The €150 exemption is gone. Per-line accuracy is now the whole game.

As of today the EU's €150 customs duty exemption for low-value consignments is gone, replaced by a per-declaration-line duty charge.

If you move parcels into the EU, the arithmetic of your operation changed overnight and your process almost certainly didn't.

What actually changed for operations

Under the exemption, a €40 consignment carried no customs duty. Which meant that for an enormous share of e-commerce volume, the commodity code was a formality. Right code, wrong code, same outcome: nothing owed.

A lot of operational practice grew up comfortably inside that. Generic descriptions passed through from marketplace listings. Default codes applied across whole product categories because the category was close enough. Classification treated as a mandatory field to populate rather than a determination to make.

None of that was reckless. It was a rational response to a rule where the answer genuinely didn't affect the money. Nobody spends review time on a field that can't be wrong in any way that costs anything.

Now it can be wrong. On every line.

Why volume makes this worse than it sounds

A freight forwarder filing a hundred entries a month with a five percent classification error rate has five problems a month. Unpleasant, findable by a person, fixable by hand.

A parcel consolidator filing tens of thousands of lines a week at the same five percent has a systematic exposure that no human review process can catch by inspection, that grows every single day, and that an auditor can reconstruct in full three years later from records the consolidator is legally required to keep.

The error rate didn't change. The exemption was absorbing it. Remove the absorber and an unchanged process produces a materially different risk profile.

There's a second-order effect worth noticing. Under the exemption, nobody was measuring classification accuracy at parcel volume, because there was nothing to measure it against and no consequence to calibrate it with. Which means most operations don't currently know what their error rate is. The number I used above is illustrative. Yours is unknown, and finding out is now a priority rather than an academic exercise.

The bit that catches people out

Duty exposure is the obvious problem. It isn't the expensive one.

The expensive one is that authorities look backwards. An error pattern that's systematic — the same wrong code applied to the same product category, tens of thousands of times, because it was a default — is exactly what an audit finds. And it doesn't find it once. It finds it across the whole retention period, applies it to every affected line, and adds interest.

A one-off mistake is a correction. A default that was wrong for two years is an assessment.

Which is why "we've always used that code" is the most dangerous sentence in a customs operation. It converts an individual error into a pattern, and patterns are what get assessed.

The asymmetry is worth stating plainly: being wrong in the customer's favour on one parcel is trivial. Being wrong the same way on four hundred thousand parcels is a balance sheet event, and the mechanism that produced it was a decision somebody made once, quickly, years ago.

What to do this month

Start by finding your defaults. Every operation has codes applied by rule rather than by determination — a category mapping, a fallback for missing descriptions, a per-client convention somebody set up during onboarding. List them, with volumes. That list is your systematic exposure and it's where an auditor will start, because it's where the volume is.

Then sort by volume multiplied by duty differential. You can't review everything and shouldn't try. You can review the ten codes where being wrong costs the most per line, weighted by how often you apply them. In every operation I've seen, that list is short and the concentration is severe — a handful of codes account for most of the risk.

Fix descriptions before you fix codes. Most classification errors at parcel volume aren't classification failures, they're input failures. "Spare parts" cannot be classified correctly by anyone, human or machine, and neither can "gift" or "sample" or "personal effects." If the description reaching your declaration is a category label rather than a description of goods, no amount of downstream cleverness rescues it. That's a conversation with the shipper or the marketplace, not a software problem, and it's the highest-leverage thing on this list.

And start capturing the reasoning now. Not just which code, but why — which heading, which note, which rule, which prior decision. Doing that from today is nearly free if the system has a field for it. Reconstructing it in two years, for a period nobody remembers and staff who've moved on, is not.

The wider point

This is the second structural change to low-value cross-border trade inside eighteen months, alongside the tariff turbulence on the US side.

The direction is consistent across markets even though the mechanisms differ. Authorities noticed that an exemption designed for occasional personal imports had become the default channel for a very large share of retail trade, and they're closing it. The EU has done it by removing the threshold. Others are doing it by rate changes, by reporting obligations, by pushing liability onto platforms.

If your classification process was designed for a world where small parcels were fiscally invisible, it was designed for a world that stopped existing this morning.