The audit's real question is whether the origin claim survives a document request
Treat the visit as a rehearsal for a request for information, not as a quality inspection with a compliance section bolted on. The legal shape of the question is old and stable: under 19 C.F.R. § 134.1(b), the operations performed or materials added in another country must amount to a substantial transformation before that country becomes the country of origin. Assembly that leaves the identity of the goods intact does not get there.
Two limits are worth carrying into the room so you do not overstate the rule. That definition is written within the meaning of its own part, and Part 134 is the marking regime; for duty purposes the standard applied comes from case law, and goods claiming preference under the North American agreement are judged under a different set of rules altogether. The practical version is unchanged: whichever test applies, it is answered with production records, not with a certificate. What the test actually asks for is a longer subject in its own right — what counts as substantial transformation in a Vietnamese factory is where the detail lives.
Red flag one: a local-content percentage that does not exist in Vietnamese law
If a factory tells you the goods are Vietnamese because local value added crosses some percentage, ask which text says so. Vietnam's non-preferential origin rules do not contain a blanket value-added threshold. The principle is the country of the last operation that substantially transforms the goods, and the operational criteria sit in a product-specific rules list organised by HS code, offering either a change in tariff classification or a local value content figure that varies by heading. The percentages exist, but per product, not as a general pass mark.
The number you will hear most often is the wrong one. According to reporting in the Vietnamese official press, the widely quoted rule that local value added of 30% plus a non-simple final operation makes goods Vietnamese has lived only in draft since 2019, and as of the ministry's submission of a draft decree in June 2026 it still was not law, with the draft itself built on voluntary self-declaration and no published percentage. A supplier quoting it is not lying to you on purpose; they are quoting a summary of a rule that was never enacted.
The replacement question is narrow and answerable: which HS heading does this finished good fall under, which product-specific rule applies to it, and does this bill of materials satisfy that rule. A factory that can answer that has done the work. A factory that answers with a percentage has not.
Red flag two: the Vietnamese step is on the list of operations that do not count
Vietnamese origin rules list operations that are disregarded when origin is determined, and the list is unusually concrete. According to the text of Decree 31/2018/NĐ-CP, it covers preserving operations during transport and storage such as ventilation, drying, chilling, salting and sulphur treatment; dust removal; sifting and sorting; matching and making up sets; changing packaging; bottling and boxing; affixing labels and marks; simple mixing; simple assembly of parts; any combination of these; and slaughtering animals.
Read that list against what you just watched on the floor. If the entire Vietnamese contribution is boxing imported goods and putting a label on them, the origin claim is not weak, it is absent — and the certificate a supplier offers to arrange changes nothing about that. This is the flag that most often survives a smooth audit, because the operation is real, the workers are real, and the building is real; only the legal effect of the work is missing.
One number in the same decree gets misused in the opposite direction. The de minimis tolerance, which allows non-originating materials up to 15% of the FOB value, with textiles measured by weight, is a tolerance rule sitting on top of a rule already satisfied. Hearing it offered as the reason goods are Vietnamese is itself a red flag: it means the person explaining the origin claim has the framework upside down.
Red flag three: nothing can be produced while you are standing there
A document that arrives after the visit is a document that may have been made after the visit. The test is simple and slightly rude: ask for input purchase invoices and packing lists for the specific lots in your order, the work orders covering the production run, and the output records for those lines, and ask for them at the desk, now. What comes back in five minutes is evidence. What comes back in five days is a narrative.
Build the request list before travelling, so the visit is a verification rather than a discovery exercise; the documents worth asking a Vietnamese supplier for mostly need to be requested in the contract, not in the meeting room. And capture what you see while you are still in the building — photographs of lines and labels, serial numbers, the names on the documents — which in an industrial park with uneven connectivity means checking beforehand whether your team's tooling really works offline, because notes that failed to sync read exactly like notes nobody took.
- Ask for records tied to specific lots in your order, never for samples of what the factory can produce in general
- Photograph the input packaging, not only the finished goods — it is where the previous country of origin is usually still printed
- Note who signed each document and in what role; a file with no names is not a file
- Leave with copies, or with the file logged in your own system before you get in the car
Red flag four: capacity, headcount and the order do not line up
Arithmetic catches what interviews do not. Count lines, count people, ask about shifts, and compare that against the volume you are about to buy plus what the factory is already committed to. A plant that promises to add capacity on demand is often describing a subcontractor you have not audited, and origin questions do not survive a step that happened at an address nobody wrote down.
Undeclared subcontracting is the quiet version of the same problem. It is not necessarily fraudulent — capacity smoothing is normal in the region — but it puts part of your production outside every record you just inspected. Ask directly whether any process step leaves the site, and get the answer in writing, because that is the sentence you will want on file when someone asks you where the goods were made.
The other side of the flag is a Vietnamese penalty, not just a rejected shipment
The exposure is not only at the border of the destination market. Vietnam replaced its customs penalty framework with Decree 169/2026/NĐ-CP, in force from July 1, 2026, superseding Decree 128/2020/NĐ-CP in full and Article 2 of Decree 102/2021/NĐ-CP. Its Article 18 deals specifically with falsely claimed Vietnamese origin across export, import, temporary import for re-export, transit and transhipment, with fines in five bands scaled to the value of the goods, plus confiscation of the goods and forced destruction or recovery of an equivalent amount.
Time works against a quiet fix as well. The decree sets limitation periods of five years for tax-administration violations such as evasion or under-declaration that has not reached the criminal threshold, and two years for other customs violations — long enough that a shipment cleared this season stays open well past the point where anyone remembers which factory ran it. That is the practical reason to stop an order on a documentary red flag rather than negotiate around it, and it is the same file that answers the buyer-side question later, when a customs request for information asks you to prove the origin claim.
