Brokerage is a per-entry cost, but landed cost lives per SKU. The allocation methods, when each is fair, and the one that breaks under scrutiny.
Customs brokerage is billed per entry, but landed cost is managed per SKU. Somewhere between the broker's invoice and the product margin, that per-entry fee has to be divided across the units in the shipment, and the method chosen changes which products look profitable. A fee spread evenly across units flatters heavy, low-value goods; a fee spread by value flatters the expensive ones. The allocation is not neutral, so it should be deliberate.
The stakes are higher than the fee's size suggests. Brokerage might be a few hundred dollars per entry, but for brands making assortment decisions, the per-unit landed cost is the number that decides which SKUs survive. A systematically wrong allocation quietly subsidizes losers and taxes winners, and the distortion compounds across every pricing and purchasing decision built on top of it.
The simplest method is per-unit: divide the brokerage fee by the number of units in the entry. It is easy to compute and easy to explain, and it is fair when the entry contains similar goods where the broker's work does not vary much by SKU. Its weakness is obvious with mixed entries: the fee per unit is identical for a $5 accessory and a $500 device, even though the broker's effort and the compliance risk skew heavily toward the latter.
Value-based allocation divides the fee in proportion to each SKU's entered value. Expensive goods absorb more of the fee, which matches the intuition that higher-value goods drive more broker attention and more duty complexity. Weight or volume-based allocation suits bulky goods where the broker's work correlates with the physical shipment. Some brands use a hybrid: a fixed per-line-item charge for the broker's per-SKU work plus a value-based split of the remainder.
The method that survives is the one tied to cost causation: allocate the fee the way the broker's work actually distributes. Broker effort concentrates on line items, classifications, and valuation questions, not on unit counts. That points toward a per-line-item base charge with the remainder split by value, because it mirrors what the invoice is actually paying for.
Per-unit allocation breaks under scrutiny the moment entries mix high and low-value goods, which is to say almost always. It is defensible only for homogeneous entries. Value-based allocation is the safer default for mixed catalogs, with one caveat: it can distort decisions for low-value, high-volume SKUs that look worse than they are. The honest answer is that no single method is right for every brand; the right method is the one whose distortions you have examined and accepted.
Whatever method you choose, implement it as a parameter in the landed-cost model, not as a hardcoded assumption. The model should take the broker's per-entry fee schedule, the entry's SKU mix, and the chosen allocation rule, and produce per-SKU brokerage cost automatically. When the broker changes the fee schedule, one parameter updates everywhere.
Review the allocation annually or whenever the assortment mix shifts meaningfully. A brand that moves from homogeneous to mixed entries, or that adds a high-value category, should re-examine whether the old method still fits. And keep the broker's invoices reconciled against the model's assumptions: if the model assumes one entry per container and the broker is filing two, the allocation is wrong before the math even starts.
By value for most DTC catalogs, because broker effort and compliance risk track value more closely than weight. Weight-based allocation makes sense for bulky, low-value goods where the physical handling dominates. If the assortment mixes both profiles, the hybrid per-line-item plus value-split method handles the tension best.
Flat-fee entries simplify the invoice but not the allocation problem; the fee still has to land on SKUs somehow. Apply the same allocation logic to the flat fee. The flat fee does make the per-entry cost predictable, which is its own win for the landed-cost model.
The allocation rule itself rarely needs to change, but its inputs change with every entry. Recompute per entry automatically, and revisit the rule when the business changes: new categories, new value mix, or a new broker fee schedule. An annual review tied to the broker contract renewal keeps it honest.