Take a mixed container: premium electronics and basic cast iron goods in the same box. Duty follows the value and rate of each line. Ocean freight follows the space and weight consumed. Entry fees follow the shipment and its line count. Any single key, value is the usual one, gets at least two of those wrong for at least one product, and the error is not random: it systematically flatters the products that consume the shared resource least in the key chosen.
The cast iron problem is the famous one. Cheap and heavy, it can be landed at a true cost that eats the margin whole, while a value based model shows it carrying only its small share of freight and looking like the safest line in the range. Businesses discontinue the wrong products for years on this arithmetic.
Cost each SKU in layers and keep them visible: purchase cost, duty on its own line rate and value, ad valorem fees by value, freight by the greater of weight or cube share as your carrier charges it, per line charges by line, and flat shipment costs by a stated convention. Layered costs can be audited line by line against the broker invoice and the carrier bill, which is what makes them trustworthy in a pricing argument.
Dimensional weight deserves its own attention in the freight layer. Bulky light goods are charged on cube by air and increasingly by ocean consolidators, so allocate on the chargeable weight the carrier used, not the scale weight in your warehouse system. The carrier's basis is the cost's basis. Anything else is a story.
Pricing is the obvious one: a tail of products reprices or exits once their true landed cost is visible, and a few products thought marginal turn out to be fine. Assortment is the quieter one: minimum order quantities, pack sizes, and supplier choices all shift when cube is priced, because a supplier whose packaging wastes a third more space is charging you freight rent on air.
Promotions change too. A discount approved on a value allocated margin can be a loss at true cost on the heavy lines, which is why promotional sign off should pull the layered cost, not the catalogue margin. The products that feature in promotions are disproportionately the bulky ones customers like delivered free. Of course they are.
A SKU level model dies if every shipment needs a week of spreadsheet care. Automate the boring parts: carrier and broker invoices captured by line, allocation rules coded once and reviewed quarterly, exceptions flagged where a shipment's profile breaks the assumptions. The analyst's job becomes investigating exceptions, which is analysis, instead of retyping invoices, which is not.
Review the keys annually against how carriers actually charge you now. Freight contracts change their weight and cube logic, fee schedules move, and a model faithfully applying a three year old basis is precisely wrong instead of roughly right. Roughly right, honestly labelled, is the standard a costing model can actually hold.
Keep reading: Landed cost variance alerts: set the threshold where a human will still read it, Duty changes while goods are on the water: which rate applies to your shipment?.
Because value only drives some of the costs. Freight follows weight and cube, entry charges follow lines and shipments. A single value key systematically misprices heavy, bulky, or cheap products, and the error always hides in the same direction.
The basis your carrier charges on: chargeable weight, which is the greater of actual and dimensional weight where dimensional pricing applies. Allocate each SKU its share of that basis so the products consuming the purchased capacity carry its cost.
Per shipment for actuals, with the allocation rules and rate assumptions reviewed quarterly and whenever carrier contracts or broker fee schedules change. Pricing can work from a rolling standard cost, provided variances to actual shipments are reported.
Margin surprises with a pattern: heavy or bulky lines that always disappoint, or a container mix change that moves total margin without any price changing. That pattern is the allocation key confessing.