.svg)
.svg){kind=icon}
The letter is two paragraphs long and it arrives on a Tuesday in 2029. It references an entry filed in August 2026, names one tariff line, and asks the importer to produce the records supporting the origin claim made on that entry. Thirty days.
Nothing about that letter is unusual. What is unusual is how many operations are now exposed to receiving one.
Official U.S. import data show that the share of import value from Mexico entering under a claimed USMCA preference rose from 44.8% in January 2025 to 88.7% in November 2025, with the sharpest move between June and July, when the figure went from 47.6% to 86.1%. Over the same stretch, trade volume did not slow. Mexican goods exports to the United States reached 50.692 billion dollars in April 2026, up 21.1% year over year and the first month above 50 billion, according to the U.S. Census Bureau.
Read those two numbers together and the story is not that trade grew. It is that in roughly twelve months, close to half the corridor moved from not claiming preference to claiming it. Not claiming was never the same as not qualifying: a good carrying a zero most-favored-nation rate had no reason to claim anything, and many never did. What the surge represents, then, is a large population of operations that started making origin claims without ever having built the file behind one. Every one of those claims is a representation that the goods qualify, made by someone who has to be able to prove it for five years.
What doubled was not just the benefit
A preference claim is not a discount applied by the customs authority. It is a statement made by the importer, exporter or producer, and it carries an evidentiary obligation that outlives the shipment by years.
Under Article 5.2 of the T-MEC, a claim rests on a certification of origin. There is no prescribed form. The certification simply has to contain the nine minimum data elements set out in Annex 5-A, describe the good in enough detail to identify it, and it can sit on an invoice or on any other document. That design was deliberate and it was a genuine improvement over the rigid form the agreement replaced.
It also means the claim is easy to make and hard to unmake. A blanket certification covering twelve months of shipments can be issued in an afternoon by a supplier who has never run a regional value content calculation. The entry clears. The duty saving is real and immediate. The question of whether the good actually qualified is deferred, and it stays deferred until somebody asks.
The five-year tail
Article 5.8 sets the retention period. An importer claiming preferential treatment keeps the records supporting that claim for five years from importation. An exporter or producer who completed a certification keeps its records for five years from the date the certification was completed. On the U.S. side those obligations sit alongside the general recordkeeping rules in 19 CFR Part 163, and records generally have to be produced within 30 days of a request, or sooner if the request says so.
The records in question are not the certification. The certification is the conclusion. What has to survive five years is the reasoning behind it: the bill of materials as it stood at the time, the classification of each input, the supplier declarations, the cost data behind the regional value content calculation, and the evidence supporting whichever origin criterion was cited.
Quick check: pick one entry from the last quarter where you claimed preference. Ask your team to produce the full supporting file, not the certification, in thirty days. Whatever comes back is what an operation actually has.
Article 5.9 sets out how the authority tests the claim. It can send a written request for information to the importer, exporter or producer, issue a questionnaire, or conduct a verification visit at the premises of the exporter or producer in the other country. For textile and apparel goods, Article 6.6 provides a separate and more aggressive site-visit regime. When a verification concludes that the good did not qualify, preference is denied, and the denial reaches back to the entries covered by the claim. The duty that was never paid becomes payable, with interest, and the conduct behind the claim determines whether anything else follows.
Why nearly universal adoption changes the calculus
For most of the agreement's life, a company that qualified its goods and documented them properly held a real commercial edge, because a large share of the corridor did not bother. At a 44.8% claim rate, qualification was a differentiator.
At 88.7%, it is the floor. When nearly nine tenths of the import value crossing the border enters under a claimed preference, claiming does not distinguish an operation from its competitors. It only keeps it level with them. The distinguishing variable moved one step down the stack, from whether you claim to whether your claim survives being examined.
There is a second-order effect that follows arithmetically. Verification resources concentrate where claims concentrate. A doubling of the claim population is a doubling of the population available to verify, arriving at exactly the moment when preferential entry became the dominant path rather than the exception. Add the annual review cycle the agreement entered on July 1, 2026, which puts rules of origin on a negotiating table every twelve months, and the direction of travel on origin scrutiny is not ambiguous.
Across more than 190,000 customs operations a year at 39 ports, the pattern we see on the corridor is consistent and it is not about intent. The claims that fail verification are rarely fraudulent. They are claims that were correct when made and could not be reconstructed three years later, because the supplier changed, the bill of materials changed, the person who ran the calculation left, and nobody wrote down why the good qualified in the first place.
Three moves that build the provable claim
Separate the certification from the qualification file, and treat them as different artifacts. Most operations hold certifications. Far fewer hold the analysis that justifies each one, indexed to the entries it covers. The certification is what you present. The qualification file is what you defend. Build the second one deliberately, per product and per certification period, and store it where a five-year-old entry can be located from an entry number.
Requalify on every change that touches the calculation. A supplier substitution, an input reclassification, a cost movement large enough to shift a regional value content margin, a new plant. Each is a trigger to rerun the analysis and, where the result changes, to reissue or withdraw the certification for the affected period. A blanket certification that covers a year during which the bill of materials moved twice is a claim with three different truths inside it.
Audit the thin margins first. Not every claim carries the same risk. A good that clears its threshold by twenty points survives a recalculation. A good that clears by one point does not survive a supplier price change, let alone a rule adjustment out of an annual review. Rank your preference-claiming volume by how much headroom each product has, and spend the review effort at the bottom of that list.
The claim takes an afternoon. The proof has to last five years.
The doubling of preference claims across the corridor was a rational response to a tariff environment that made non-qualification expensive. It was also, for a large number of operations, a decision made at the speed of a purchase order and documented at the speed of a purchase order.
The tariff environment has changed twice since, and it will change again. The obligation attached to every claim made in 2025 and 2026 has not changed at all, and it does not expire until 2030 and 2031.
The advantage in this corridor is no longer held by the companies that qualify. It is held by the companies that can still explain, years later, exactly why they did.
Talk to a Joffroy expert about an origin verification readiness review across your preference-claiming volume, on both sides of the border.
TRADE. UNDER CONTROL.
