The shipment fails your check, and the factory’s answer is that it matches the sample you approved. You pull out your photographs of that sample and the factory pulls out theirs, and the two do not quite agree. Most disputes at this point are not about honesty — they are about which sample, at which stage, was actually the one both sides meant to lock. The stages exist to stop that argument before it starts, and most gaps appear because a stage was skipped or its approval was never written down anywhere either side can point to later.
💬 Mid-way through a sample round and unsure which stage you are actually on? Send us the product and the samples so far and we will help you read where you stand.
What are the stages, in order?
Four stages, each answering a different question, and each one only meaningful once the stage before it has actually been approved rather than assumed.
| Stage | Question it answers | What is being tested | Made in what quantity |
|---|---|---|---|
| Proto (prototype) sample | Can this be made at all, roughly as designed? | Feasibility of the design — whether the concept translates into a physical part | A single unit, sometimes by hand, often on different tooling than production will use |
| Fit / size-set sample | Does it fit, in every size or configuration ordered? | Dimensional accuracy across a size run, or fit against a mating part | One unit per size or variant, on tooling closer to production intent |
| Pre-production (golden) sample | Is this exactly what mass production should match? | Every specification point together — material, finish, dimensions, function, packaging | A small number of units, on the actual production tooling and line |
| Shipping sample | Does the packed, finished unit match what was approved, drawn from the real shipment? | The finished product exactly as it will arrive, pulled from the lot itself | Drawn from the completed lot, at or near pre-shipment inspection |
Each stage answers a narrower question than the one before it. Approving a proto sample as if it were a pre-production sample — treating “yes, this concept works” as “yes, mass-produce exactly this” — is where most disputes trace back to, because the proto was never meant to carry that weight.
What is a counter-sample, and why keep one?
A counter-sample is a duplicate of the approved sample kept by each side, so a dispute later is a comparison between two physical objects rather than between memory and a description.
When a sample is approved — most often the pre-production sample — make (or ask the factory to make) a second, identical unit at the same time, from the same production run. One stays with the buyer, one stays with the factory. Both are marked, dated and referenced by the same approval record. If a later shipment is challenged, both sides are comparing the delivered goods against the same physical reference rather than against a photograph, a verbal description, or a memory of what the sample looked like weeks earlier.
This matters more than it sounds like it should, because photographs compress colour, texture and finish in ways a physical sample does not. Where the product has a colour, texture or surface-finish tolerance, a counter-sample is close to essential; where it is a simple mechanical part defined entirely by drawing dimensions, a well-kept drawing and inspection record can substitute.
💬 Sunsho Trade holds counter-samples on file for orders it coordinates and can arrange third-party storage for larger programmes. Ask about sample handling for your product.
How should an approval actually be recorded?
In writing, against a specific physical sample identified by a reference number, signed or confirmed by name on both sides — not in a chat message that says “looks good.”
A usable approval record has:
- A reference number or code unique to that sample, written on the sample itself and in the record.
- The date, and who approved it by name.
- What was checked against — the specification, the drawing revision, or the prior stage’s sample.
- Any deviations noted and accepted at that stage, in writing, rather than assumed to be fixed later.
- Where the approved sample and any counter-sample are physically kept.
A message reading “looks good, go ahead” against an emailed photograph is not this. It works until the first disagreement, and then it is one person’s recollection against another’s, with neither able to point to the object in question.
The common gaps
Most disputes trace back to one of a small number of repeated gaps — not to bad faith on either side.
| Gap | What it looks like | What it causes |
|---|---|---|
| Skipping the pre-production sample | Going from a fit sample straight to mass production to save a week | The first thing anyone sees on production tooling is the finished lot itself |
| Approving on a photograph alone | Sign-off given from an image, with no physical sample kept by the buyer | A colour, texture or finish dispute has no shared reference to settle it |
| No counter-sample | Only the factory holds the approved unit | The buyer has nothing physical to check a disputed shipment against |
| Verbal approval | ”Go ahead” over a call or a chat message, never written into a record | Neither side can point to what was actually agreed, or when |
| Deviations noted but not recorded | A known compromise accepted verbally at the pre-production stage | The compromise resurfaces as a complaint at pre-shipment inspection |
| One sample approved for a whole size range | A single size approved, sizes either side assumed to scale correctly | A size or fit problem appears only once the full run is complete |
Most of these are cheap to fix before production starts and expensive to argue about afterwards. The fix in every case is the same: name the stage, keep a physical object, and write the approval down against a reference number.
None of these gaps needs to be closed by adding process for its own sake. A small first order between two parties who already trust each other can reasonably run on a shorter chain — a pre-production sample and a written approval, no proto stage, no formal counter-sample register. What breaks orders is not the absence of a heavyweight system; it is skipping the pre-production stage entirely, or leaving its approval unwritten, on the assumption that an earlier, looser sample already settled the question.
💬 If you would rather someone in Vietnam managed the sample rounds and kept the records straight, that is our sourcing and negotiation service. Tell us where your sample process is today.
Where sample approval connects to the rest of the order
The pre-production sample is the reference that a pre-shipment inspection actually checks the finished lot against — the two stages only work together if the sample they share is the same physical, dated object.
Once the pre-production sample is approved and a counter-sample is kept, that same reference is what an inspector checks the finished shipment against — see setting up pre-shipment inspection and AQL sampling for how the defect list and the reference sample fit together. And before any of this starts, the twelve specification lines that a comparable quotation needs already assume a sample or limit reference exists — see comparing Vietnamese supplier quotations.
Frequently asked questions
1. Do I need all four stages for every product?
No. A simple, low-tolerance product might reasonably skip the proto stage and go straight to a pre-production sample; a garment with multiple sizes needs the fit stage explicitly. The pre-production sample is the one stage worth insisting on for almost any order, because it is the last checkpoint before the real production run starts.
2. How many units should a pre-production sample be?
Enough to see the process behave consistently rather than a single lucky unit — commonly a small handful, run on the actual production tooling and line rather than a separate sample line. The exact number depends on the product and the process; agree it with the factory rather than assuming one unit is representative.
3. What if the factory says keeping a counter-sample is unnecessary?
It costs one extra unit and some storage space, and it is the cheapest insurance in the entire order against a shipment dispute. A factory with nothing to hide has no reason to resist keeping one; treat resistance to the idea itself as worth asking about.
4. Can a sample be approved with known defects, to save time?
Yes, but only if the deviation is written into the approval record explicitly — “approved with the following exceptions” — rather than accepted verbally and forgotten. An unrecorded exception at the sample stage becomes an unexplained defect at inspection.
5. Who should physically hold the approved sample on the buyer’s side?
Whoever will be asked to judge a disputed shipment later — often the same person or team that will review the pre-shipment inspection report. If sourcing happens through an agent or a trading company, agree in writing whether the agent or the buyer holds the physical counter-sample.
Sources: general international manufacturing practice on sample approval stages and counter-samples; no single primary legal or regulatory source applies to this topic. Checked on 25 September 2026.