CBP Move on Small Parcels: All Entry Type 11 Under $2,500 Must Be E-Filed by Import Date

The Analysis

A nine-kilo carton of accessories leaving Shenzhen on a Tuesday night used to be cleared into Los Angeles on nothing more than a commercial invoice and an informal entry filed a day or two after the wheels touched down. That particular habit is what CBP is coming for. The proposed rule published on 8 October reaches every informal entry of goods valued at 2,500 dollars or less, and it moves Entry Type 11 off the old rhythm of file it once it has landed and onto an electronic filing that has to exist on or before the date of importation. There is a new Entry Type 13 built for mail, the Universal Postal Union S-10 tracking number has to ride on the inward cargo manifest, and a carrier that signs as consignee without owning the goods has to put a licensed broker in place as importer of record.

Comments close on 7 December. None of this is law yet. But if you sell into the States below 2,500 dollars a shipment, the real question is not whether you file earlier. It is whether your data can be ready earlier, because that is a different problem and it belongs to a different party.

Begin with what the document actually says, because the headline version is useless to anyone who has to file something. The proposal touches four separate things. Every informal entry valued at 2,500 dollars or less is in scope, not just the consumer parcels everyone keeps using as the example. Entry Type 11 has to go in electronically by the import date. A new Entry Type 13 is carved out for mail, which CBP has wanted for years because mail has never fitted the consumption-entry codes cleanly. And the inward cargo manifest has to carry the S-10 tracking number, which CBP priced at 18.7 million dollars for carriers in 2027 and more than 100 million dollars from 2027 through 2034. Fifty-five days of comment period closes on 7 December.

That 18.7 million dollars for one identifier is the number that tells you where the work is. S-10 is not something you can invent at the airport. It is a Universal Postal Union barcode standard assigned by the origin post or by whoever prepares the mail at induction, printed on the item, and scanned along the postal chain. To put it on an inward cargo manifest, the identifier has to be captured at origin, survive every handover, and reach the US border in the same form as the one physically on the parcel. That is an origin problem wearing a US filing costume, and it explains why CBP's own cost curve puts the heavy lifting in 2027, the build year.

Why any of this is happening now is not hard to work out. De minimis treatment was suspended during 2025, so the enormous flow that used to walk in free under 800 dollars suddenly had to enter formally and pay. On today's map most of that flow lands in Entry Type 11, because it is valued under 2,500 dollars. Informal entry was designed decades ago for genuinely small movement: a sample, a returned part, a repair under warranty, the occasional odd lot. No continuous bond is customarily required, no merchandise processing fee attaches, and the data burden is light enough that a decent broker clears it off a single invoice. It tolerated sloppy inputs because it was never meant to be the front door. It is the front door now, and the entire point of pre-arrival data is to let the agency target risk before an aircraft lands. You cannot target risk against an entry that does not exist yet.

Run it along my usual map and you can see exactly where the stress sits. Classification, then declaration elements, then exam, then duty, then release. On an informal entry the middle is very short: there is usually no exam worth mentioning, duty is frequently electronic the same day, and release follows quickly. What does not move is that reasonable care for classification, valuation and origin sits with the importer of record, and that party has to be named in the system before the filing goes in. Every sharp edge in this proposal lands on that one sentence.

And now the part the news coverage skipped, which is the part that will actually decide whether your operation survives this. The draft says that a carrier acting as consignee but not owner has to appoint a licensed broker as importer of record. Read that slowly and think about who holds what. The party that physically holds the parcel, owns the manifest, can see the tracking number and can push a file at eleven at night is the carrier. The party the draft says is legally permitted to be importer of record is a licensed broker, who in plenty of models does not see the data until the goods are on the ground.

So the rule takes filing responsibility away from the party that has the timing and hands it to the party that does not have the data, and then tells both of them the file has to be in before arrival. If your low-value US volume runs on today's model, where the carrier or the platform files as consignee, it is that model being pulled apart here rather than the deadline itself.

Bring that down to one missing field. Right now, a missing ten-digit HTS or a blank country of origin on a low-value informal entry costs an email. The broker sends it to the seller, the seller answers the next morning, the entry goes in late and nobody notices, because late was allowed. Under this draft the same missing HTS costs days on a clock that starts before the aircraft is even down, and those days are billed. If the goods sit in an airport facility, the facility bills. If they sit in a bonded warehouse, the warehouse bills and the answer does not arrive faster because somebody is paying. Who carries that cost is decided by contract, not by regulation, and for a lot of shippers the first unpleasant surprise will be discovering that nobody has ever written down whose fault a late data element actually is.

Who eats it, in rough order of how badly. The most exposed are the flows where the importer of record today is a carrier or a platform rather than the buyer, which is precisely the shape of most direct-to-consumer and marketplace volume. Next come the non-sale movements nobody ever designed a data source for: warranty replacements, repairs, samples, returns, goods sold on consignment where there is no transaction value to declare and you need a different valuation basis agreed before you can file at all. Then come the small B2B sellers sitting between 800 and 2,500 dollars who were quietly enjoying informal entry as a cheap compromise. Finally come brokers themselves, who pick up volume and earned liability and pre-arrival labour in the same envelope, and who will reprice accordingly.

Timing is the piece people misjudge, because a proposal is not a compliance date. Published 8 October, comments close 7 December, roughly sixty days for the industry to put its arguments in writing. After that nobody can promise you a date, but CBP's own economics give the game away: if the agency is pricing carrier system work into 2027, agency staff are not expecting this to be live this winter. Assume a final rule somewhere in 2027 with a delayed effective date, and assume the manifest requirement is the long pole. Your own software you can fix in a quarter. An origin post in another jurisdiction that has to print and scan S-10 at induction for every service level is not yours to fix at all. Work backwards from the manifest rather than from the entry.

Here is the arithmetic, and I will write the assumptions out so you can argue with them. Assume you send 400 shipments a month into the US valued at 2,500 dollars or less, average declared value 600 dollars, average nine kilos. Assume 12 percent arrive today with at least one declaration element missing, which is 48 shipments a month, about two a working day. Assume clearing that element takes one and a half calendar days and today costs you nothing beyond correspondence. Assume that once filing has to precede importation those 48 shipments wait in a facility at 45 dollars a day each after a 48-hour free window, plus 25 dollars for a second presentation.

Forty-eight times one and a half days times 45 dollars, plus 48 times 25 dollars, is 4,440 dollars a month, roughly 53,000 dollars a year. Now assume instead that the missing element belongs to a marketplace seller twelve time zones away and you need three days rather than one and a half: the same arithmetic gives 7,680 dollars a month and about 92,000 dollars a year. Neither number has ever appeared on a freight invoice. Both will show up in your accounts.

Put that against unit economics, because this is where low-value gets dangerous. Assume your contribution margin on a 600 dollar shipment is 18 percent, call it 108 dollars a unit. Assume shifting a low-value informal entry from post-arrival paperwork to pre-arrival electronic filing, with a licensed broker sitting in as importer of record, adds 25 dollars to the cost of handling it. That is 23 percent of the unit contribution before you have bought a single day of storage. Across 400 shipments a month it is 10,000 dollars a month and 120,000 dollars a year. Those are my assumptions, not CBP's numbers. CBP priced the carrier side at 18.7 million dollars in 2027 and more than 100 million dollars across 2027 to 2034, which averages above 12 million a year with 2027 heavier because it is the build year. That estimate covers carriers putting S-10 on manifests. Nobody has published a figure for the broker-appointment piece, and that is the piece most importers will receive as an invoice.

The 2,500 dollar line deserves its own paragraph, because it is a boundary rather than a round number. Below it you informal entry: no continuous bond in the usual case, no merchandise processing fee, thin data. Above it you have a formal consumption entry with a bond requirement and the ad valorem merchandise processing fee at 0.3464 percent, subject to a floor and ceiling that get reset each fiscal year. Take one consolidated carton declared at 2,400 dollars and assume a general duty rate of 5 percent: duty due is 120 dollars and that is about the whole story. Take the same carton at 2,600 dollars and you have crossed into a different regime entirely. Freight, assists and rounding have a habit of pushing declared values over a line like that, which is why classification and value belong at the booking stage now, not at six in the morning at the gateway.

So this is the week to find out your actual exposure rather than guess at it. Pull ninety days of entry data and count entries at or below 2,500 dollars, split by how they arrived: air express, inward mail, ocean consolidation, truck from Mexico. If you cannot split them by channel, fix that first, because the mail answer and the express answer are going to be different answers. Then take the HTS lines that cover 80 percent of that volume and ask one question of every declaration element: can this be known at booking time, or does somebody physically have to touch something to learn it? Everything in the second bucket is a future hold, and the honest count of those is the true size of this change for your business.

Then run the physical test, which costs you twenty shipments. Ten going through a postal channel and ten going through an integrator, and for each one ask whether the number on the label is the number that can be carried onto the inward cargo manifest. On the postal side it should already be there where the origin post issues S-10 at induction, and plenty of small operations do not issue anything usable for every service level. On the integrator side it will probably not be a Universal Postal Union number at all, and that gap is what you either map or argue about.

This is precisely why there is a comment period. Dr. Ingrid Voss puts it better than I can: the exceptions get written while the rule is still a proposal, and once it is final you simply live inside whatever somebody else managed to put in writing. Two specific things are worth arguing for. What identifier satisfies the manifest requirement when the operator is not a postal operator and issues tracking of its own. And whether any post-arrival window at all survives for the shipment whose data genuinely cannot exist before the aircraft lands.

Now the ways this could turn out differently, because a proposal is a proposal. If the final text accepts filing by import date with even a short post-arrival grace window, the cost story above shrinks and possibly disappears. If the S-10 requirement gets narrowed to postal traffic only, the integrators' definitional fight evaporates along with most of CBP's 18.7 million dollar estimate. If your volume is mostly above 2,500 dollars already, none of the Entry Type 11 mechanics touch you beyond manifest data for consolidated loads. And if the consignee on your entries genuinely is the owner, which is normal in ordinary B2B imports where the buyer is named, then the licensed-broker-appointment rule has nothing to bite on. It bites specifically on the arrangement where the carrier signs as consignee for goods it does not own, which is the standard shape of consumer-direct and marketplace flows. There is nothing to pay today. Plan against it, argue about it, and do not rebuild anything until you have read a final rule.

Three ways people will lose money here regardless of how the text lands. Shippers will assume the carrier handles it because the carrier handled it yesterday, without checking whether their own contract makes the carrier consignee rather than owner, which is exactly the pair of words the draft tests. Sellers will discover that they never learned their own HTS codes because for years nobody asked, and the correct answer will not arrive faster for being urgently required. And somebody will suggest splitting shipments so each piece stays under the threshold. Do not go near that. Splitting to stay under a line is not planning, and now that a tracking number ties every piece of a movement to the inward manifest, a pattern like that is going to be visible rather than buried in one broker's desk drawer.

These parcels are still going to arrive. All that changes is that they have to be described properly before they do, instead of after. Get the description early enough and the file goes in on time and none of the above is ever billed. Get it right late and you are buying days at whichever facility happens to be holding your goods. The 2,500 dollar threshold, the import-date filing clock and an S-10 on the manifest are three lines on one page, and that page is due before goods are. This one is salvageable, but the work has to start this week rather than on the day the final rule lands.

Before the checklist, one thing worth pricing separately, because it is the flow everybody forgets. Assume 6 percent of those 400 monthly shipments are not sales at all: warranty replacements, repairs, samples, returns, consignment stock. There is no transaction value to declare, so the entry needs a valuation basis agreed in advance, written down, and defendable if anybody ever asks. Assume the missing valuation basis costs two days rather than one and a half. Twelve shipments a month at two days and 45 dollars is 1,080 dollars a month, about 13,000 dollars a year, before any second presentation fee. That is not a large number next to the 53,000 dollars above, and it is precisely the kind of number that never gets a project attached to it because nobody owns that category internally. Somebody does need to own it, and under a rule that requires the file before arrival, "we will work it out when it lands" is not an answer you will be able to give.

And one structural alternative worth asking about now, even though I would not rely on it yet. If a meaningful part of your low-value volume arrives as an ocean or air consolidation rather than as individual pieces, ask your broker whether a weekly entry or admission into a foreign trade zone changes the count of entries you file. It changes the count; it almost never changes the data requirement, because the manifest still has to describe what is moving. So treat any such scheme as a way to compress filing labour rather than as a way out of the S-10 problem, and do not let a vendor sell you restructuring until you have seen a final rule.

↑ Back to the key points

— By Derek Xu