A pre-shipment inspection is a sampling audit at the end of production, not a guided tour of the factory. The goods are finished and mostly packed, an inspector draws a sample from the lot, checks quantity, workmanship and function against your purchase order, and writes a report that either releases the balance payment or stops the container. It usually takes one working day, and nearly everything that decides the outcome was agreed weeks before the inspector walks in.
A PSI answers a narrow question: does this lot match this order. It is not a factory audit — that looks at the system that produced the lot, not the lot itself. It is not a laboratory test either: nothing is sectioned, salt-sprayed or pulled apart, because the samples have to ship.
That matters when you read the report. "Passed" means the sample met the limits you agreed on the day of inspection. It does not mean the factory can repeat the result next quarter, and it does not cover a defect your specification never mentioned.
Most third-party inspections are booked days in advance, and most failed inspections are decided by the calendar rather than the workmanship. If an inspector arrives while the last cartons are still coming off the line, the sample is no longer random: it is whatever happened to be finished, usually the easiest part of the run.
The clause worth writing into every order is simple — full quantity produced, most of it packed, before the inspection date. Packing counts because carton condition, quantity per carton and shipping marks are part of what is being checked.
Who books the inspection matters too. A buyer-booked inspection is paid by the buyer and reported to the buyer. A supplier-booked one is cheaper and weaker evidence, even when the inspector is scrupulous. Name the firm, the sampling standard and the acceptable quality level in the purchase order, not in an email after the goods are ready.
An inspector who starts by opening cartons is guessing. The sequence is documents first: purchase order, the specification you signed, the approved sample, the packing list and any previous report on the same item. Each one closes a different door. The approved sample settles arguments about finish and shape; the packing list turns a count into a comparison; a previous report tells the inspector which defect was already raised and whether it came back.
This is also where a vague order is exposed. If the specification says "good quality" and nothing more, the inspector has no limit to measure against and falls back on his own judgement — which is not the same as yours.
No inspection opens every carton, and any inspector who claims otherwise is selling theatre. A lot is accepted or rejected on a sample drawn to a statistical plan, commonly an AQL scheme such as ISO 2859-1: a sample size derived from the lot size, and accept and reject numbers derived from the level you agreed.
The practical consequence is that a passed lot can still contain defective pieces. What the plan controls is the risk of accepting a bad lot, not the number of bad pieces inside a good one. Two things follow. Choose the level by what a defect costs you downstream — a weeping valve body is not the same risk as a scuffed carton — and write down what counts as critical, major and minor before anyone starts counting.
Workmanship is judged under light, not in the hand. A raking light at a shallow angle exposes polishing swirls, thin plating and sand pits that disappear under straight-on light, and a full rotation of the part finds what the front face hides.
Function is where a cheap lot separates from a good one. Every sampled tap, valve and drain is operated: the handle turned through its full travel, the spindle felt for side play, threads run in by hand, and any part that holds water filled and watched on the outside wall. A weep under standing water never shows up in a visual check.
| Stage | What is checked | Evidence worth keeping |
|---|---|---|
| Documents | Order, specification, approved sample, packing list | Photo of the signed sample from four sides |
| Quantity | Cartons counted against the packing list | Photo of the sealed pallet with the count sheet |
| Workmanship | Surface, plating, edges, threads under raking light | Close-up of each defect found, with a rule in frame |
| Function | Handle travel, spindle play, thread fit, water hold | Short video of the sample being operated |
| Packing | Carton grade, inner protection, marks, label data | Photo of one opened carton and one closed |
A carton is part of the product. Inspectors check the board grade against the order, how parts are separated inside, and whether a drop or compression test leaves corners crushed. Then the marks: shipping data against the packing list, label position, a barcode scan where one exists, and whether the carton count per pallet matches what the forwarder was told.
Most damage claims that surface after a sea voyage are not manufacturing defects at all. They are a carton that was adequate for a truck and not for a container, or an inner divider that let two brass bodies rub for six weeks.
A useful report classifies defects rather than describing them: critical, major, minor, each counted against the accept and reject numbers. Photographs should be dated and scaled, not decorative. A result marked pending is not a pass; it means a decision was left to you, usually about a defect sitting on the border of the agreed limit.
Read the sample size before the verdict. "Passed" on a small sample and "passed" on a large one are different statements about the same lot.
It cannot find a design fault, because the sample was made to the drawing you approved. It cannot predict corrosion two years into a coastal installation, because nothing was tested to failure. It cannot prove the next lot will match this one, and it cannot see a production change that happens after the inspector leaves.
Those gaps belong to the other two checks in a normal programme: a during-production inspection that catches a problem while there is still time to correct it, and a factory audit that looks at the system rather than the shipment.
We cast, machine, polish and plate brass drainage and water-supply fittings, so an inspection on our floor tests work we control from casting onwards — including the models in our own catalogue. That range includes the BQ20157-Q brass floor drain for tiled wet areas, the JF-07 brass angle valve used at fixture connections, and the SZ-13-30-6T-F European style bibcock for wall-mounted outdoor supply.
What we ask a buyer to fix before booking is the part that makes an inspection meaningful: the approved sample, the finish and the faces that must stay cosmetic, the accept and reject rule, and the packing method. Given those four, a lot can be checked in a day and a report can be argued with on facts.
Send the item you are buying, the finish you need and the packing method you expect — a photo of the carton you received last time is enough to start. We will reply with the matching part, the sample approval sheet and the inspection points a third party should check on that item, so your inspector and our line are measuring the same thing.