Rules of origin for processed food: four tests, and yours is only one of them

For a primary product, origin is close to obvious. Fish caught by a Canadian-flagged vessel, grain grown in Saskatchewan, an animal born and raised here: these are originating because of where they came from, and the paperwork exists to record something nobody seriously disputes.

Processed food is different, and the difference is where most preference claims either fail or are never made. A product manufactured in Canada from a mix of Canadian and imported inputs is not originating because it was manufactured here. It is originating if it satisfies a specific rule written for its tariff classification, and that rule may be nothing like the rule for the product next to it on the same production line.

The four tests, and how to tell which is yours

Rules of origin under an agreement such as CPTPP take a small number of forms. Knowing which one applies to your product is the whole job, and it is a lookup, not a judgement.

**Wholly obtained.** The product is entirely from the territory: grown, harvested, caught, or made only from materials that are themselves wholly obtained. This covers most Canadian primary exports and is the easiest to evidence, because the question is a factual one about where things happened.

**Change in tariff classification.** The rule says that every non-originating input must move from one part of the nomenclature to another as a result of the processing done in Canada. A rule requiring a change at chapter level is a demanding rule; one requiring a change at heading or subheading level is easier.

The test is about the inputs that were not already originating, which is why a product with entirely Canadian inputs usually passes without arithmetic.

**Regional value content.** The rule sets a percentage of value that must originate. These rules come with a prescribed calculation method, and the method matters as much as the threshold, because different methods measure different things and produce different numbers from the same facts.

**A combination, or an alternative.** Many rules for processed food offer a choice: satisfy the classification change, or meet a value threshold. Where both are available, most companies can satisfy one more cheaply than the other, and which one is cheaper is worth establishing rather than assuming.

Two supporting provisions apply across the tests and are worth knowing because they rescue claims that would otherwise fail.

**De minimis** allows a small proportion of non-originating material to fail the classification change without disqualifying the product. It is a tolerance, it is capped, and for some agricultural goods it is restricted or unavailable, so it is not a general safety net.

**Accumulation** lets material originating in another party to the agreement count as originating for your purposes. Under CPTPP that means inputs from other member countries can help you qualify, which changes sourcing decisions for anyone buying ingredients internationally.

The rule is written against the classification

This is the connection exporters most often miss, and it makes classification a prerequisite, not a parallel task.

Rules of origin are organised by tariff line. To find your rule you must first know your classification, and if the classification is wrong you will read the wrong rule, apply it correctly, and arrive at a confident answer that is worthless. The goods might have qualified under the right rule; the claim is still defective, because it was made under a rule intended for something else.

So the order is fixed: classify, then find the rule, then test the product against it. Doing it the other way round, or doing them in parallel by different people, is how a claim ends up resting on a mismatch nobody noticed.

What the test actually costs to run

For a wholly obtained product, very little. The evidence is a description of the process and records showing the inputs were what you say.

For a classification change rule, the work is a bill of materials in which every non-originating input is classified. That is the hidden cost: you are not classifying one product, you are classifying every imported ingredient in it. A formulation with thirty inputs, eight of them imported, requires eight classifications before the rule can be applied once.

For a value content rule, the work is accounting. The prescribed method determines which costs count and how they are valued, and the calculation has to be reproducible from records rather than reconstructed from memory. Companies frequently discover that the numbers their finance system produces are not the numbers the method asks for.

None of this is difficult. All of it takes longer than the week before a shipment, which is when it is usually attempted.

Sourcing decisions change the answer

Once the rule is known, it becomes a constraint on procurement, and treating it as one is where the real value sits.

If the rule requires a classification change and one imported ingredient does not make it, the options are to source that ingredient from Canada, source it from another party to the agreement so accumulation applies, or process it further so that it does change classification. Each of those is a business decision with a cost, and each can be compared against the duty saving the claim would produce.

That comparison is worth making explicitly. A preference claim is worth the duty it avoids, multiplied by the volume, minus the cost of qualifying. For a high-volume product facing a meaningful rate, reformulating to qualify can pay for itself quickly. For a low-volume product facing a small rate, the honest answer is sometimes to pay the duty and stop spending on the question.

Very few companies make that calculation deliberately. Most either claim without checking, which is a liability, or never claim at all, which is a cost.

Advance rulings exist and are underused

Companies can approach the customs administration in the market they are targeting to receive an advance ruling on the origin of their product.

That is a genuinely useful facility and it is used far less than it should be. A ruling settles the question with the authority that would otherwise raise it, before the goods move, and it converts an argument you might lose later into a document you hold now.

The obvious objection is time, and it is a real one: a ruling takes weeks that a first shipment may not have. The answer is to start it early instead of skipping it, and to recognise that a ruling obtained for the first shipment serves every subsequent one.

A ruling binds the authority that issued it. For a product going to several of the six markets, expect to make the case more than once, and keep the technical description identical across all of them.

The claim has to be supportable when nobody is watching

The last point is the one that determines whether any of this was worth doing.

A preference claim is a statement that the goods satisfy a rule. Satisfying it is a fact about the product; being able to demonstrate it is a fact about your records, and only the second one survives an audit two years later.

What has to exist, from before the claim rather than after a question:

• The classification of the finished product, and the basis for it
• The bill of materials, with the origin of every input
• The classification of every non-originating input, where the rule requires a change
• Supplier declarations supporting the origin claimed for inputs
• The calculation, where a value rule applies, in the prescribed method
• A record of formulation changes, so the file describes the product as it actually shipped

That last item is the one that quietly breaks otherwise excellent files. A company that qualified a product in 2024 and changed an ingredient supplier in 2025 has a file describing a product it no longer makes, and a claim that has been wrong ever since without anybody noticing.

Three markets are inside and three are not

Worth keeping in view, because the work above only pays where an agreement exists to pay it.

CPTPP is in force between Canada, Australia, Brunei, Chile, Japan, Malaysia, Mexico, New Zealand, Peru, Singapore and Vietnam. Of the six markets on this route, Malaysia, Singapore and Vietnam are members and Indonesia, the Philippines and Thailand are not.

For the three outside, the ordinary applicable rate governs, and rules of origin in the preference sense do not arise. Origin still matters for the entry, and the labelling claim still matters, but there is no qualification exercise to run and no certification of origin to make.

That divides the work usefully. A company selling into all six is doing origin qualification for three of them and not the other three, and the file it builds for the first three is the file that serves any future agreement covering the others.

It also has a sequencing implication for a company choosing where to start. Where a product faces a meaningful rate and would qualify, the three member markets carry an advantage that compounds with volume, and it is a structural advantage over a competitor outside the agreement, not a temporary one.

Where to start

• Classify the finished product first, because the rule is written against it
• Look up the specific rule for that line instead of assuming a general one
• Establish which test applies, and whether an alternative is offered
• Build the bill of materials with origins before doing anything else
• Compare the cost of qualifying against the duty actually at stake, and be willing to conclude it is not worth it
• Consider an advance ruling for anything significant or borderline, and start it early

The theme is that origin for preference is a technical exercise with a definite answer, not a general impression that a product is Canadian. The impression is usually right. The claim has to be more than an impression.