Freight does not split evenly

The container arrives. Inside are forty different products from three vendors, and on the table sit the customs broker's invoice, the carrier's, the handling crew's, and the bank statement with the wire transfer fee.

The question is easy to ask and hard to answer well: how much of that belongs to each product?

Almost everyone does this calculation. Almost nobody does it right, and the error is never visible — it dissolves into a margin that comes out slightly below expectations, month after month, with nobody quite sure why.

What you paid is not what it cost

The purchase price is only the first line. On top of it go:

• International freight — ocean, air or road.
• Customs broker fees and the costs of clearing the entry.
• Import duties, where they apply.
• Bank fees on the transfer to the vendor.
• Handling, palletising, labelling and everything done on the floor before the goods can be sold.
• Cargo insurance.

Added up, on a normal import those items run from 15% to 40% on top of the invoice value. That is not an accounting detail: it is the difference between believing you make twenty points and making eight.

The three allocation bases

Here is the point of this article. Not every cost is split the same way, because not every cost is caused the same way.

There are three reasonable bases, and each item belongs naturally to one:

• By volume — when what you pay for is space. Ocean freight, storage.
• By units — when you pay to handle pieces. Labelling, handling, palletising.
• By value — when cost is proportional to money. Insurance, bank fees, ad valorem duty.

The rule is common sense: allocate each cost by whatever caused it. You pay the carrier for cubic metres, so freight is split by cubic metres. You pay the insurer on insured value, so insurance is split by value.

The mistake that costs money

The classic mistake isn't picking the wrong base. It is using one base for everything — usually value, because it is the one closest to hand.

An example with round numbers. A 30 m³ container with two products:

• Product A — 1,000 units, $10,000 USD purchase, occupying 5 m³.
• Product B — 100 units, $10,000 USD purchase, occupying 25 m³.

Freight cost $3,000 USD. Both products are worth the same, so splitting by value gives $1,500 to each. Clean and quick.

Now split it by volume, which is what you actually paid the carrier for. A took 5 of 30 m³: it gets $500. B took 25: it gets $2,500.

Watch what happens to unit cost:

• Product A, 1,000 units. Purchase $10.00. Splitting freight by value, $11.50; splitting it by volume, $10.50.
• Product B, 100 units. Purchase $100.00. Splitting by value, $115.00; splitting it by volume, $125.00.

Splitting by value made A subsidise B. Product A looks 9.5% more expensive than it is, and B looks 8% cheaper than it cost.

The consequences are the two worst possible ones at the same time. On A you are declining to compete over a cost that doesn't exist. And on B you have spent months selling at eight points less margin than you think you have — which in distribution is usually the entire margin.

When a cost has no natural base

Not every item fits so cleanly. Customs broker fees, for instance, are charged per entry: they are a fixed cost of the whole shipment and no single product caused them.

There, choosing a base is a convention, not a deduction. And what matters isn't which one you pick, but two things:

1. Write it down. Document why that item is allocated that way.
2. Don't change it. If one shipment splits broker fees by value and the next by units, historical costs stop being comparable and any per-product profitability analysis turns into noise.

An imperfect base applied consistently beats the perfect base applied to half your shipments.

The same product, two different costs

This is unsettling the first time it happens: the same SKU, in two different shipments, ends up with different costs. Sometimes even within the same container, if it came from two orders on different terms.

It is not a data-entry error. It is what actually happened. January's freight didn't cost what March's did, the dollar wasn't where it is today, and February's container travelled emptier. If your system shows the same cost for all three, what it is doing is averaging — which is fine for valuing inventory, but misleading for setting a price.

Replacement cost and historical cost are different numbers and serve different purposes. Worth not confusing them.

The exchange rate that matters

One detail that gets overlooked: the relevant exchange rate is the one you actually bought the dollars at.

Not the official rate on the receipt date, not the one on the day customs released the shipment, and not whatever the system has loaded. If you bought dollars at 18.40 and the system values at 17.90, you have half a peso per dollar of cost sitting nowhere — and on a fifty-thousand-dollar shipment that is twenty-five thousand pesos that vanished from cost and will reappear later as an exchange loss nobody connects to that import.

And when you need the number

Here is what makes all of the above urgent rather than academic.

Between placing the order and the goods reaching the warehouse, two or three months can pass. Commercial operations do not stop during that time: you quote, you promise and you sell goods that are still on the water.

If real cost is only known once the container lands, every one of those decisions was made on an estimate. And an estimate that is eight points off on a fast-moving product cannot be corrected afterwards: it has already been sold.

That is why import costing isn't a month-end accounting task. It is a commercial tool, and it is useful to the extent that it is ready before receipt.

How we solved it

We built Containers around exactly that idea: group the orders travelling together in one place, load their expenses as the invoices arrive, allocate them on the base each item deserves, and have the real cost of every unit before receipt.

It is not magic and it is not an algorithm: it is the same arithmetic as this article, done in the system instead of a spreadsheet, with the result available while it can still inform a price.

If you import and that calculation currently lives in a spreadsheet only one person understands, it is worth talking.