Inline and final inspection are two different QC passes with two different jobs: inline inspection catches defects while the line still runs, when a fix costs minutes; final inspection samples finished, packed goods against AQL, when a fix costs a reorder. Orders that rely on the final pass alone find paint and seam problems after the cartons are sealed, which is exactly when defects are most expensive to correct.
On a duck line running to a 1,000-piece MOQ, both passes are scheduled, not optional. This guide separates the two passes, what each one samples, and how the results are written down so a claim, if one ever happens, reads from records instead of memory.
Inline inspection: catch it while the line runs
Inline checks sample units off the line at intervals, commonly at startup, then every hour or two, against a golden sample approved before mass production. The checklist watches the defects that drift with time and temperature: paint registration on pad-printed eyes and logos, seam flash from mold wear, color drift against the Pantone control, and squeaker function on units fitted with sounders. A 1,000-piece MOQ order runs the inline pass at startup, then every 60 to 120 minutes of cycle time. The value of the pass is arithmetic. A registration problem caught at hour two costs a stencil adjustment and a bin of rework; the same problem caught at final inspection costs hundreds of finished units. Our product development page describes where the golden sample comes from, and the defect vocabulary is covered in the general quality inspection guide.
Final inspection: sample the packed order
Final inspection samples from finished, packed cartons using an AQL sampling level written into the contract, checking what the buyer will actually receive: counts per carton, barcode and polybag print, workmanship sampling, and carton drop performance for the packing style. The pass ends with a written report, photos and a lot reference, and third-party inspectors can run the same AQL plan on the buyer’s behalf. Buyers booking independent inspection should reference the recognized testing bodies, such as Intertek, whose toy screening covers the material safety side alongside the visual pass.
| Pass | When it runs | What it samples | What it catches | Cost of a miss |
|---|---|---|---|---|
| Inline | Startup plus hourly checks | Units off the line vs golden sample | Paint drift, flash, color drift, squeaker | Minutes of rework if caught here |
| Final | After packing, before shipment | AQL sample from packed cartons | Counts, labeling, workmanship, drop | Reorder or claim if missed |
How the two passes write their records
Inline checks log three lines per interval: units checked, defects by type and the correction applied on the spot. Final inspection logs the AQL table used, the cartons opened, defects by code and the accept-or-reject call with photos. Two pages, one per pass: an inline log for a 1,000 pcs order runs 8 to 12 entries per shift, and the final AQL report opens 3 to 5 cartons per thousand with photos, so the shipment file answers any post-delivery question without a meeting. The records also feed the next order: a defect code that repeats across orders points at a tool or a training gap, which is information a buyer wants before the next PO, not after a claim. Claim photos carry a 30 day acceptance window in the contract, and responses run inside 48 hours.
Video: the production line both passes watch
Frequently asked questions
Can I skip inline inspection on a small order?
You can, but the final pass only samples; a drift problem that inline would catch becomes a finished-goods problem. Small orders benefit most from the startup inline check.
What AQL level should the final pass use?
The level is a contract line, chosen by product risk and budget, and the sampling plan is written into the PO so both sides read the same table.
Who pays for third-party inspection?
Usually the buyer books it, and the factory supports access, carton handling and re-inspection if the first pass finds defects.
What is a golden sample?
The approved reference unit from sampling that both sides measure production against; inline checks compare the line to it, hour by hour.




