Landed cost is supposed to be every cost of getting a unit saleable in the destination warehouse. Two costs routinely escape the model: cargo insurance, bought per shipment or on an annual policy, and inspection fees, where a buyer pays for pre shipment checks on quality or quantity. Both are caused by the shipment, both scale with volume, and leaving them out understates unit cost by a small percentage that compounds across every pricing decision the number feeds.
The landed cost model usually grows from two documents: the commercial invoice and the freight bill, with duty from the entry. Insurance often sits on a separate policy, sometimes an annual open cover paid as a lump, and inspection is invoiced by a service company that appears nowhere in the logistics chain. Costs that do not arrive attached to a shipment do not get allocated to one, not because anyone decided they were free, but because the spreadsheet never met them.
The same blind spot hides brokerage disbursement gaps, document fees, and currency conversion spreads. The pattern is the tell: if a cost is paid to someone other than the supplier, the carrier, or customs, check whether the model knows it exists.
Insurance allocated by value makes sense, because the premium is charged on insured value: spread the period's premium across the period's shipped value and apply the resulting rate per unit. Inspection allocated by shipment or by inspection event fits its cause: a per container inspection belongs to that container's units, not smeared across the year's volume. The method matters less than the consistency and the honesty: pick the driver that matches why the cost was incurred, document it, and use it every period so margins stay comparable.
Do not chase the last cent. A model that allocates insurance to four decimal places and guesses freight is precision theatre. The goal is that no material cost line is zero by accident.
Unit margins tighten by a point or so on insured, inspected supply lines, which is enough to reorder a product range's apparent winners, and enough to change a make or drop decision at the margin. Pricing built on the corrected cost stops quietly subsidising the most inspected, most insured product line in the catalogue, usually the one with the quality history that made inspection necessary, which is information in itself.
There is a negotiation dividend too. Once inspection cost is visible per supplier, it can be discussed per supplier: fixed at source, charged back where the contract allows, or engineered down by fixing the failures the inspections keep catching. A cost nobody allocates is a cost nobody manages.
Related reading: How do peak-season surcharges change your landed-cost math?, How do you allocate freight across SKUs for accurate landed cost?.
For unit economics it belongs in landed cost, because the shipment causes it and it scales with what you ship. Finance may classify the annual policy as overhead in the statutory accounts, and both treatments can coexist: the management model that prices products should carry every cost the product causes, wherever the ledger parks it.
To the inspected shipments, and within them to the units inspected. Spreading one supplier's inspection costs across a clean supplier's units hides exactly the signal the allocation exists to reveal: which supply line costs more to buy from safely.
Include it on the same logic: reinspection fees, rework charged by the supplier where it is separately billed, and the freight consequences of a failed lot all belong to that lot's cost. A supplier whose goods fail inspection twice a year is a more expensive supplier than the invoice price suggests, and the model should say so.
Whenever the underlying premium or fee schedule changes, and at least quarterly for the value based insurance rate, since it floats with what you ship. A stale rate is just a more respectable version of the original omission.