinosaur marble-run set, a flower stacking set, and a stick-building kit. Two destinations: one EU retailer pack, one US retail pack. The EU pack needs its own marking and file set; the US pack needs tracking-label content and a CPC matched to each SKU.
Steps 1 to 3 pass — the sealed sample is present, revision-labelled, and the document pack is complete. Step 4 passes on colour and function. Step 5 surfaces the hold: the dinosaur marble-run carton for the US destination carries a tracking label with manufacturer and date but no lot code, and the CPC on file is addressed to the factory rather than to that SKU. Under the buyer's own release rule, a required marking that is incomplete blocks the shipment. The lot is isolated, the carton is re-labelled, and the CPC is reissued to the SKU before release. The EU pack never gets opened past step 4, because the hold is destination-specific.
One missing field, one held lot. That is the audit working as designed.
Audit steps, what to check, and the red flag that should hold the lot
| Step 1 — Freeze the reference piece | Check: sealed golden sample present, sealed, tool number and revision on label | Red flag: only a photo, a Fair booth unit, or an unlabelled sample exists |
|---|---|
| Step 2 — Build the SKU and document pack | Check: product spec, packing spec, instruction sheet and market/lot/file map per SKU | Red flag: documents exist for the container but not for one or more SKUs |
| Step 3 — Tag statement vs observation vs recommendation | Check: each audit line carries one tag, in separate columns | Red flag: compliance sentences recorded as if they were test results |
| Step 4 — Physical comparison to the sealed piece | Check: colour, fit, flash and function against the sealed sample | Red flag: bulk approved against a screen image or a revised texture with no new sample |
| Step 5 — Apply the PO inspection plan | Check: defect classes, sampling plan, nonconformance record with lot, quantity, containment, disposition | Red flag: no inspection plan on the PO, or a pass/fail note with no defect class |
| Step 6 — Paper-to-lot match | Check: each certificate mapped to market, lot and test; tracking label content; CPC matched to SKU | Red flag: one certificate reused for two different requirements, or a blank lot field on the label |
| Step 7 — Commercial documents | Check: tooling fee, ownership, storage period, idle-tool release conditions; Incoterms version and named place | Red flag: verbal tooling terms, or FOB with no named place |
| Step 8 — Reorder flags | Check: categories that fit small space and move fast, logged as recommendations | Red flag: turnover guesses written into the measured-observation column |
Does a sealed sample audit replace laboratory testing?
No. A sealed golden sample is a production-QC reference, and the audit around it is a sampling and documentation exercise. It does not replace required market testing, and it does not prove a lot is defect-free. AQL sampling is statistically grounded, but it is a sampling framework — not a 100 percent unit check.
Where a market requires testing, the test file has to exist and has to match the lot. The audit's job is to catch the mismatch before the container moves.
One short note on compliance timing
Treat regulatory dates in three states: entered into force, actually applicable to this product category, and not confirmed by official text. A regulation that has entered into force is not automatically applicable to every toy SKU on the date it was published, and secondary articles that put a year in a headline are not a legislative timetable.
Where the official text does not confirm an application date for a product category, write 'timeline is not confirmed from official text' in your audit notes and keep the file mapped to the market as it stands today. Keep Digital Product Passport requirements, GPSR, and toy safety directive obligations as separate line items — they are different regimes and do not share one deadline.
What to ask suppliers: audit RFQ checklist
Send these as written questions and require written answers. Verbal confirmations do not survive a change of account manager.
- What is the tool number and revision on the sealed sample you are holding, and when was it sealed?
- Which inspection plan on the PO governs this lot, and what defect classes does it name?
- Which certificate or file covers which market, lot and test for this SKU?
- What manufacturer, date and lot appear on the product or pack tracking label?
- Is the CPC matched to this SKU, or to the factory door?
- What are the tooling fee, ownership, storage period and idle-tool release conditions, in writing?
- Which Incoterms version and named place — Shantou or Shenzhen — are actually booked?
- Can you confirm the bulk matches the sealed piece on colour, fit, flash and function?
- After the last change to material, colourant, mould or decoration, was a retest decision documented?
FAQ
What is a sealed sample inventory audit in toy sourcing?
It is a document-and-sample review that compares a production lot against a sealed golden sample and the SKU document pack. It produces a release-or-hold decision and a record of defects, not a laboratory test result or a safety certificate.
How many units should I pull for the audit?
The sample size comes from the agreed inspection plan on the PO, using ISO 2859-1 or its identical US counterpart ANSI/ASQ Z1.4, with General Inspection Level II as the default for most consumer products. Do not accept a sample size quoted from a catalogue or a blog post.
Can a sealed sample replace EN71 or ASTM F963 testing?
No. A sealed sample with approved material, colour, decoration, dimensions and photos is a production-QC reference only, and it does not replace required market testing. Keep the test file separate and confirm it matches the market, the lot and the SKU.
What should trigger a retest even if the SKU has not changed?
Any change that can affect safety or compliance — material, colourant, mould revision, decoration or production site — should have a documented retest decision, even when the product name and SKU stay the same.
Which single finding should hold an entire container?
An incomplete or unresolved required test, marking, document or critical inspection should block the shipment. In practice that is often a missing lot code on a tracking label or a compliance file addressed to the factory rather than to the SKU.
How do I audit a mixed building blocks order with several SKUs and two destination packs?
Build the document pack per SKU, then per destination pack, and run the paper-to-lot match before the physical comparison. A hold can be destination-specific: one retail pack can clear while the other stays isolated on a marking or file defect.
Sources
- ANSI/ASQ Z1.4 (ISO 2859-1) AQL Sampling Inspection Guide
- AQL Inspection Standards: Acceptable Quality Limit Levels for FBA Imports From China
- AQL Table for Inspection Sampling: Standards, Sampling Levels & Acceptance Numbers
- How to Inspect Plush Toys Before Shipment
Related Products & Categories
Run the audit on your next building blocks lot
Send us your SKU list, destination markets and the inspection plan you intend to put on the PO. We will tell you which documents, sample labels and packing specs we can line up against it — and which questions belong in writing before the first carton is sealed.