EU ends €150 de minimis exemption, reshaping cross-border e-commerce logistics

The Analysis

Start with one parcel, because the rule that changed prices the parcel, not the shipment. A carton leaves a consolidation warehouse in Shenzhen on a Monday night with twelve hundred pieces inside, and every piece is somebody's order, with its own buyer, its own invoice line and its own declared value. Before July 1 that carton moved on the €150 duty exemption and the file behind it was thin: a manifest, a commercial invoice with a total on it, one tracking number per piece. From July 1 the European Union has removed that exemption and put a €3 fee per item in its place, with an expected €2 handling charge from November. The file has to describe items now. That is the whole change, and it lands on paper.

What I have been doing since the rule took effect is unfashionable work. I take one file at a time and follow it through our own chain, writing down at each desk what can still be changed, how long a change takes, and who owns the record afterwards. Not the policy, the paperwork. And I want to be clear about my lane: how this applies to a particular consignment, and where the exceptions sit, is not a call I make. That is the kind of boundary Dr. Ingrid Voss sets out properly. My question is narrower and, for the next six weeks, more useful: which field has to be right, by what time, and who can still touch it when it is wrong.

Here is what the first week taught me. The €3 is the number everyone budgets for and it is not the number that hurts. The number that hurts is rework. A file that goes out with a vague goods description and comes back from the broker with a query costs three days and two people, and by the time that is sorted the November €2 is close and nobody has re-priced anything. Get the documents right in July and August and the €5 a piece you will face from November is a line in a spreadsheet. Leave them and it arrives as a surprise with a handling charge stapled to it.

Let me map the file before I talk about money, because the fee is charged per item while the item only exists on paper in four places. In a China-to-EU small-parcel flow I meet the same four holders every time, and each one has a different window for change. The listing and order system holds the goods description, the SKU, the material, the country of origin and the unit price. The consolidation warehouse holds the packing list, the piece count and the weight. The carrier holds the waybill and the manifest it transmits. The broker or clearance agent holds the declaration itself. The further right you go, the more a correction costs, and that cost is not a published fee, it is delay and somebody's afternoon.

At the left end there is room. A goods description can be fixed right up to the moment the pick list is released, which in practice means the evening before the carton is built. The failure here is not a typo, it is a habit carried over from the marketplace listing. Descriptions written to sell are not descriptions written to clear: a phone accessory listed as a gift idea, a textile listed by colour instead of fibre content, a set listed as one article when the box contains four things sitting under four different tariff lines. While the value threshold decided most questions, those habits were cosmetic. Under a per-item charge with the duty exemption gone, they are the reason a file comes back.

The warehouse is where the piece count turns physical. You can reopen a carton before it is manifested. After that you are asking people to cut tape, re-weigh, re-photograph and re-label, and each of those steps is somebody standing still for a quarter of an hour. I treat the generation of the manifest as a hard gate: before it, corrections are free; after it, a correction is a favour. Get the cut-off time from your consolidation partner in writing and put it in the calendar, because that is the real deadline in this process. Everything downstream is downstream of it.

The waybill comes next. Once the carrier has the data, a change to the description or the piece count is a data amendment, and amendments on consolidated parcels are handled in batches rather than one at a time. My honest answer on timing is one to two working days if you catch it before the flight, and an unpredictable number of days if you catch it after. This is where a sentence I repeat too often comes from: that shipment can be saved, but it has to be before this afternoon. The afternoon is not magic. The flight cut-off is, and once the flight has gone the file is in another country and you are asking a foreign desk for a favour.

Then the declaration, and here I have to watch my own words: this is where it depends on the port of entry. Two ports, one file, two different questions. One wants a more granular description, another wants proof of value, a third will not release without the consignee's identification number. I will not tell you which port asks for what, because that moves and because anyone who tells you a port behaves the same way all year is selling you something. The operational consequence is what matters: build the file to the strictest version you have personally met, because building to the loosest version means rebuilding it at the desk of whoever is strictest that week.

There is an order of work I use and I push it on anyone who asks. Look at the tariff code first, then look at relief and exemption. Not the other way round. While the €150 threshold stood, a lot of small-parcel work was done on the assumption that value settled the question and the code was an afterthought nobody checked. Now the code sets the duty rate sitting under the fee, and the origin decides whether any preference applies. Get the code wrong and every number beneath it is wrong, and the fix is no longer a data correction, it is a re-declaration with a paper trail. One hour on the code before the carton closes saves three days after it lands.

Now the part that catches everyone: the fee is per item, the declaration is per line, and those are not the same unit. A consolidated carton of twelve hundred pieces can be declared as twelve hundred lines, as one line for the whole carton, or somewhere between. Declare it as one line and the file cannot be reconciled against a per-item charge, because nobody can say which piece carried which fee. Declare it as twelve hundred lines and you have created a data-entry job nobody budgeted and a bill for lines nobody priced. The middle answer depends on how uniform the carton is. One SKU across twelve hundred pieces? One line, quantity twelve hundred, a clear per-unit value, defensible and cheap. Forty SKUs mixed together? Forty lines, and you need to know that before the carton is built rather than after.

There is one document rule I enforce, and it is the one people find pedantic right up to the week it saves them. The commercial invoice, the manifest and the declaration have to tell the same story in the same words: same item count, same per-item value, same description, same origin. When they do not, the file rarely fails at the desk. It fails later, as a query you cannot answer quickly, because two of your own documents disagree and you have to work out which of them was the honest one. Paper that does not match is where the trouble actually starts, and it is not the fee that catches you, it is the week you spend proving which of your three numbers to believe. I check the three against each other at day minus one, on paper, five minutes, every single consignment.

That pushes a decision upstream that most teams have never made out loud: how do we build cartons. Build by destination and you get mixed contents and long declarations. Build by SKU and you get short declarations and more cartons, which is more freight and more handling at the far end. I am not going to pretend there is a free answer. What I will say is that a per-item fee changes the arithmetic of that trade-off, and the arithmetic should be redone in the next few weeks using the November number, not the July one.

Here is the time account for one consignment, the way I actually track it. Day minus two, the order file locks and the data freezes; if a description is not right by then it will not be right. Day minus one, a pre-check runs on four fields only: description, tariff code, declared value, country of origin. Four, because a longer list does not get done on a Friday. Day zero, the carton goes out and the manifest is generated; that is the gate. A few hours later is the flight cut-off, and that is the last moment a change is cheap. Day one to two, the file is presented at entry. Clean, and release follows inside a day or two. Pulled for examination, add two to five days, and the number depends on the port. Then one to three days to the door. A clean file is about a week door to door. An untidy one is two, with storage running while it sits.

Numbers, with every assumption on the table. Assume one consolidated consignment of twelve hundred pieces a week. At €3 per item from July that is €3,600 a week, roughly €15,600 a month. Assume the €2 handling charge lands in November as expected: the same volume at €5 a piece is €6,000 a week, roughly €26,000 a month, a step up of about €10,400 a month on that one lane. Now assume a quarter of consignments carry descriptions weak enough to draw an examination, and each examination costs three extra days. Assume the stock in that pipeline is worth €120,000 to you and your cost of capital is eight per cent a year.

Three extra days across a quarter of the flow is a weighted delay of well under a day, and the carrying cost of that is around €200 a month, which is small next to the fee. Then add rework: half a person-day per examined consignment to answer questions, re-issue documents and chase the broker, at whatever your internal day costs. One consignment a week means two person-days a month. The fee is what you can see. The rework is what quietly eats the schedule.

So what gets done, by whom, and by when. Before the end of July, the goods master has to be complete on the four fields, and complete has a threshold: no SKU without a tariff code and a country of origin enters a consolidation list, and there is no override granted over the phone. During August, the descriptions get rewritten as material, function and article type, starting with the two hundred SKUs that carry the most volume, and the words gift, sample, accessory and set stop being acceptable on their own. From September, the four-field pre-check runs on every EU-bound consignment at day minus one, with twenty consignments a week pulled for a full file review. Before the fifteenth of October, every EU lane is re-priced from €3 to €5 per item and the storefront prices and shipping fees are refreshed on that basis. Owners: master data sits with whoever maintains the catalogue, the pre-check sits with the shipping desk, the re-pricing sits with whoever owns margin.

Alternatives, and what they cost in paper. Moving stock into a European warehouse and clearing as a conventional import changes the document set completely. You go from a parcel manifest to a formal import declaration. You need an economic operator identification for the importer of record. You take on the local tax handling at the destination. You move the cash cycle from days to weeks, because the stock is sitting in Europe before it sells. In exchange, the parcel-level per-item charge stops being the unit that decides the inbound leg, and the delivery promise you make to the buyer gets shorter. Whether that trade is worth it depends on your volume and on how predictable your demand is, and it is a decision you make with a spreadsheet rather than a feeling.

Changing channel is the other option people reach for, and the trap is assuming a description that passes on one channel passes on another. Postal clearance and express clearance ask for different data at different granularity, and a template written for one will fail on the other. Ports behave the same way: change the port of entry and you change the questions you get asked, and the tuition fee for learning that is a fortnight of stuck parcels. If you are going to change, change one lane, run it for four weeks, and measure the examination rate before you move anything else.

The trap I most want to wave a flag at is splitting. When a threshold disappears, somebody in the room will suggest splitting orders so each one lands under some number, or trimming declared value to soften the duty under the fee. Do not. I am not making a legal argument, I am telling you what happens to the file. Split orders create a reconciliation problem you cannot clean up later, because the pieces stop matching the orders, and an understated value is the fastest way I know to turn one examination into a standing suspicion on every file you present afterwards. What you declared stays on the record, and it is the first thing anyone reads when they do not like something.

Three smaller traps. One, the value basis. Put one sentence in the procedure note saying whether the declared value includes freight and insurance, and hold everyone to it, because the quickest route to an inconsistent file is two people filling the same field on two different bases. Two, returns. A parcel that comes back either re-enters or is resold locally, and both need documents you have probably not drafted yet. Three, free time. Storage during an examination is billed by the day and the clock does not stop while you wait for an answer from another time zone. Find out what the free time is at the port you actually use and write it next to the examination line in your tracker.

On responsibility I keep it to one name per lane. Not a team, not a shared mailbox. One person who can say, for a given consignment, that the four fields were checked, and who approved the tariff code on it. When a question comes back from a port eight weeks later, a shared mailbox produces four people each assuming somebody else has answered. A named owner produces an answer the same afternoon. The same rule applies to amendments: whoever asks for one records the time the request went out, and whoever gets the answer records the time it came back. Those two timestamps are how you later show whether a delay was yours or somebody else's, and they cost ten seconds each.

On records I keep it boring. One folder per consignment: the manifest, the commercial invoice showing per-item value, the tariff codes used and who approved them, any amendment request with the time it was sent and the time it was answered, the examination notice if there was one, and the release confirmation. Retention is whatever your own compliance function tells you and I am not going to invent a number. The point of the folder is not audit theatre. It is that the third time the same question comes up you can answer it in two minutes from your own file, and when a fee is queried you can show which pieces it attached to.

One more thing, because the reporting on this change said something I keep coming back to. The removal of the threshold is already redirecting China-origin small-parcel flows, and it is cooling one of the big growth engines in air cargo. On the document side that means the lane you have been running for two years may not be the lane you are running in six months. Different channel, different gateway, different cut-off, different questions at a desk you have never dealt with. Every one of those changes resets the learning curve I described above, and the learning is paid for in held parcels. Build the file to a standard that travels, not to a standard that works on one lane.

None of this is clever. The European Union took away a threshold that let a lot of parcels travel on thin paper, put a per-item charge where the threshold used to be, and signalled another one for November. The response is not a strategy deck. It is four fields that have to be right before a carton closes, a gate you respect, a tariff code you check before you check anything else, and a cost model that already contains the November number. Do that in July and August and the rest of the year is arithmetic. Leave it and you will spend the autumn explaining to buyers why their parcels are sitting somewhere, which is a conversation I have had more often than I want to have again.

↑ Back to the key points

— By Derek Xu