The letter is a records request
The envelope does not look like much. A short questionnaire, a handful of questions about how the entered value on your entries was arrived at, and a response window that is fixed and not especially generous. You read it twice and notice what it is not doing. It is not inviting you to argue that first sale for export is available to you. It is asking whether the records that would support it already exist, and whether they existed before the letter turned up.
That distinction is the whole exercise. Importers who have used first sale for years tend to assume the exposure sits in the legal theory, and the theory is usually the least fragile part of the arrangement. The elements have been settled long enough that any working customs practitioner can recite them from memory. What comes apart under a records request is everything downstream of the theory: invoices from a factory you have never contracted with directly, proof that money moved the way those invoices say it moved, evidence that the goods were bound for the United States at the moment the first sale happened rather than at some later point when a trading company decided where to ship them.
You can usually tell within a day which position you are in. Either the file assembles itself out of systems you already run, or somebody starts emailing a middleman in another time zone asking for paper that may never have been retained.
What first sale actually asks you to prove
The framework has three moving parts, and every one of them is a documentary question wearing legal clothing.
There has to be a bona fide sale between the factory and the middleman, which means title and risk genuinely passed and money genuinely changed hands. The merchandise has to be clearly destined for the United States at the time of that sale rather than redirected afterwards. And the price in that first sale has to be arm’s length, unaffected by any relationship between the seller and the buyer.
What stands out about CBP’s 2026 first-sale valuation campaign is how little of the questioning turns on interpretation. The questions read closer to an inventory. Did you claim first sale on these entries. Do you hold the manufacturer’s invoice. Can you show the payment against it. Can you show the goods were headed to a US customer when the factory sold them. Each one is answerable in a single sentence and provable only with an attachment, which is why the response burden lands on whoever keeps the files rather than on whoever wrote the valuation memo years ago.
The element that trips people up is rarely the bona fide sale. It is the destination question. A factory selling to a Hong Kong trading company may have no idea, and no documentary reason to know, where the goods end up. If the order it received says nothing about a US buyer, if the cartons are marked only for the middleman, if the shipment moves into a regional warehouse before anyone decides which market it serves, then the goods were not clearly destined for the United States when the first sale took place. You can hold every invoice in the chain and still fail on that point alone.
The memo still matters. It just stops being the thing under examination.
Contemporaneous beats reconstructed
A file built while the business was running reads differently from a file built in the fortnight after a letter arrived, and you do not need to be a valuation specialist to see it. Dates run in the right order. Purchase order references carry through from the factory invoice to the middleman’s invoice to the remittance advice. Quantities agree across the whole set rather than most of it. Nothing has been helpfully retyped to make it tidier.
Reconstructed packets have a texture of their own, and it is recognisable. Invoices that all carry the same creation date. A spreadsheet standing in for payment records. A letter from a supplier confirming, in general terms, that the goods were always intended for the US market, written long after the fact and signed by somebody who was not there when the orders were placed. Consistent numbers with no source system behind them.
None of that is fraud, and treating it as fraud misreads what usually happened. Most of it is a company doing its honest best with whatever survived a systems migration, a change of freight forwarder, or the departure of the one person who understood the sourcing structure. It still fails the test, because the test is whether the documentation supported the claim at the time the claim was made. A valuation method you can only evidence retrospectively was never really a method. It was a position you happened to be able to defend.
The trail a genuine multi-tiered transaction leaves
When the structure is real, the paper follows it without anyone having to work hard. The set worth holding usually includes:
- the manufacturer’s commercial invoice to the middleman, with terms of sale stated
- purchase orders covering both legs of the transaction, cross-referenced to each other
- proof of payment in both legs, traceable to specific invoices rather than to monthly totals
- the bill of lading or air waybill, with consignee and destination consistent with a sale into the US
- any assists, royalties, commissions or design contributions that feed into value
- the intercompany or agency agreement explaining why a middleman sits in the chain at all
- packing lists and production records tying particular goods to a particular US order
Notice how much of that sits outside the compliance team’s own systems. Sourcing negotiated the structure. Finance holds the remittances. A freight forwarder holds the transport documents, sometimes only for as long as its own retention policy requires (which is often shorter than yours). Nobody set out to scatter the evidence, and yet it is scattered.
Related parties make the price question harder
Arm’s length stops being a safe assumption the moment the factory and the middleman sit inside the same corporate group, and that is an extremely common shape in Asian sourcing chains. Valuation practice has mechanisms for testing it: comparing the price against sales of identical or similar goods to unrelated buyers, examining the circumstances of the sale itself, or building the price up from all costs plus an amount for profit and general expenses consistent with the industry. Every one of those is an evidentiary project rather than an arithmetic one.
The thing that catches otherwise careful importers is a transfer pricing study written for income tax purposes and then quietly reused for customs. They answer different questions. A study can be entirely defensible to a tax authority and still say nothing useful about whether one particular first sale price was set at arm’s length for valuation purposes. Reviewers read it, find it addresses the group’s overall margin profile across a whole product family, and ask the question again in narrower terms.
Who owns the answer inside your own company
The importer of record carries the duty of reasonable care, and a questionnaire is the moment that duty becomes concrete rather than theoretical. Brokers file entries from the data you give them. A broker can tell you what was declared and can pull entry summaries out of ACE, but it cannot manufacture the underlying commercial documents, and it was not in the room when the sourcing structure was designed.
So the response usually gets assembled by a small group who have never worked on anything together: someone from trade compliance, someone in finance who can actually trace payments, someone from sourcing who remembers why the middleman was inserted, and outside counsel where the structure is aggressive or the exposure is material.
Naming that group before a letter arrives costs almost nothing. Naming it afterwards costs you most of the response window, because the first several days go on working out who has access to what and which system the payment records actually live in.
There is a version of this that goes smoothly, and it looks dull. A short internal note saying which entries use first sale, which supplier chains they run through, where each document class is stored, and who to call at the middleman. Dull, and worth more than any memo on valuation theory.
The valuation basis itself could move
There is a second thing running alongside the questionnaires. A Last Sale Valuation Act was introduced in the Senate in February 2026 which would assess duty on the last sale before importation and so curtail first sale as a route. It is proposed legislation and not law, and plenty of introduced bills go nowhere at all. Reading it as settled would be a mistake, and so would ignoring it.
What it does change is how you think about dependency. A duty saving that works only because of one valuation method has a single point of failure built into it, and importers who already keep a firm grip on the documents behind every import fee sit in a different position from ones whose whole advantage rests on a structure they cannot fully evidence. The ordinary correction mechanisms stay available either way, through post summary corrections before liquidation and a protest after it, and those only help when the underlying records exist to support the corrected position. Paper first, position second.
Which is where the questionnaire actually points. Not at the theory, and not at the structure, but at the filing cabinet.
