Getting samples right before you order

Stock sample (现货样). An existing product taken off a shelf. It proves the product exists and lets you feel the general quality of what that supplier sells. It proves nothing about your version, your materials or your tolerances, and it is the sample most often sent when a buyer asks vaguely for "a sample".

Prototype or hand sample (手板). Made by hand or by 3D printing, usually to confirm shape, size, ergonomics and appearance before tooling exists. It is genuinely useful for exactly that and is evidence of nothing else — the material is often not the production material and the process is not the production process.

Pre-production sample (产前样, PPS). Made on the actual tooling, from the actual purchased materials, by the actual process. This is the first object in the sequence that predicts your production run, and it is the one to approve against. If the tooling is new, insist the sample came off it rather than off a prototype route.

Golden sample (封样). The signed reference. Both sides sign and seal identical units; the factory keeps one and you keep one. It is the physical version of the specification and the thing a dispute is measured against.

Shipment sample (大货样). Pulled from finished goods, usually at inspection. It proves what was actually produced — which is a different claim from all four above, and the only one made after the money question is live.

A sixth meaning, and it is not the same word at all: in quality control "sampling" means the statistical selection of units to inspect from a lot. That is AQL sampling, an entirely separate subject, and confusing the two is common enough to be worth naming once.

Buyers are routinely surprised that one unit costs many times the quoted piece price, and read it as opportunism. It usually is not. A sample is a production setup performed for a quantity of one: machine time, a line stopped and reconfigured, material bought in a minimum quantity you are not consuming, an engineer's attention, and often an entirely different process because the tool does not exist yet.

A sample charge is normally negotiated as credited against the order if the order proceeds. That is common practice rather than a rule, so agree it in writing before you pay, and be specific about what triggers the credit and when.

Where a tool is involved the two costs separate. The tooling charge builds the mould; the sample charge runs it once. Agree who owns the tool at the same time you agree the charge — a tooling payment is not the same thing as tooling ownership, and it is a much easier conversation before the money moves.

Be suspicious of a free custom sample rather than pleased by it. A supplier giving away engineering time on an untested enquiry is either absorbing it into a unit price you have not seen yet, or sending you something that was already sitting in the sample room — which is a stock sample with a new label.

Most samples travel by international express, and the paperwork matters more than the size of the parcel. A sample still needs a commercial invoice with a realistic declared value, an honest description and the right classification. Undervaluing a sample to save duty is a customs declaration, not a favour, and it also sets the value your own market sees for the goods that follow.

Anything containing a lithium cell is a different problem. A power bank, a battery-powered device or a loose cell cannot travel as an ordinary express parcel: dangerous-goods rules govern how it is packed, marked, documented and carried, and couriers refuse or seize non-compliant shipments regularly. Plan for it in the schedule rather than discovering it, and let the factory's freight people handle the declaration — this page deliberately publishes no threshold, because the guides on lithium shipping and watt-hour ratings carry those figures and one set of numbers in one place is how they stay right.

Ask what is in the box before it ships. A sample with a battery installed, a sample with a magnet in it, an aerosol, a liquid or a powder each carry their own transport rules, and the sender knows what they packed.

Have it sent to where it will be examined. A sample forwarded twice arrives late and sometimes arrives damaged in a way nobody can attribute, which wastes the whole exercise.

Photograph it before you touch it — packed, unpacked, from every side, with something in frame for scale. If there is later an argument about condition on arrival, these photographs are the record and they cost nothing.

Measure it against the written specification, not against the photograph you were sent. Dimensions, weight, materials, finish, markings, the components you named. Write the result down beside each line of the specification so the comparison is a document rather than an impression.

Test the thing it is for. Charge it, run it, load it, wash it, drop it if that is what it will meet. Test to the limits you actually care about, and note them: a sample that survives one cycle proves less than most buyers assume.

Buy two and destroy one. Open it up. See what is inside — which cell, which chipset, how the joints are made, whether the internals match the specification you were quoted on. A single unit that is admired and shelved is the least informative outcome available.

Then seal, label and date the one you keep. Note who approved it, what was approved, and what was still open at the time. That last clause prevents the commonest argument: an approval that meant colour to you and meant everything to them.

A sample is made once, carefully, by someone who knows it is being judged. Production is made repeatedly, to a schedule, by whoever is on shift. The gap between them is not dishonesty; it is the difference between a best case and an average, and managing it is the actual job.

Close it with the pre-production sample rather than with assurances. A PPS made from the purchased materials on the real tooling converts a promise into an object, and any difference between it and the hand sample you liked is information you want before the run rather than after it.

Then decide what happens at the limits before production, in writing. What is a cosmetic defect and what is functional; what is acceptable variation in colour; what happens to units that fall outside it. These are cheap sentences before the run and are unresolvable arguments afterwards, because both sides are then describing the same units in good faith and disagreeing about the words.

And treat the first production batch as evidence, not as a formality. A shipment sample checked against the sealed golden sample, before the balance is paid, is the only point in the whole sequence where a difference is both discoverable and fixable.