The four traps that cost buyers the most

First, the pattern. This is the most common complaint in this business, not bad luck. A sample shows what a factory can build once, under conditions it chose. It does not show what the line builds on a normal day.

What is usually missing is one of these three:

• No sealed golden sample, signed and dated, one copy with each side.
• No written acceptance range — what deviation is acceptable in colour, size, registration, weight and seal strength, and what is not.
• The sample came from a sample room, not from the line that ran the order.

What to do now, in this order:

1. Do not ship your only approved physical sample back to the factory. Photograph it, weigh it, measure it, and store it dark and dry.
2. Collect your written evidence: the approval email or chat, with a date and a version number, plus any photos sent at approval.
3. Compare bulk against the approved sample under the same light and with the same instruments, and write each difference down as a number, not as an adjective.
4. Send one written notice that states the facts and asks for a position. Keep it factual — "this looks cheap" is not a claim.
5. Check what your contract and PO actually say about the standard.

What I can't promise: I am not a lawyer, I do not file claims, and I do not guarantee recovery. Courts have dismissed buyers' claims outright where no sample was sealed, and cut agreed penalties sharply where the sample could no longer be examined. What I can do is fix the evidence set and tell you honestly how strong it is.

This is the problem I see most often: not a fake report, but a real one that does not cover your goods.

Check four things on the report itself, line by line:

1. The applicant — is it your supplier, a resin supplier, or a company you have never heard of?
2. The sample name, model and material grade — do they match your order and your specification?
3. The lab's accreditation scope — is the specific test method on the report inside the scope, and is the accreditation still valid?
4. The date — is it older than the last material change, and inside the validity your market accepts?

Then verify with the lab, not with the paper. China's market regulator publishes a three-step method: check the report number in the official system; check that the body's CMA accreditation is valid and covers each test item; then contact the issuing lab directly.

What I can't promise: I cannot certify that any report is genuine. Two hard limits. CNAS states that to confirm the authenticity of a report you must contact the laboratory itself — its own website only confirms whether a body is accredited. And the national report-number system is filled in by the labs themselves, so a number that cannot be found is a red flag, not a conclusion.

One more thing worth knowing: a US food contact notification is specific to the company that submitted it and to the specified use. An upstream resin maker's notification does not automatically cover the converter's finished bag.

Sometimes it is, and sometimes it is a downgrade with a friendly name. Nobody can tell the difference from a description — only from numbers agreed in advance.

Here is the uncomfortable part: there is no unified standard in China that draws that line. In a 2026 People's Daily investigation, two professors said the same thing — normal variation and a quality downgrade need to be defined, and the test is whether the seller disclosed the change truthfully and completely. Until that is settled, the only thing that protects you is your own specification.

So put this into the PO or its annex, per characteristic: the requirement, the first batch's measured value, the tolerance, the measuring method and conditions, the instrument model, and who decides. At a minimum: colour (ΔE, illuminant, observer, SCI/SCE), grammage and thickness, material structure, print registration, seal strength, total solvent residue, and barrier (oxygen and water vapour).

Two data points explain why this matters. In China's own national spot checks from 2023 to 2025, the failures that recur in films and bags are solvent residue and barrier performance — items a visual inspection cannot see. And in one Qingdao converter's case, a miscalculated curing time put an entire batch over the residue limit; the regulator's fine was about RMB 4,500, while the buyer's loss on the same batch would be orders of magnitude larger.

What I can't promise: I cannot guarantee that a factory will not drift or substitute. What I sell is verifiability, traceability and clause design — a change that happens gets found, and when it is found you have something to point at. And a standard pre-shipment inspection will not catch a different resin, a thinner layer or a changed adhesive: those are not visible.