What is landed cost, and why is my customs broker's number always wrong?

By Saurabh Goyal, Founder & CEO of Phlo Systems. Published 21 May 2026.

Ask three customs brokers to quote the landed cost for the same shipment and you will get three different answers — and none of them will match what actually hits your bank account when the goods clear. This is not because brokers are dishonest. It is because "landed cost" is not the simple number it appears to be, and the way most brokers calculate it omits several components that have become material in the last 18 months.

For a commodity trader, an importer, or a procurement lead, this matters more than it used to. Tariff volatility — US Section 301 and Section 232, the EU Carbon Border Adjustment Mechanism (CBAM), the UK's CDS-era rate revisions, and the rolling sanctions overlay — has turned duty from a small predictable cost line into a margin variable. If the number you priced your deal against is wrong by three percent, your gross margin can disappear before the container leaves the port.

Here is what the components actually are, why brokers under-quote, and what an importer needs to do about it in 2026.

The seven components of true landed cost (and the three brokers usually omit)

A complete landed cost has seven components:

1. Goods cost — the invoice value from the supplier, FOB or EXW depending on the incoterm.
2. Freight — sea, air, road, or multi-modal carriage to the destination port or door.
3. Insurance — marine cargo or equivalent.
4. Customs duty — the MFN rate or applicable preferential rate, applied to the customs value.
5. VAT or sales tax — applied at import on the duty-paid value.
6. Other border charges — anti-dumping duties, countervailing duties, excise where applicable, CBAM levies (from 2026), and regime-specific charges (e.g. US Merchandise Processing Fee, Harbour Maintenance Fee).
7. Inland and handling — port handling, terminal fees, demurrage risk, broker fees, last-mile transport, warehouse-in.

Most broker landed-cost quotes cover 1, 2, 3, 4, and 5 reliably. The three that get omitted are:

• Component 6 in full. Anti-dumping and countervailing duties are surfaced when the broker knows to look. CBAM levies — applicable to EU imports of cement, iron and steel, aluminium, fertilisers, electricity, and hydrogen — are routinely missed because they live outside the standard tariff lookup workflow.
• Component 7's variable risk. Demurrage, detention, and storage costs are treated as exceptions rather than expected costs. For a container that misses a discharge slot, the demurrage bill alone can be a material percentage of the goods value.
• Currency conversion timing. The customs value is converted at the official exchange rate on the date of import, not the rate at which you priced the deal. For volatile pairs over a 30-day shipment window, this can move the landed cost by 1–3%.

Add those three and "landed cost" looks materially different from the broker's number.

Why HS code classification accuracy is now a P&L issue, not a compliance issue

The traditional view of customs classification is that it is a compliance task — get the HS code right to satisfy the customs authority, avoid penalties, move on. That view is outdated.

In 2026, the same product can attract three different duty rates depending on what HS code it is classified under:

• The MFN rate under the destination's standard tariff.
• A preferential rate under an applicable trade agreement (UKCA, DCTS, EU FTAs, USMCA, etc.).
• An additional rate from Section 301, Section 232, CBAM, or anti-dumping orders that hook onto specific HS chapters.

The gap between rates for the same physical product can be 0% versus 25%+ depending on classification. That is not a compliance question — that is a pricing question, a margin question, and increasingly a deal-survives-or-doesn't question.

Two recent examples make this concrete. EU steel imports under chapter 72 attract CBAM levies; the same product reclassified under chapter 73 (further-processed articles of iron or steel) may or may not — the line is in the chapter notes and is being actively litigated. US imports of certain semiconductor-adjacent goods classified under 8541 versus 8542 can differ by double-digit percentage points after Section 301 layering. These are not edge cases. They are common product categories with high trade volume.

For a commodity trader pricing a deal, getting the HS code right at the pre-trade stage is now part of margin protection, not a back-office task to do after the goods arrive.

CBAM, Section 301, and the rise of duty as a margin variable

Three policy regimes have moved duty from a stable input cost to a margin variable that has to be modelled actively:

CBAM (EU). The transitional phase ran 2023–2025 with reporting-only obligations. The definitive phase began on 1 January 2026. EU importers of cement, iron and steel, aluminium, fertilisers, hydrogen, and electricity must purchase CBAM certificates corresponding to the embedded emissions of imported goods, less any carbon price paid in the country of origin. For a steel importer running 50,000 MT/year, CBAM exposure can run into seven figures annually depending on supplier emissions intensity and certificate price.

Section 301 and 232 (US). Section 301 tariffs on Chinese-origin goods (lists 1–4) remain in force with periodic adjustments. Section 232 tariffs on steel and aluminium imports apply origin-based rates regardless of where the goods are shipped from. The combined effect is that "country of origin" is now a P&L lever — and origin engineering (legitimate, aggressive, or reckless) is being practised at scale.

Sanctions overlay. Russia-origin goods, Belarus-origin goods, and certain Iranian and DPRK-origin goods face prohibition or punitive rates in most G7 markets. Sanctioned-entity screening at the supplier and ultimate-beneficial-owner level is now part of the duty determination, not a separate compliance task.

The practical implication: a trader who priced a steel deal in October 2025 against the prevailing duty profile, and ships it for delivery in February 2026, is exposed to a different duty regime than the one priced against. Duty is no longer a stable cost — it is a curve that has to be modelled and re-modelled across the shipment window.

Pre-trade vs post-trade landed cost: why both numbers must reconcile

Most import operations carry two landed cost numbers and rarely reconcile them:

• The pre-trade landed cost is the number the trader or buyer uses to price the deal. It is built from supplier quotes, freight indications, and an estimated duty number — sometimes calculated, often guessed.
• The post-trade landed cost is the actual cost that hits the GL after the goods clear. It comes from the broker's clearance entry, the actual freight invoice, the actual duty paid, and the actual incidentals.

When these two numbers diverge — and they usually do — the variance is the trader's margin. If the pre-trade number was too low, margin erodes. If it was too high, the firm priced itself out of deals it could have won.

A serious landed-cost discipline requires:

• Pre-trade calculation against current tariff data — including any active anti-dumping orders, CBAM exposure, and Section 301 layering for the specific HS code and origin combination.
• Reconciliation on every import — comparing pre-trade estimate to post-clearance actuals, with variance reasons captured (rate change, classification change, FX move, freight overrun).
• Feedback into pricing models — so the next deal's pre-trade estimate reflects what was learned from the last one.

This is the discipline most SME importers do not run, because it requires data flow between the deal-pricing workflow, the classification engine, the customs entry, and the GL. The four typically live in four different systems.

How AI-driven classification changes the economics of small-shipment trading

Until recently, the economics of customs classification favoured large shipments. A 5,000 MT bulk steel shipment can justify professional classification time. A €20,000 mixed container of specialty chemicals cannot — the broker uses a default code, accepts the rate, and moves on.

AI-driven classification compresses that cost. A modern AI classification engine — including customs-compliance.ai, which Phlo Systems builds — runs a three-stage process (cache lookup, embedding search across hundreds of thousands of commodity codes, LLM re-ranking with written justification) at marginal cost per classification near zero. The economic floor on "is it worth classifying this properly" drops from £100s per line to pennies.

The practical effects:

• Small-shipment importers can now price accurately. A 200-line packing list that previously got default-coded can be properly classified in minutes, with a written justification for each code.
• Preferential-rate qualification becomes routine. Many SME importers leave preferential rates on the table because qualifying each line under rules of origin is too time-consuming. AI brings the cost of that work down to where it is worth doing.
• Tariff changes can be re-scored at scale. When the EU updates its Combined Nomenclature, when the US updates Section 301 lists, when the UK publishes new tariff chapters, an importer can re-classify its product master against the new rules in hours, not weeks.

The economic argument for proper classification at every shipment size is now decisive. The argument used to be cost-versus-accuracy. AI removes the cost side of that trade-off.

Building a landed-cost model that survives audit and re-pricing

For an importer or commodity trader who wants a landed-cost model that holds up under audit and supports re-pricing under changing duty regimes, four properties matter:

1. Current data, not snapshot data. The tariff schedules behind the model must update on the same cycle as the underlying authorities publish changes — daily for HMRC, weekly or rolling for EU CN amendments, ad hoc for US executive orders. A static tariff table is wrong within weeks.

2. Written justification per classification. Every HS code attached to a product must be defensible — chapter notes considered, alternative classifications evaluated, reasoning recorded. This is a Rule 6 due diligence requirement for brokers; for self-classifying importers, it is the evidence base in a post-clearance audit.

3. Pre-trade and post-trade reconciliation. The same model must price a deal pre-trade and accept post-trade actuals, with variance captured. Without this, pricing drifts away from reality and no one notices until quarter-end.

4. Integration with the ERP or accounting system. Landed cost feeds inventory valuation, COGS, and margin reporting. If it lives in a spreadsheet disconnected from the GL, it cannot survive audit and it cannot drive pricing decisions reliably. Native connectors into Xero, Acumatica, or a unified platform like opsPhlo close that loop.

The firms that build this discipline now will have a structural pricing advantage over competitors still working from broker-quoted numbers. The firms that don't will discover their margin in the variance.

Frequently Asked Questions

What is the difference between landed cost and total cost of ownership?

Landed cost is the cost to get the goods from supplier to your warehouse, customs-cleared and ready to sell or use. Total cost of ownership includes downstream costs — financing, storage, depreciation, disposal. For a commodity trader reselling within weeks, landed cost is usually the relevant number. For a manufacturer holding inventory long-term, TCO matters more.

How accurate are AI-generated landed cost calculations?

Accuracy depends on the quality of the underlying tariff data and the classification logic. A well-maintained AI tool that syncs daily against official tariff sources and produces a written classification justification can match or exceed broker accuracy for the duty and classification components. Freight and incidentals are still inputs the user supplies; AI does not invent shipping rates.

Why don't brokers include CBAM in their landed cost quotes?

Two reasons. First, CBAM is administered separately from the standard customs entry workflow — importers register with their national competent authority, purchase certificates, and submit quarterly returns outside the customs declaration. Second, the embedded emissions data needed to calculate CBAM exposure comes from the supplier, not the broker. Brokers reasonably defer that calculation to the importer, but the cost is real and must be in the landed-cost model.

Can I claim a preferential rate retrospectively if I find I qualified?

In most jurisdictions yes, within a defined time limit — typically 3 years in the UK and EU, varying elsewhere. The mechanism is a post-clearance amendment supported by valid proof of origin documentation. Many SME importers miss preferential rates at clearance and never go back to reclaim them. A landed-cost discipline that runs reconciliation will surface these as a recovery opportunity.

How does Section 301 affect non-Chinese suppliers?

Section 301 is origin-based, not shipment-route based. Goods of Chinese origin attract the tariff regardless of where they ship from or where they are processed lightly. Goods of non-Chinese origin do not, even if shipped from a Chinese port. The determinative question is substantial transformation: where the goods were last substantively manufactured. Origin engineering — moving production to circumvent Section 301 — is a separate workstream with its own legal and compliance risks.

What's a realistic budget for getting landed-cost discipline right?

For an SME importer running 500–5,000 entries per year, a modern AI classification tool plus reconciliation discipline costs in the range of £500–£5,000 per year all-in (tools plus analyst time). The recovered margin from properly classified preferential rates and properly modelled CBAM exposure typically exceeds this by an order of magnitude for any importer with meaningful trade flow.

How Phlo Systems helps

customs-compliance.ai is the customs and freight intelligence layer Phlo Systems built for traders. It runs AI-driven HS classification against 51 country tariff schedules at full national-level depth (8–10 digit), syncs daily against official tariff sources, calculates full landed cost including duty, VAT, preferential rates, and applicable border charges, and produces a written classification justification for every lookup.

For importers running Xero or Acumatica, the X-Customs and AcuCustoms connectors push the classification and landed-cost results directly into the accounting or ERP system — closing the pre-trade-to-post-trade reconciliation loop natively.

The free tier handles 50 classifications per month, no card required. The Business tier (£39/month) covers unlimited lookups and landed-cost calculation. Enterprise pricing (£99/month) unlocks API access and ERP connectors.

If you want a side-by-side comparison against your own product master and actual trade lanes — including CBAM and Section 301 exposure modelling — request a demo at customs-compliance.ai.

---

Related reading:

• SimplyDuty vs customs-compliance.ai: A Feature-by-Feature Comparison for 2026
• CargoWise Alternatives for SME Customs Brokers and Freight Forwarders in 2026
• What is the difference between Commodity Management (CM) and CTRM?

Saurabh Goyal is the Founder & CEO of Phlo Systems. He has built customs and trade infrastructure for commodity traders for over 15 years; customs-compliance.ai is used today by trading firms including Omni, Konexus, and Torq Commodities.