Published: Nov 2026 · by HYCTON, sourcing office for hydraulic fittings in China
A sample from a Chinese hydraulic factory is not evidence of what you will receive. It is evidence of what the plant can do when it is paying attention.
This is the most expensive misunderstanding in small-volume sourcing, and almost every buyer makes it in the same order: request a sample, judge the sample, place the order, discover at the container that the batch is not the sample.
The mistake is not in judging the sample. It is in the sequence. By the time a sample exists, the plant has already decided how much attention your order deserves — and that decision was made before the sample was cut.
What a sample actually tells you
A sample proves three things and only three things:
- The plant can hit the drawing when it tries.
- The plant understood your specification well enough to make one.
- Someone at the plant is capable of answering you.
It does not tell you whether the production line will be set up the same way, whether the same operator will run it, whether the material will be the same heat, or whether the plating batch will be the same thickness. Those are the variables that fail.
And there is a structural reason the sample is always good: on a single piece, the cost of doing it properly is trivial. A machinist can take an extra ten minutes on one part. That same ten minutes across two thousand parts is a conversation about margin — and the answer to that conversation is decided by the kind of plant it is, not by how good the sample looked.
Why the order matters
Verification before the sample changes the sample from a demonstration into a measurement.
If you have already checked the plant — its licence, its export record, its actual production floor, the machines it owns — then when the sample arrives you are no longer asking “can they do this?” You are asking “did they do this?” Those are very different questions, and only the second one is answerable.
If you have not checked the plant, the sample is doing work it cannot do. You are extrapolating from one good piece to a production run, with no information about the system that will produce it.
The five checks, in order
Check 1: Is it a plant, and does it export?
Two documents and ten minutes. The business licence tells you the registered activity and the registered capital. The export record tells you whether this company has actually shipped goods under its own name.
What you are looking for: the registered activity matches what you are buying, and there is a real export history. What should worry you: a licence registered as trading or consulting while the website says factory, or a company with no export record that is promising to ship you a container.
Neither document is proof of quality. Both are proof of existence, and existence is the first thing to establish.
Check 2: What machines are on the floor?
A plant that runs the right machines for your part is a different proposition from a plant that outsources the operation you care about.
For adapters and fittings, the questions are concrete: do they have the CNC turning capacity for the body, do they forge or buy forgings, and do they run their own plating or send it out?
The plating question is the one that matters most and gets asked least. If a plant sends parts out for plating, it does not control the finish — and finish is where most rejected hydraulic batches fail. Ask who plates the parts and whether you can see that plant too.
Check 3: Does the sample match the drawing, measured?
Not photographed. Measured, with the numbers written down.
For a hydraulic fitting the measurements that matter are: thread standard and pitch confirmed with a gauge, across-flats, overall length, and the sealing surface. If the part is a JIC 37 degree flare, the angle is a measurement, not an impression.
Ask for the actual readings against each callout on your drawing. A plant that can produce a dimensional report for a sample is a plant that has a quality function. A plant that sends a photo is a plant that has a camera.
Check 4: Is there a first-article inspection, or just a part?
The difference between a sample and a first article is documentation. A first article says: here are the measurements, here is the material certificate, here is the batch, here is who inspected it.
For a first order this may be more than the plant normally provides. That is exactly why asking for it is a useful test. A plant that can produce a material certificate when asked has the certificate. A plant that cannot has been buying material without traceability, which is a fact about the plant you needed to know.
Check 5: Is the trial order a fair test of the production line?
The last check is on your own order, not on the plant.
A fifty-piece trial run on a machine set up for that run tells you about the setup, not the line. It is still worth doing — it confirms the thread, the finish and the packaging — but do not mistake it for a guarantee about volume.
What makes a trial meaningful is agreeing, before it runs, what would make you reject it: which measurements, to what tolerance, and what happens if they fail. Written down, agreed, before the first cut. That single document converts a trial from a favour into a test.
The sequence, as a checklist
| Step | Ask for | What a bad answer looks like |
|---|---|---|
| 1. Existence | Business licence, export record, factory name and location | “We are a factory” with no documents |
| 2. Capability | Machine list, forging or buying, who does the plating | Everything in-house, nothing specified |
| 3. Communication | Questions about the drawing before quoting | A price within the hour, no questions |
| 4. First article | Dimensional report, material certificate | Photos of a part on a desk |
| 5. Trial | Agreed accept and reject criteria, in writing | “Do not worry, it will be fine” |
The uncomfortable part
All of this requires someone to be physically near the plant. A video call can show you a machine; it cannot tell you whether the machine was running that day, whether the operator was the one who will run your order, or whether the plating shop down the road is the reason the finish is inconsistent.
This is the actual reason a sourcing office exists, and the reason “we can check it for you” from a supplier six thousand kilometres away is worth very little. Verification is a physical act. It requires a person who can stand on the floor, look at the rack, and ask the question that was not on the form.
That is not a service you can buy as a document. It is why buying from China without anyone on the ground is a bet, and why the bet is usually decided before the sample ever reaches you.
What to do differently on your next order
- Before you request anything, ask for the factory name and its export record. If you cannot get them, stop.
- Ask three technical questions about the part. Count how many come back answered.
- When the sample arrives, ask for the measurements, not a photo.
- Agree the reject criteria in writing before the trial runs.
- Keep the trial small and the specification strict. The point is information, not volume.
None of this is exotic. It is just the order most buyers never do it in — and the order is what decides whether the sample meant anything.
HYCTON — sourcing office in Dalian, China for hydraulic fittings & hose components · www.hycton.com