Customs
The HTS Classification Mistakes That Quietly Cost Importers Money
Quick answer: Most duty lost to classification is lost to ordinary error, not missed cleverness. The usual cause is that nobody chose the code deliberately. A supplier put something on an invoice, a broker filed it, and it has been repeating ever since. The categories that go wrong most are apparel, footwear, sets and kits, and parts versus accessories, because in each the deciding attribute lives in a factory spec sheet rather than on the invoice. Audit your top ten products by duty paid and you will usually find at least one you cannot justify.
Ask an importer how their HTS codes were chosen and you get one of three answers.
The supplier provided them. The broker set them up at onboarding. Or nobody actually knows, they have simply always been there.
None of those is a classification decision. They are all defaults, and defaults are how importers end up paying the wrong rate for years in either direction.
Classification is not a judgement call dressed up as one. It runs on six General Rules of Interpretation applied in strict order, and you do not reach rule two until rule one fails to resolve the question.
GRI 1 decides most cases. You classify according to the terms of the headings and any relative section or chapter notes. The notes are the part people skip, and they are binding. A note can expressly exclude your product from a heading that otherwise describes it perfectly, and route it somewhere else entirely.
GRI 2(a) covers incomplete, unfinished, unassembled and disassembled goods. If the article as presented has the essential character of the finished article, it is generally classified as the finished article. This is why shipping something in pieces usually does not achieve what people hope.
GRI 2(b) extends headings referring to a material to goods partly of that material, and sends mixtures on to GRI 3.
GRI 3 is where the money is. When two or more headings could apply:
GRI 4 classifies goods most akin to something else, and is genuinely rare.
GRI 5 deals with fitted cases and packing materials.
GRI 6 applies the same logic when choosing between subheadings.
The practical takeaway is that GRI 3(b) essential character is where competent people legitimately disagree, and where the largest duty differences sit. It is also the point at which a documented reasoning trail stops being paperwork and starts being your defence.
Classified by fibre content, construction and gender, and the thresholds are precise rather than approximate.
Whether a blend crosses 50% of one fibre can change the heading and the rate. Knit and woven go to different chapters. Gender specific headings exist and matter. Base rates in this space commonly run from around 16% to 32% before any Section 301 duty is added, so a heading error here is expensive per unit and repeats on every single order.
The failure mode is almost always the same. The tech pack says 60/40 cotton polyester, production drifts, the actual garment is 48/52, and nobody tells the person choosing the code because nobody realises that person exists. If you are running an apparel brand out of China, fibre content is a customs document.
Chapter 64 turns on the constituent material of the upper, the constituent material of the outer sole, whether the upper covers the ankle, and in some headings the value per pair. The rate range across the chapter is very wide.
Constituent material is determined by the greatest external surface area, which is a measurement, not an impression. Shoes with mixed uppers get classified on what dominates the external surface, which is frequently not what the designer would name as the main material.
Under GRI 3(b) a set put up for retail sale is classified by the component giving it essential character, and the whole set can take that component's rate.
That produces genuinely counterintuitive outcomes. A modestly priced item can set the rate for an expensive bundle, or the reverse. It also means the same items shipped loose and shipped as a retail set are different classification questions. Worth settling before you commit to kitting and bundle assembly at source, because the decision is easier to make before the boxes are printed.
A persistent problem in electronics and machinery. Section and chapter notes frequently exclude parts from the heading covering the machine, and direct them elsewhere. Whether something is a part, an accessory, or a standalone article with its own heading changes the answer, and the notes govern rather than intuition.
Whether a battery is installed, packed with the item, or absent can shift classification, and it interacts with dangerous goods requirements. We covered the shipping side in the guide to batteries and electronics from China. The classification side is a separate question and worth asking separately.
Nobody in the chain has an incentive to find it.
Your supplier puts a code on the invoice because a field requires one. In the US that code is a suggestion with no legal weight. Suppliers classify for export, if at all, and are not exposed to US penalties.
Your customs broker files what you instruct. A broker exercising reasonable care will query something obviously wrong, but they are not paid to re-derive classifications for a catalogue, and they do not have your spec sheets.
CBP has no reason to tell you that you have been overpaying. If the error runs the other way, discovery often comes through a request for information, a focused assessment, or an audit, by which point it has been repeating for years.
So the error compounds silently in whichever direction it happens to point.
This is a half day of work for most catalogues and it is the highest return half day available in customs compliance.
Rank by duty paid, not by revenue. Pull the last twelve months of entry summaries and sort by duty actually paid. Classification effort should follow duty, and the distribution is usually far more concentrated than people expect.
Take the top ten and write down five things for each. The code in use. The exact heading text, copied out. The section and chapter notes that apply. The physical attributes that drove the decision. Who decided, and when.
Most importers cannot complete that table for most products. The gaps are the finding.
Check the attributes against reality. Not against the tech pack, against a physical sample and the current production spec. Fibre percentages drift. Sole materials get substituted. Suppliers change subcomponents without telling anyone because it does not affect anything they are measured on.
Search CROSS. CBP's Customs Rulings Online Search System at rulings.cbp.gov holds decades of rulings, searchable by keyword and HTS number. There is a good chance CBP has already ruled on something close to your product, and a ruling on comparable merchandise is far better evidence than an internal opinion.
Escalate what you cannot justify. Anything where you cannot write down a defensible reason is a candidate for a binding ruling request, or for a conversation with a licensed customs broker or trade attorney.
As at 31 August 2026, Chinese goods entering the US face the MFN base duty for their code, Section 301 duties from the 2018 lists at 25% on Lists 1 to 3 and 7.5% on List 4A, and a Section 301 forced labour duty of 12.5% in effect since 24 July 2026. Section 232 duties apply on top where relevant. IEEPA tariffs were struck down by the Supreme Court on 20 February 2026, and the Section 122 surcharge expired on 24 July 2026.
Section 301 duties attach by HTS code. So the code does not only set your base rate, it determines which list you fall into and therefore whether you are paying 7.5% or 25% on top. A single misclassification can move both layers at once.
On the penalty side, 19 USC 1592 sets statutory maximums of 20% of the domestic value of the merchandise for negligence, 40% for gross negligence, and the full domestic value for fraud, on top of unpaid duty and interest. Negligence is failure to exercise reasonable care. The audit table above is, more or less, what reasonable care looks like written down.
Rates in this section are current as at 31 August 2026 and have changed several times this year. Verify before relying on them.
Classification decisions are only as good as the product information behind them, and that information is generated at the factory and the warehouse rather than in a spreadsheet.
From our Shenzhen facility we capture product attributes at intake, record materials and weights consistently, keep declared values and descriptions aligned across shipments, and hold audit ready records. When a supplier quietly changes a subcomponent, being at that end of the chain is where it gets noticed.
We are not a licensed customs broker and we are not trade counsel. We do not sign your entries and we do not take on importer of record responsibility. For a classification you intend to rely on, or anything that has already become a question from CBP, engage a licensed broker or trade attorney.
Classification is not exotic. It is a decision that most importers never consciously made, repeating on every shipment, in whichever direction it happened to land.
Rank by duty paid, take the top ten, and try to write down why each code is right. What you cannot justify is what to fix. Do that before you think about anything cleverer, because tariff engineering built on top of a wrong base code just makes the error more efficient.
For the wider framework see our guide to HS code classification and tariff engineering, or get in touch if you want a second look at the product data behind your codes.
$0.99 per order pick and pack. DHL/FedEx/UPS to 200+ countries. Tracking auto-syncs to Shopify. DDP so your customers never see a duty charge. 30 days free storage.
See eCommerce Fulfillment →The importer of record. Your supplier's code on the commercial invoice is a starting suggestion with no legal weight in the US. Your customs broker files what you instruct. The obligation to exercise reasonable care in classification sits with you, along with a five year record keeping requirement.
Six rules applied in strict order that determine tariff classification. GRI 1 classifies by the terms of the headings and any section or chapter notes, and resolves most cases. GRI 2 covers unassembled and incomplete goods. GRI 3 handles goods that could fall under two or more headings, using most specific description, then essential character, then last in numerical order. GRI 4 covers goods most akin to others. GRI 5 covers cases and packing. GRI 6 applies the same logic at subheading level.
Only within limits. Entries that have not yet liquidated can generally be corrected, and liquidated entries can be challenged by protest within the statutory window. Older entries are usually closed. This is why classification errors that overstate duty tend to be discovered long after most of the money is unrecoverable, and why an audit is worth doing sooner rather than later.
Apparel, footwear, retail sets and kits, and parts versus accessories in electronics and machinery. What they share is that classification turns on a physical attribute recorded in a factory specification rather than anything visible on a commercial invoice, such as fibre percentage, sole material, ankle coverage, or which component gives a set its essential character.
Rank your products by duty paid over the last twelve months, take the top ten, and for each one write down the code, the exact heading text, the section and chapter notes that apply, and the physical attributes that drove the decision. Search CROSS at rulings.cbp.gov for comparable products. Anything you cannot justify in writing is a candidate for a binding ruling request.