Import duty is not a fixed cost of doing business. It follows from how the product is classified, where it originates and what it is valued at — and the first two follow from decisions a brand makes during design and sourcing. Tariff engineering is making those decisions deliberately instead of discovering their cost at the first import.
What tariff engineering is, and what it is not
- Tariff engineering
- Designing and sourcing a product so that its genuine characteristics and genuine supply chain carry a lower duty burden — and being able to substantiate that with the product itself, the manufacturing process and the records.
The distinction that matters is between the product and the paperwork. Deciding that a garment will be knitted rather than woven, or that assembly happens in a country whose output genuinely qualifies for a preference, changes the thing being imported, and the declaration then follows honestly. Deciding that an unchanged product will simply be declared differently changes nothing except the accuracy of the declaration.
That gives a sequencing rule: tariff engineering happens before the product exists. Once the tech pack is frozen and the factory booked, almost every lawful lever has closed — which is why duty belongs in the design review rather than in the import file.
How a duty rate is actually decided
The European Commission sets out three factors that determine the duty on imported goods: the tariff applicable, the value of the goods, and their origin. Duty is normally charged as a percentage of the customs value, so all three have to be settled before the amount is knowable.
- Classification. Which heading of the tariff the goods fall in. The WCO Harmonized System covers more than 5,000 commodity groups in six-digit codes, is used by more than 200 countries and economies, and — the WCO states — classifies over 98% of merchandise in international trade. The EU extends those digits in its Combined Nomenclature.
- Origin. Where the goods are considered to have been produced, decided by legal rules about the manufacturing process rather than by where the container was loaded. Origin decides whether a preferential rate can apply at all.
- Customs value. In the EU the primary basis is the transaction value — the price actually paid or payable for goods sold for export to the Union, adjusted under specific rules. Transport to the point of entry is included; transport after entry and the duties themselves are not.
Other mechanisms move what you pay without touching any of the three. Tariff quotas let a set volume enter at a reduced or zero rate within a defined period, so timing and quantity can matter; agreements often dismantle duties gradually, so the rate in two years may not be the rate today; and anti-dumping, countervailing and safeguard duties can sit on top of the normal rate.
Five lawful levers
1. Product design and material composition
Classification follows what a product is — its material composition, its construction, sometimes its function. Different headings carry different rates, so a real design decision (a different fibre blend, construction, closure or level of assembly) can genuinely place a product in a different heading, at a different rate.
Three conditions make this design rather than declaration. The change has to be real, and the product honestly describable in the new terms. It has to be decided while the design is still open, alongside everything else a material change affects — performance, durability, cost, testing. And the reasoning has to be written down, because a classification you cannot explain in two years is one you cannot defend.
2. Comparing sourcing countries on duty, as one input among several
Because origin is one of the three factors, the same product can carry different rates depending on where it is genuinely made, and choosing where to place production is an ordinary commercial decision. Treat the rate as one line in a landed-cost comparison rather than the answer: freight, transit time, working capital, factory capability and quality risk sit in the same comparison, and a lower rate on a lane you cannot run reliably is not a saving.
The condition is that the origin must be the real one. A country where a factory actually manufactures is a sourcing decision; a country the goods merely pass through is not.
3. Trade preferences, where the product genuinely meets the rules of origin
A preferential agreement can let goods enter at a reduced or zero duty — but only goods that qualify. The Commission describes two routes to originating status: goods wholly obtained in the partner country, and goods sufficiently worked or processed there. The second is a product-specific test in one of three forms.
- A value-added rule, capping non-originating material value as a percentage of the ex-works price of the finished product.
- A change of tariff classification rule — the Commission's example is paper of Chapter 48 made from non-originating pulp of Chapter 47.
- A specific operation that must be carried out, such as spinning natural fibres into yarn. Rules of this kind are common for textiles, clothing and chemicals.
Around those rules sit three mechanisms that are routinely misread. Cumulation — bilateral, diagonal, regional or full, depending on the agreement — lets materials or processing from partner countries count towards origin. A general tolerance, around 10% of the ex-works price, allows a small share of non-originating material that would otherwise break a rule, but cannot breach a maximum-value threshold the rule already sets. And minimal operations never confer origin alone: preserving, packaging, simple cutting, simple assembly, mixing, ironing textiles, painting and polishing are named as insufficient — while exceeding them is not automatically enough, because the product rule still has to be met.
The qualifying processing also has to happen in the partner country, and the goods must not be altered in transit. And the claim has to be evidenced — an exporter's statement on origin, a movement certificate such as EUR.1, or importer's knowledge. If the rules are not met the ordinary most-favoured-nation rate applies, so an unsupported preference claim is an exposure rather than a saving.
4. Supply-chain configuration
Where value is added, where assembly happens and which market the goods enter are configuration choices with duty consequences: placing the origin-conferring operation in a country whose output actually qualifies, sourcing components from partner countries so cumulation applies, or importing an item as components rather than finished, which classify differently. These change the chain, not the description of it, and they only work when the operations are substantive.
5. Landed-cost modelling before design and sourcing lock
This is the lever that makes the other four usable, and the one most often skipped. If the duty is first computed when the entry is filed, every design and sourcing option has already been spent. Modelling two or three viable configurations, with the classification and origin estimated for each, turns duty into an input to a decision that is still open. The estimate need not be certain: it has to be early, include the customs value and not only the rate, and be revisited when the design changes. See what landed cost includes and the landed cost guide.
When to ask an authority rather than decide yourself
Ambiguity in classification is a reason to ask a question, never a reason to pick the cheaper heading. In the EU, Binding Tariff Information is a legal decision issued by a member state customs authority on the classification of a product: it binds all EU customs administrations and the holder, is normally valid three years, cannot be backdated, and is published in the EBTI database. Binding Origin Information does the equivalent for origin. The Commission also publishes a self-assessment tool (ROSA) in Access2Markets, good for orientation.
- The product could credibly sit in more than one heading and the rates differ materially.
- The volume is large enough that being wrong is expensive, or the product will be imported for years.
- A preference claim rests on a rule you are interpreting rather than plainly satisfying, or a supplier gives you the classification and cannot show how they arrived at it.
The documentation is the deliverable
A duty position you cannot substantiate is not a saving; it is a liability that has not been called yet. Whatever lever you use, the output is a file, cheaper to build during development than to reconstruct later.
- Before you rely on a duty position
- The classification, written down, with the product characteristics and reasoning behind it.
- The origin determined by the applicable rule, not by where the goods shipped from.
- For a preference: the specific rule, the calculation, and the proof of origin the agreement requires.
- Supplier declarations and a bill of materials with the origin and value of each input.
- The customs value basis and the adjustments applied to it.
- A named person accountable for the position, and a review date, because no input is permanent.
Where this goes wrong
- Claiming a preference without holding valid proof of origin and the evidence behind it.
- Assuming that where the factory sits is the origin. Origin is a legal test on the manufacturing process, not a shipping address.
- Reading the general tolerance as a broad allowance rather than a narrow one.
- Modelling the rate but not the customs value it applies to, then finding that freight to the border is dutiable.
How Library of Trade approaches it
We treat duty as a product-development question rather than a shipping one, which mostly means raising it while materials, construction and origin are all still open. AI gives our team the scale to compare many material, factory and lane combinations against a brief; our experts judge which of them is actually defensible. Where a position is genuinely uncertain the recommendation is to get a ruling or professional advice, not to pick the cheaper reading and hope. See sourcing from Vietnam for how this plays out in one origin.