Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 31 July 2026, obtain the executing text and the annex listing the 60 partners, and register every rate from that text rather than from secondary reporting.
- By 7 August 2026, build a register of every origin and tariff line in the active book, with annual volume, and mark each pair as low band, high band, or outside the schedule.
- By 14 August 2026, pull the evidence file behind every exemption claim, targeting 100% of USMCA claims backed by a certification valid at the date of entry and every sector-tariff exclusion matched to the measure that governs it.
- By 21 August 2026, run a date-line test on all entries filed within 10 days either side of 24 July, listing goods whose entry date and arrival date fall on opposite sides of the line.
- By 4 September 2026, read the price-adjustment and change-of-law clauses on the 20 lines carrying the most duty, and record which party absorbs a rate reclassification.
The new 10%-12.5% US tariffs on 60 trading partners took effect at midnight on July 24, 2026, replacing the expiring 10% global tariff. Goods already under sector-specific steel and aluminum tariffs and those entering under USMCA are exempt, but logistics providers face higher compliance and documentation burdens.
The Analysis
At midnight on 24 July 2026, the new tariff schedule on 60 US trading partners took effect at rates between 10% and 12.5%, the same day the temporary 10% global tariff expired. Goods already covered by the sector-specific steel and aluminium tariffs sit outside it, and goods entering under USMCA are exempt. Coverage has treated that as one event with two exceptions. In the text it is a rate schedule, a date line, two exits of different character, and a classification step that did not exist the day before.
Strictly speaking, what changed at midnight was not the level of protection but the basis on which the level is set. A single global rate attaches to a shipment by default. A schedule of sixty rates attaches only after someone has established which partner the origin belongs to, and after someone has established that neither exit applies. That is a classification question before it is a cost question, and it is answered per line, not per shipment.
The documentation burden described as falling on logistics providers is not a side effect. It is the mechanism. A differentiated schedule cannot be collected without per-line origin evidence, and an exemption that is claimed rather than granted is only as durable as the file behind it.
The new rate applies to an entry when three conditions hold at once: the entry is made after the date line, the origin is one of the sixty, and neither exclusion reaches the goods. Take any one away and the new schedule is not the operative rule. That reads as obvious and is routinely lost, because the three are tested by three different people at three different times. The date line by whoever files. The origin by whoever holds the supplier declarations. The exclusions by whoever reads the scope of the measure. When those answers live in three systems, the entry filed is the one that passed the first test and assumed the other two.
The replacement matters more than the rate. Had the temporary tariff simply lapsed, midnight would have been a reduction, and the planning question would have been how to hold goods back or push them through. Because a replacement took effect at the same moment, there is no gap, no window, and no arbitrage on timing. Any plan built on a lapse-and-refile window rests on an assumption the facts do not support.
The two exits do not mean the same thing, and this is where internal reporting goes wrong. A steel or aluminium item outside the new schedule is outside it because a different measure governs it. Exemption from one layer is a statement about which regime applies, not a statement that the burden is gone, and the sector measure may well be the heavier of the two. A USMCA-qualifying good is a different case again: the preference is claimed, it rests on a certification the importer has to hold, and it can be withdrawn from a line that was treated as exempt last week. Reading both exits as not affected collapses two exposures into one row of a summary.
Size decides who absorbs the classification work, not who absorbs the duty. A small importer with a narrow range and mixed origins has the worst ratio of exposure to capability: few lines, several origins, and no origin file beyond what the supplier put on an invoice. In the absence of evidence the conservative treatment is the higher of the plausible readings, and that default is paid silently on every entry. A large importer faces the same legal question with better machinery and can usually answer it per line. Direction matters as much as size. An exporter in one of the sixty does not remit the duty, but the incidence is settled in the contract, and a contract silent on a rate change allocates it to whoever has less leverage. A domestic manufacturer buying imported components never sees a customs entry; it sees a component price move one tier back, with no line item to appeal.
Legal time starts at midnight on 24 July and it is the least useful clock to watch. The first operational signal arrives earlier, at the documentation cut-off before the first entry under the new schedule, because that is the moment somebody has to decide what origin to declare. The second arrives at the first entry filed after the line, where the origin code and the exemption basis are fixed for goods that may have been loaded weeks earlier. The commercial signal arrives on the next price-adjustment cycle, a quarter later for most supply contracts. The audit signal arrives last and is the only one that looks backwards. Importers watching only the first clock meet the other three in the wrong order.
What the coverage missed is that a range is not a spread. Ten to twelve and a half percent reads like a band around a midpoint and is nothing of the kind. It is a lookup: the rate is a function of origin, so the effective rate on a consignment is a weighted average of the rates on its lines, and it moves when the origin mix moves. A buyer who swaps one supplier for another at the same unit price can change landed cost without changing anything a procurement system tracks. The headline number describes no actual shipment. What the coverage also missed is the unit of work. Sixty is a count of jurisdictions. The operational unit is the pair of origin and tariff line, and a mid-sized importer's active book typically holds several hundred of those pairs. Sixty is the number in the headline; several hundred is the number of decisions.
A small example, with the assumptions mine and not the article's. A consignment valued at one million dollars, split into four blocks of two hundred and fifty thousand, from four of the sixty. Two blocks land at the low end and two at the high end, so the blended rate is 11.25%, a number that appears nowhere in the text and is the number that actually prices the consignment. Move one block from low to high and the duty on it rises by 6,250 dollars, which is 2.5% of a quarter of the value. Nothing physical moved, no route changed, and a cost line shifted because a documentary question got a different answer. Small per line, large per book, invisible in any total not built from lines.
Where this stops holding is worth stating plainly. If the executing text carries a product annex, or an exclusion broader than the summary suggests, the register changes and the blended arithmetic changes with it. If the origin of a line is contested, the question stops being a tariff question and becomes an origin question, answered by rules the rate schedule says nothing about. If one exit is narrowed later, the files that supported the exemption become the record of what was claimed, and a claim that cannot be re-supported is a different category of problem from an underpayment. None of those is remote, and each is testable with the text in hand this week.
↑ Back to the key points