CBP Issues Interim Final Rules Indefinitely Suspending De Minimis Exemption

The Analysis

Yesterday this shipment was the easiest file on my desk. A carton worth under eight hundred dollars, handed to the post or to an integrator, moving on the strength of being small. Nobody asked who the importer of record was, because there was no entry to file and nothing to pay. That is over. On 23 June 2026 CBP issued two interim final rules indefinitely suspending the eight hundred dollar de minimis duty exemption for low-value imports arriving by all modes, including the international postal network. The rules took effect on 24 June. Low value no longer means no paperwork. A formal or informal entry is now required, and the duties and fees that apply are now payable. One narrow piece of it runs later: certain international mail shipments have a delayed compliance date of 22 October. That is the only runway anybody is going to get, and it is four months long.

I read rules the way I read everything else, by asking where the paper goes and who is holding it at each hour of the day. So I pulled one low-value file this morning and walked it through the new sequence, start to finish. What came out of that walk is not a legal opinion; I am not the person who gives those, and this trade has enough people speaking confidently about instruments they have not read. What came out is a map: where a document sits, when it can still be changed, how long the change takes once it cannot, and what each day of the delay is worth.

So the useful question for anyone moving low-value freight into the United States by any mode is not whether the exemption is gone. It is gone. The useful question is whether your data package is complete the day before the vessel, the aircraft or the mail bag gets there, because after that point you are no longer fixing data. You are filing corrections against a released entry, and those are two different jobs with two different price tags.

Assume this one is sixty cartons, declared at six hundred and forty dollars, moving into the United States through the international post. Under the exemption that file was finished before it started: nothing to classify, nothing to value, nobody to name as importer. From 24 June it is an entry, formal or informal depending on what the goods are and how they arrive, and the duties and fees that apply are payable on it.

Here is the custody chain on a file like this, hour by hour. The seller raises a commercial invoice at dispatch. If there is a marketplace in the middle, the marketplace raises its own version, usually with its own product description and its own idea of what the value is. The carrier builds a manifest line from whatever it received, which in most low-value flows is the marketplace version and not the seller's. The broker engaged by the importer receives the manifest. And the importer of record, the party the entry is actually filed against, is in plenty of operations nobody in particular, because nobody ever had to decide. On 23 June that stops being an untidy habit and becomes a hole in the file. When the classification turns out to be wrong nine weeks later, the only things that show who decided it are the email trail and the version history on the invoice.

Three chokepoints decide whether a file like this moves or sits. The tariff number: in the version I walked, the classification column was empty, because the seller had been writing plain-language product descriptions for two years and nobody had pushed back. The origin: the carton marking and the invoice said different things, which is the kind of mismatch that costs days rather than minutes. The value: the number on the invoice was the price after a promotional discount, with no note anywhere about whether freight and insurance sat inside it or on top of it. If the papers do not match, nothing downstream does. I know that sounds like a scold, but it is literally what happens. One mismatch between the invoice and the carton is a request for information, and a request for information against a released entry is a different job from a correction on an unreleased one.

I work in a fixed order: classification, then value, then any relief that might apply. Classification comes first because it sets the duty rate, it decides whether a relief or exemption is even available for that line, and it decides whether another agency has an interest in the goods at all. Working backwards from the money is how people end up with a tariff number in the entry that nobody can defend when it is questioned. I would rather spend forty minutes getting one line right than spend three weeks explaining four hundred lines that were filed on a guess.

Valuation deserves its own turn, because it is where low-value files quietly go wrong. The entry needs a customs value, and the number on a marketplace invoice is often not that number. It may be net of a promotion, it may exclude freight and insurance, it may be in the wrong currency, and it may be a price that only exists because the platform subsidised it. Which of those is the right basis for your goods depends on the transaction, and it is exactly the kind of question I would rather see answered by somebody reading the regulation than by me guessing. What I will insist on is that the invoice states what the number is and what is inside it, so that whoever files the entry is not inventing a basis on the fly.

Then comes the part that actually decides the outcome, which is the clock. Assume dispatch and arrival are thirty-one hours apart. Assume the seller's office is seven hours ahead of the broker's. That leaves a working window of less than a day in which anything on the invoice can still be corrected before the entry is filed, and most of that window falls while one of the two offices is shut. Every correction carries a price in hours, and the price steps up the moment the entry is filed. Before filing, a wrong origin statement is a corrected invoice and an email. After filing and before release, it is a request to amend, and the amendment has to be processed by whoever is holding the entry. After release, it is a post-entry correction, which means a separate submission, a separate review, and a separate line in a record that somebody will read two years from now.

How long the middle one takes is where I have to be honest and say that it depends on the port. I am not going to hand you a number of days as though every location runs the same, because they do not. The same correction on the same commodity can clear in an afternoon at one place and sit for most of a week at another, and the variables are the entry type, whether the shipment has been selected for any kind of examination, and how that office is handling volume that week. What I can give you is the shape of it. A pre-filing correction is measured in hours. A post-filing, pre-release correction is measured in days. A post-release correction is measured in weeks and leaves a permanent mark on the file. Plan against the shape, not against a number.

Examination changes the arithmetic again. If a low-value shipment gets pulled, the clock stops being about documents and starts being about physical access to the cartons: who can open them, where, and how fast the goods can be presented. On freight that used to walk through under the exemption, nobody had built a process for that, because there was nothing to pull. Now the same freight can be held, and the hold runs on the facility's calendar rather than yours. Assume two and a half days as a planning figure, and assume you have to produce the invoice, the packing list and a coherent description of the goods within the first few hours of that hold to have any chance of shortening it.

Let me put the whole time account on one line so it can be argued with. Assume half a day to assemble a complete data package before dispatch, a quarter of a day for the broker to review it and come back with questions, half a day for the seller to answer if the answer needs a human, the filing itself, and then either release, or a hold of two and a half days, or a two-day correction cycle if something is wrong at filing. Add it up and the gap between a file that was complete at dispatch and a file that was completed after arrival is somewhere around three to five calendar days, on freight that used to clear without anybody touching it. Three to five days on one carton is annoying. Three to five days on forty cartons a day is an operations problem.

So let me do the arithmetic, and I will write the assumptions out so you can argue with them. Assume an operation moving forty low-value shipments a day into the United States, twenty-two working days a month. Assume an average declared value of two hundred and forty dollars. Assume the combined duty and fee load on those lines works out to fifteen per cent of value under the classifications as filed; that is an assumption about a product mix, not a rate anybody should copy. Forty times two hundred and forty times fifteen per cent is one thousand four hundred and forty dollars a day, and across twenty-two days that is thirty-one thousand six hundred and eighty dollars a month of duty and fees that did not exist in the budget before 24 June.

That is the number everybody stares at, and it is not the number that will hurt. Assume that eight per cent of those shipments get held because the invoice is missing an origin statement or the tariff number is blank. Forty times eight per cent is three point two shipments a day. Assume each hold costs two and a half extra days and each day of storage and handling on a held low-value shipment costs forty-five dollars. Three point two times two and a half times forty-five is about three hundred and sixty dollars a day, or roughly seven thousand nine hundred dollars across a month. Now assume the fix: one person pre-clearing the data package before dispatch, six minutes a shipment, forty shipments, four hours a day, at a loaded cost of twenty-eight dollars an hour.

That is a hundred and twelve dollars a day, about two thousand four hundred and sixty dollars a month, set against the seven thousand nine hundred it removes. On these assumptions the pre-clearance pays for itself a little over three times over. Change any of the assumptions, the value, the hold rate, the number of days, and the ratio moves with them, which is exactly why I have written them down instead of quoting a clean multiple.

One more thing belongs in the file and usually is not there: who pays. An entry now carries duties and fees, somebody has to be the party paying them, and somebody has to be the party that receives the bill. In a lot of low-value operations the seller quotes a delivered price and has never had to think about what happens when a duty is assessed against the buyer. That is a commercial conversation rather than a customs one, but it produces a document, an agreement about who is the importer of record and who bears the duty, and that document is precisely what the file will be missing when a dispute arrives. Get it in writing per lane rather than per dispute.

The work between now and 22 October is a data project, not a customs project. Certain international mail shipments have until that date, which means the postal side of a low-value business gets roughly four months that the rest of the modes did not get. I would not spend those four months waiting for further clarity. I would spend them running postal shipments through the same entry process the non-postal modes went through on 24 June, on a small volume at first, so that when the delayed date arrives the question is not whether the process works but whether it scales.

I should say plainly that this is one file, walked on one morning, and I am not going to claim it is everybody's situation. Low-value flows differ enormously by product, by marketplace, by whether the seller is also the shipper, and by whether anybody in the chain has ever filed an entry before. What transfers is not the numbers. What transfers is the order of operations: decide who the importer is, get the classification right, make the invoice carry what the entry needs, and do all of it before the goods move.

Records are the part people skip and then need. On every file I open now I want four things kept: the version of the invoice that was used to file, with the date it was received; a short classification worksheet saying what the tariff number is, how it was arrived at, and who decided it; every correction request and its answer, in writing; and a note of who was the importer of record on that entry. None of that is glamorous. All of it is the difference between an enquiry you answer in an afternoon and one that takes a month. I also want the amendment trail kept with the entry rather than in somebody's inbox, because the person who needs it in two years will not be the person who filed it.

The legal reading of the interim final rule, what it says, what it excepts, and how the 22 October delayed compliance date for certain international mail shipments is actually framed, is not mine to give. Dr. Ingrid Voss has been through the text. What I can tell you is what the paper has to look like on Monday morning and who has to be holding it.

There are a few routes people will try around this, and most of them move the problem rather than solve it. Consolidating several low-value shipments into one entry is the obvious one: fewer entries, and any fixed cost per entry gets spread thinner. It also means one bad line can hold the whole entry, and it means the origin and value statements on the consolidated invoice have to cover goods that may not share a classification. Filing separately keeps the blast radius small and multiplies the per-entry cost. Which way that trade falls depends on how homogeneous the product mix is and on whether the delay risk or the cost risk is the one that actually hurts you.

Changing mode is the second route. Moving low-value freight off the post and onto an integrator, or consolidating it into ocean freight under a single importer of record, buys a cleaner document chain and a party who files entries every day. It also buys a different cost structure, and on ocean a transit time that makes the document lead time less forgiving rather than more. Changing the sourcing country changes the origin statement and therefore the classification work, and it usually means starting the supplier conversation from zero. Holding inventory inside the United States, a domestic buffer fed by properly entered bulk shipments, converts thousands of small entries into a smaller number of large ones, which is attractive on paper and expensive in working capital and storage.

The traps I would watch on each of those. Free time: whatever the facility allows before storage starts charging runs from a moment defined in the facility's own terms, not from when you noticed the hold, so the clock usually started before you knew there was one. Per-entry versus per-line charging: some costs scale with the number of entries and some with the number of lines, and consolidating changes which of those you are exposed to. The billing basis for any accessorial: read whether it is charged on declared value, on weight, or per entry, because a fifteen per cent duty load and a per-entry fee respond to consolidation in opposite directions.

Transit-time promises: a carrier's published transit time almost never includes customs hold, and a two-day service with a two-and-a-half-day hold is a four-and-a-half-day service. And the one I see most often, assuming the party that booked the freight is the party filing the entry. Ask who the importer of record is before the goods move, in writing, or find out during the hold.

I keep coming back to the same thing on files like this. The rule changed on a date, and for most modes that date has passed. What decides whether this lands as a thirty-one-thousand-dollar-a-month duty line or a forty-thousand-dollar-a-month mess is not the rate. It is whether the invoice that reaches the broker the day before arrival carries a tariff number, an origin that matches the carton, and a value somebody can explain. That is a document problem, and document problems are solvable in days if you start before the goods move and in weeks if you start after. This one can still be saved, but not after this afternoon, which is the version I say out loud on the phone, and it is only half a joke.

↑ Back to the key points

— By Derek Xu