The Tariffs Got Refunded. De Minimis Did Not Come Back.
In February 2026 the Supreme Court struck down the IEEPA tariffs and refunds began flowing. The $800 de minimis exemption stayed dead, because it was never a tariff. That distinction decides how you should be rebuilding DTC pricing and fulfilment, and most brands have it backwards.
Annie Chan··17 min read
Through most of 2025 the sensible-sounding advice inside consumer brands was to wait. The tariff actions were extraordinary, they were legally contested, and a lot of experienced people expected the courts to unwind them. Holding your fulfilment architecture steady while the litigation resolved looked like the disciplined choice rather than the lazy one.
The litigation did resolve. On 20 February 2026 the Supreme Court held, 6 to 3, that the International Emergency Economic Powers Act does not authorise the President to impose tariffs at all. The IEEPA duties came off. Refund claims opened on duties paid from early February and early April 2025 onward. For a lot of importers this was the single largest cash event of the year.
And the $800 de minimis exemption did not come back.
That is the whole lesson. Duty rates are political and reversible. Entry procedures are administrative and sticky. De minimis was an entry procedure that everybody had been modelling as a duty rate, which is why so many brands hedged against the wrong risk for eighteen months.
What actually happened, in order
It is worth laying the sequence out precisely, because the compressed version circulating in most brand teams (the $800 rule went away in 2025) is missing the part that matters.
Section 321 of the Tariff Act of 1930, codified at 19 U.S.C. 1321(a)(2)(C), let shipments valued at $800 or less enter the United States free of duty and, more importantly, free of a formal or informal customs entry. That second half is the part almost nobody priced.
2 May 2025. Executive Order 14256 removed de minimis eligibility for products of China and Hong Kong. Postal traffic was put on a choice between an ad valorem duty and a per-item specific duty. The announced figures moved several times over the following weeks as the US and China negotiated, which is itself a useful data point: nobody should have been planning against any single published rate from that period.
30 July 2025. A further executive order, titled Suspending Duty-Free De Minimis Treatment for All Countries, extended the suspension worldwide for goods entered for consumption on or after 12:01 a.m. EDT on 29 August 2025.
20 February 2026. The Supreme Court held that IEEPA confers no tariff authority. Both of the executive orders above rested on IEEPA.
24 February 2026. A Section 122 balance-of-payments surcharge of 10 percentage points ad valorem on imports from all origins took effect as a stopgap. Section 122 is capped by statute at 150 days, so it expired on 24 July 2026 absent an act of Congress. The durable instruments are Sections 232 and 301, which is where you should expect the rate structure to keep being rebuilt.
24 June 2026. US Customs and Border Protection published two interim final rules of its own. One amended 19 CFR 10.151 to indefinitely suspend the de minimis exemption for merchandise arriving by every mode other than the international postal network. The companion rule did the same for mail and created a new postal informal entry process.
1 July 2027. Congress has already repealed the statute. The One Big Beautiful Bill Act (H.R.1, section 70531) strikes the Section 321(a)(2) de minimis privilege for commercial shipments with effect from that date, and adds civil penalties of up to $5,000 for a first violation and up to $10,000 for each subsequent violation of the customs laws via the administrative exemptions.
Read that list again and notice the structure. The tariff authority was struck down in February. CBP re-founded the de minimis suspension in June on its own regulatory authority, which the Supreme Court decision did not touch. And behind that sits a statutory repeal already passed by Congress with a fixed date. Three separate legs. Knock out one and the thing still stands.
"A brand waiting for de minimis to return in 2026 was waiting for a court to overturn an executive order, a regulator, and an act of Congress, in that order, on the same question."
The category error: it was never a duty rate
When finance teams modelled the change, almost all of them modelled it as a duty. Take the landed cost, add a percentage, see what it does to contribution margin, decide how much to pass through. That work was competent and it was answering the smaller half of the question.
De minimis relieved two different things at once. It relieved duty, which is a variable cost proportional to value. And it relieved the requirement to file an entry, which is a fixed cost per shipment and has almost nothing to do with value.
Look at what the new postal informal entry process requires: a customs bond, ten-digit HTSUS classification, and filing by a licensed broker or another party with entry rights. None of those requirements gets cheaper because the parcel is worth $19 instead of $190. A classification decision costs the same to make either way. A bond costs the same. A broker filing costs roughly the same.
The repeal of de minimis is a regressive tax on low average order value. Not because the duty is regressive, but because the paperwork is.
This is why the brands hurt worst are not the ones with the thinnest margins. They are the ones with the smallest baskets. A supplements brand shipping a single $28 bottle and a furniture brand shipping a $1,400 chair face the same entry obligation, and one of them can absorb it invisibly.
Three costs, and your model probably has one
Rebuild the model with three lines instead of one.
Cost one: the ad valorem stack
This is the part you already modelled. It is also the part most likely to change again, because it now depends on Section 232 and Section 301 actions that move by product category rather than by blanket order. Practical consequence: this line should be maintained per HTS code, not per country, and it should be owned by someone who reads Federal Register notices. If your current model has a single country-level tariff assumption applied across the catalogue, it is already wrong for part of your catalogue.
I am deliberately not giving you a rate. Rates in this environment have a shelf life measured in months, they vary by classification, and any number printed in an article is a liability. Get your top twenty HTS codes classified properly and get current rates from your broker in writing.
Cost two: the per-entry fixed cost
Brokerage, bond, classification, data preparation, and the internal labour of maintaining accurate product data good enough to survive an entry. Your broker will quote you the external portion. The internal portion is usually larger and almost never gets counted.
The single most useful number to extract from this exercise is your fixed cost per entry, and then the order value at which that fixed cost falls below whatever percentage of revenue you are willing to spend on customs administration. Below that value, direct cross-border shipment of a single order is structurally uneconomic no matter how you price it.
Cost three: variance
The one nobody puts in the model. Entry processes fail. Classifications get questioned. Parcels get held. Under de minimis a mis-declared parcel simply went through; now it becomes a hold, a query, a delay, and occasionally a refusal. The cost shows up in support tickets, in refunds, in the delta between promised and actual delivery dates, and in the churn that follows.
If you serve a category where delivery-date reliability drives repeat rate, this third line can exceed the first two combined, and it will not appear in any spreadsheet until you go looking for it in your support data.
Why the answer is re-bundling, not repricing
The reflex response to a cost shock is to raise prices. In this case that is the wrong lever, and it is wrong for a specific reason: you are being charged per parcel, so the efficient response is to send fewer, larger parcels, not to charge more for the same small ones.
Four levers, roughly in order of how quickly they pay back.
Raise units per parcel by design, not by discount. Multi-packs, refill bundles, sets, and starter kits move average units per order without touching your unit price. A three-pack at a modest bundle discount can be more profitable per parcel than a single unit at full price once a fixed entry cost is in the equation.
Move replenishment onto a cadence. If a customer buys monthly, converting them to a quarterly shipment cuts your entries by two thirds for the same revenue. Subscription programmes built for retention now have a second, larger justification.
Set a real free-shipping threshold and set it above your break line. Most brands set thresholds by looking at AOV distribution. Set it by looking at your per-entry cost instead, and accept that it will be higher than the number marketing wants.
Prune the low-value tail. SKUs that mostly ship alone and mostly ship cheap are now negative-contribution items disguised as catalogue breadth. Some of them belong in bundles only.
"The instinct is to protect the $25 order by charging $29 for it. The structural answer is to stop the $25 order from existing as a standalone shipment."
Four fulfilment architectures, honestly compared
One: keep shipping direct from origin
Still viable, but only above your break line and only with entry data discipline you probably do not have today. The winners here are high-AOV, low-frequency, differentiated goods. The losers are everything that looked like a bargain.
Note what is no longer available: consolidating many small orders to dodge entry. That was the mechanic behind a large share of the sourcing-agent and freight-forwarder value proposition, and it is gone. If a partner is still selling you consolidation as a duty-avoidance strategy rather than a freight-cost strategy, that is a signal about the partner.
Two: forward-deploy inventory into the destination market
Import in bulk, clear once, fulfil domestically. One entry covers thousands of units, which collapses cost two to near zero per order and eliminates cost three almost entirely. This is the architecture the rules are pushing everyone toward, and it is the right answer for most brands with predictable demand.
The price you pay is working capital and forecast risk. You are converting a variable cost into an inventory position. For a brand with wide SKU counts and unpredictable demand, that trade can be worse than the tariff.
Three: bonded warehousing and foreign trade zones
Defers duty until goods leave the zone, and can avoid it entirely on goods that are re-exported. Genuinely useful if you are running multi-market distribution from a single hub, or if you hold significant inventory for long periods. Overhead-heavy for small operations. Worth an actual quote rather than an assumption, because the break-even is lower than most brand teams expect.
Four: change where the goods are made
The slowest lever and the one most often invoked as a gesture. Moving production is a two-to-four-year project for anything with tooling, and origin rules mean partial moves frequently do not achieve the origin change you were paying for. Have the conversation, but do not put it in this year's plan as a mitigation.
If you are at the stage of rethinking where and how you buy rather than only where you ship from, the sourcing guide for brand owners covers the supplier-side half of this decision.
DDP is now a survival requirement
Under de minimis, whether you shipped delivered duty paid or delivered duty unpaid was a finance preference. It is now a retention decision.
A DDU parcel that arrives with a demand for duty plus a carrier handling fee produces one of three outcomes: the customer pays and never buys again, the customer refuses and you eat the return, or the customer pays and then opens a chargeback. There is no fourth outcome where this goes well.
Show the full landed price at checkout, collect it, and remit it. Every parcel where the customer discovers the cost at the door is a customer you have bought once and lost once.
Practically this means your checkout needs a duty calculation that is right often enough to be worth trusting, which in turn means your product data needs HTS codes and country of origin on every SKU. Most catalogues do not have this. Fixing it is unglamorous and it is the highest-return work available to a cross-border DTC team right now.
The European mirror, and why it points the other way
Europe is making the same decision and implementing it differently, and the difference is more interesting than the similarity.
The EU has agreed to remove the EUR 150 customs duty exemption for parcels sent directly from third countries to EU consumers. In the interim regime, running from 1 July 2026 until the EU Customs Data Hub is in place around 2028, such consignments carry a flat customs duty of EUR 3 per consignment. A separate e-commerce handling fee is mandated to be collected from November 2026, with the structure still being worked out at the time of writing.
Now compare. The United States responded by removing an entry exemption, which imposes a process. The European Union responded by imposing a flat charge, which imposes a price.
"A price is something you can put in a spreadsheet on Monday. A process is something you rebuild your operations around for a year. They are not the same magnitude of change even when the euro amount is similar."
The practical consequence is genuinely counterintuitive, and I have not seen anyone state it: for the next two years, direct-to-consumer cross-border shipment into the EU is administratively cheaper than into the US for the first time in the modern history of DTC. A EUR 3 flat duty per parcel is a known, small, arithmetic problem. A mandatory customs entry with bond, classification and broker filing is not.
If your brand has been treating the US as the default first cross-border market and Europe as the complicated one, that assumption was formed under rules that no longer apply. It is worth re-running. This is separate from the demand question, and if you have not built a view on where demand actually is, the market entry framing here is the prior question.
Two caveats before anyone reorganises around this. The EU interim regime is explicitly interim, and the 2028 end state is a full customs regime, not a permanent EUR 3. And the handling fee arriving in November 2026 has not had its structure published, so the total per-parcel cost is not yet knowable.
What is still moving
Two things worth watching rather than planning against.
CBP has said it will run a voluntary test of a new electronic informal entry process, Entry Type 13, in ACE beginning 22 September 2026, covering international mail shipments valued at $2,500 or less. If that test works, the per-entry cost of the postal channel falls, and some of the arithmetic above shifts. It does not restore de minimis. It makes the replacement cheaper to comply with, which is a different and smaller thing, but for high-volume low-value senders it is the single most consequential development in the pipeline.
The tariff stack itself will keep moving under Sections 232 and 301. Expect category-specific actions rather than blanket ones, which means the discipline you need is per-HTS monitoring rather than a country-level assumption.
The next ninety days
Get HTS codes and country of origin onto every SKU in your product data, not in a spreadsheet held by one person in operations. This is the prerequisite for everything else and it usually takes longer than teams expect.
Get a written per-entry cost from your broker, including bond, filing, and classification maintenance. Compute your break-even order value. Put that number in front of the person who sets your free-shipping threshold.
Audit the low-value tail of your catalogue against that break line. Decide which SKUs become bundle-only.
Confirm you are shipping DDP with duty shown and collected at checkout, in every market. If any market is still DDU, find out what your refusal and chargeback rate looks like there before you argue about it.
Model the forward-deployed inventory option properly, with working capital cost included, for your top market. Not as a strategy exercise, as an arithmetic one.
If you have IEEPA duties paid between early 2025 and February 2026 and have not filed for refunds, that is a cash conversation with your customs counsel, and it has deadlines.
Frequently asked questions
Is the $800 de minimis exemption coming back?
There is no realistic path to it returning. The executive orders that first suspended it rested on IEEPA, which the Supreme Court invalidated in February 2026, but CBP separately suspended the exemption indefinitely by interim final rule on 24 June 2026 using its own regulatory authority, and Congress has repealed the underlying statutory privilege for commercial shipments with effect from 1 July 2027. Three independent legs support the change.
Does this mean every parcel now needs a customs broker?
Shipments valued at $800 or less arriving by modes other than international mail must now use formal or informal entry procedures. The new postal informal entry process for mail requires a customs bond, ten-digit HTSUS classification, and filing by a licensed broker or another party with entry rights. In practice most brands will route this through a broker or a fulfilment partner who provides the capability rather than building it in house.
What changed in the EU, and when?
The EU agreed to remove the EUR 150 customs duty exemption for parcels sent directly from outside the EU to EU consumers. An interim flat customs duty of EUR 3 per consignment applies from 1 July 2026 until the full customs reform takes effect around 2028, and a separate e-commerce handling fee is to be collected from November 2026 with details still pending.
Should we raise prices or add shipping fees?
Usually neither as the first move. Because the new cost is largely fixed per parcel rather than proportional to value, the higher-return response is to increase units per parcel through bundles, replenishment cadence and a free-shipping threshold set above your break-even order value. Price increases become the right lever only after you have exhausted the per-parcel economics.
Do the tariff refunds change any of this?
They change your cash position, not your architecture. IEEPA duties paid from early February 2025 (the fentanyl-related actions) and early April 2025 (the reciprocal actions) through late February 2026 are refund-eligible, and that is worth pursuing with counsel. But the de minimis suspension was re-established on separate authority and the refunds have no bearing on it.
We sell through Temu and marketplaces rather than our own site. Does this apply?
It applies to whoever is the importer of record, which under most marketplace local-fulfilment models is now you rather than the platform. The platforms restructured toward domestic inventory partly for this reason. That shift is covered in more detail in the Temu and Shein analysis.
The judgement underneath all of this
The 2025 tariff actions were a shock, and shocks get reversed. Plenty of them were. What happened alongside the shock was quieter and permanent: the United States and the European Union both concluded, independently and with different mechanisms, that a category of commerce which had been operating outside the normal customs system for a decade would be brought inside it.
That decision was not really about China, and it was not really about revenue. It was about the fact that a parcel-scale flow of goods had grown large enough that not inspecting it had become a policy in itself. Once a state notices that, it does not un-notice it.
"Tariffs are a negotiating position. Customs procedure is an infrastructure decision. Brands that read the last two years as the first thing spent eighteen months waiting; brands that read it as the second thing spent eighteen months rebuilding."
The rebuild is not exotic. Clean product data, a real break-even number, bundles instead of singles, duty collected at checkout, and inventory positioned where the entry happens once. None of it is clever. All of it is overdue.
Share:
Free · 24 pages
Read Chapter 1 of the Playbook
AI Foundation Models — how DeepSeek and Qwen crossed a billion downloads while Western boards were counting Mag7 capex, and which layers of the AI stack the rotation has not reached. No excerpt tricks: it is the full chapter.
You will also get the weekly newsletter. Unsubscribe any time — the link is in every email.
The Book · $99
China Market Entry Playbook 2026
456 pages across 22 industry chapters — AI, semiconductors, EVs, payments, logistics, beauty, F&B, healthcare, pets, toys. Every chapter runs the same five questions: the anchor number, the rotation underneath it, which layers of the global stack it has reached, and what to do about it. English and Simplified Chinese editions included.
Ex-Transsion Global Digital Marketing Director · Effie Awards Greater China jury member · Guest lecturer, Peking University School of Economics · Writes Annie Chan Talk, your insider lens on China.
