A client who assembles industrial pumps called me in a bad mood after the July tariff handoff. His Chinese components now carry a combined 37.5% duty, and roughly 40% of his finished units get exported to Latin America. He wanted to talk about resourcing. I told him to hold that conversation for a month and look at drawback first, because he had been paying duty on goods that left the country and never asked for it back.
The July change matters for the size of that recovery. Since 24 July 2026 a Section 301 forced-labor action has applied additional duties of 10% or 12.5% across 60 economies covering close to 99% of US goods imports, stacking on rates that were already there rather than replacing them. USTR Higher duty paid on goods that later leave the country means a larger refund sitting unclaimed.
He is not unusual. Duty drawback is the least glamorous line item in trade compliance and the one with the most money sitting in it right now, precisely because rates went up. When duties were 3% nobody chased the refund. At 37.5%, ignoring it is a financing decision you are making by accident.
What drawback actually is
Drawback is a CBP program that refunds up to 99% of eligible import duties, taxes and certain fees when the imported goods, or articles made from them, are later exported or destroyed. CBP The statutory logic is straightforward: the United States taxes goods entering its commerce, so if the goods do not stay in its commerce, the duty comes back.
That 99% figure is not a negotiation. It is the statutory retention, with CBP keeping 1%. What is negotiable, in the sense that it depends entirely on your records, is whether you can prove the link between a specific duty-paid import and a specific export.
Section 301 duties are eligible, and that is the headline
Per CBP guidance, duties paid under Section 301 of the Trade Act of 1974 are eligible for drawback. CBP Given that China Section 301 rates run from 7.5% to well over 100% depending on which list a product sits on, that eligibility is where the largest recovery opportunity in the program currently lives.
Do the arithmetic on my pump client. Say he imports $6 million of Chinese components annually at a combined 37.5%, which is $2.25 million in duty. If 40% of the resulting units are exported and he can document the match, the recoverable duty approaches $900,000 before the 1% retention. That is not a rounding error, and it is not a tax scheme. It is a refund of duty on goods that did not remain in the US market.
I am deliberately calling that a worked illustration rather than a promise. The recovery depends on documentation quality, drawback type and how cleanly your bill of materials ties imports to exports. But the order of magnitude is why this belongs on a CFO's desk in a high-tariff year.
Note also that the 37.5% in that example is a floor rather than a ceiling for some lines. Where the new 12.5% layer stacks onto pre-existing Section 301 rates that already ran from 7.5% upward, the combined figure can land higher, and the recoverable pool grows with it. Model your own lines rather than borrowing my arithmetic.
The exception that catches steel and aluminum importers
Here is the limit that surprises people, and I lead with it because assuming otherwise wastes a quarter of planning. Section 232 duties on steel, aluminum, copper and automobiles are not eligible for drawback. CBP
So if your import book is primarily metals under a Section 232 action, drawback does not touch those duties at all. I have sat in meetings where a metals importer built a recovery forecast for weeks before anyone checked this. Confirm which authority your duty was assessed under before you model anything, because Section 301 and Section 232 look identical on a cost line and behave completely differently here.
Check the exemptions before you count the refund
There is a step people skip that can make the whole calculation moot. The July forced-labor action carries extensive product exemptions, set out across a Federal Register notice running to several hundred pages, and more than 2,120 tariff lines are excluded from the new duties. USTR Goods already subject to Section 232, and qualifying USMCA goods from Canada and Mexico, sit outside the action as well. Honigman
Two consequences follow. If your product is exempt, there is no new duty to recover on that line, and a forecast built on the headline rate will overstate your pool. If your product is exempt because it falls under Section 232 instead, you have moved into the category that cannot be drawn back at all. Read the annex against your own HTS lines before you size the opportunity, not after.
The July forced-labor action is a Section 301 determination, which is the eligible family, but I would not treat that as settled for your specific entries without confirming the treatment with your broker or seeking a CBP ruling. New actions sometimes arrive with their own drawback language, and "it is Section 301 so it must be eligible" is the kind of inference that reads well in a memo and badly in a claim denial.
The real obstacle is proof, not eligibility
Almost every failed drawback program I have reviewed failed on records rather than on law. The requirement is not simply that exports happened. It is that exports can be tied back to duty-paid imports at the SKU and quantity level.
CBP guidance requires drawback filers to report both the Chapter 99 and the Chapter 1-97 HTS numbers, plus quantity and value, in the same order as they appeared on the ACE underlying import entry. CBP Read that again if you run the data side, because it dictates your architecture. Your claim has to mirror the original entry structure, which means the import data you keep must preserve that order rather than normalizing it into whatever shape your warehouse system prefers.
The practical failure mode is mundane. A company imports components, commingles them in inventory, builds finished goods, exports some, and six months later cannot demonstrate which units contained duty-paid material from which entry. The duty was paid, the export happened, and the claim still dies because the middle link is missing.
The clocks you are working against
Drawback has deadlines and CBP enforces them strictly. Claims generally must be filed within statutory limits that run to five years from the date of import, with a separate three-year window tied to the date of export, and the specifics vary by drawback type. CBP
Two implications follow. First, if you have never filed, you may be able to reach backward into prior entries rather than only forward, which is the single fact most likely to change a skeptical CFO's mind. Second, the five-year outer boundary means the oldest recoverable duty is expiring quietly every month you delay. A program you start in the fourth quarter captures entries a program you start next spring will have lost.
Which flavor of drawback fits your operation
People talk about drawback as one thing. It is a family, and picking the wrong member creates unnecessary work.
Manufacturing drawback applies when imported material is used to produce an article that is then exported. This is my pump client. It carries the heaviest documentation burden because you must connect inputs to outputs through a production record.
Unused merchandise drawback applies when goods are exported in essentially the same condition as imported. Distributors and companies with reverse logistics flows land here, and the documentation is lighter because no production step needs proving.
Rejected merchandise drawback covers goods that did not conform to specification or were shipped without consent and are returned or destroyed. Anyone running meaningful returns volume out of an import book should check this one, because the duty on defective goods is recoverable and routinely is not recovered.
When it pencils out and when it does not
I will not pretend this is free. A drawback program costs money to build, whether you hire a specialist broker who works on contingency or build internal capability. So the threshold question is volume.
My rough screen: if your annual duty on exported or destroyed goods is in the low six figures or above, the program almost certainly pays for itself. Below that, it depends on how clean your data already is. A company with a disciplined ERP and clear lot traceability can run a small program cheaply. A company that commingles inventory and reconciles by spreadsheet will spend more on the reconstruction than it recovers.
One thing I tell every client: do not let a contingency-fee provider talk you into a claim your records cannot defend. Drawback claims are subject to audit, and a refund you have to give back with penalties is worse than a refund you never claimed.
How drawback interacts with the other tools
Drawback is a refund after the fact. Deferral tools change when duty is assessed in the first place, and the two answer different questions.
A bonded warehouse lets you land goods and postpone the duty payment until withdrawal for consumption, covered in the bonded warehouse deferral strategy. A foreign-trade zone can go further, and for goods that are re-exported from the zone you may avoid the duty rather than paying and reclaiming it, which is structurally cleaner than drawback when the flows fit. That is covered in the foreign-trade zones guide. If a large share of your imports are destined for export, look hard at the zone option before building a drawback program, because not paying is simpler than paying and recovering.
Refund questions arising from the earlier IEEPA duties run on a separate legal track from drawback entirely, and I keep them apart in the IEEPA tariff refund guide. Drawback is a statutory program you claim; that one turns on litigation.
Also check who is bearing the duty in the first place. If your terms put the import duty on a counterparty, the drawback right may not be yours to claim, and I have seen that fight surface after a program was already built. The DDP versus DAP breakdown covers where that liability lands.
Your first two weeks, concretely
- Identify which authority your duties were assessed under. Section 301 is eligible; Section 232 on steel, aluminum, copper and autos is not. This single check determines whether the rest is worth doing.
- Measure your export or destruction share of imported value. No exports, no drawback. A 40% export share on a duty-heavy book is a serious opportunity.
- Pull five years of import entries. The outer statutory window reaches back that far, so the historical claim may dwarf the forward-looking one.
- Test whether you can tie one export back to one import at SKU and quantity level. Pick a single real shipment and trace it end to end. If that takes a week, you have a data project before you have a drawback program.
- Confirm your import data preserves the ACE entry structure, including Chapter 99 and Chapter 1-97 numbers in the original order.
- Check whether an FTZ would serve better than a refund for the export-bound share of your volume.
- Decide who owns the claim contractually before you invest in building it.
The reason I push this now rather than in January is arithmetic, not urgency theater. Duty rates stepped up in July, the recoverable pool grew with them, and the five-year window keeps closing on your oldest entries whether or not anyone is watching. My pump client's resourcing question was reasonable, and he may still move some production. But he was about to redesign a supply chain to avoid a duty he could have recovered on nearly half his volume, and that is an expensive way to solve a paperwork problem.