Six weeks to a forecast that outperforms yours See how
Back to planning notes

The allocation choice that moves total delivered cost by a third

Nov 16, 2026, 9 min read

In short

Total delivered cost is assembled rather than measured, and the allocation rules that turn shared freight, duty and handling into a per unit figure move the answer further than most sourcing decisions do. Splitting one container by units, weight, volume or value gives four defensible costs for the same item, and the rule used is rarely written down. The cost of the time a unit spends as your asset is easy to compute and usually left out, because it sits in treasury rather than in logistics. Give a buyer a delivered cost per unit target and the purchase price falls, because everything downstream of the invoice belongs to somebody else's cost centre.

Procurement has a landed cost of 12.87 for the item. Finance has 14.02. Both figures came out of the same system, both describe the same shipment, and the gap is wide enough that one of them says the item earns a margin worth defending while the other says it barely clears the shelf. The meeting where this surfaces ends with somebody agreeing to reconcile the two offline, and the reconciliation never happens, because there is nothing to reconcile. Both numbers are arithmetically correct. They answer different questions, and nobody in the room knows which question they asked.

Total delivered cost gets assembled rather than measured. Every component arrives from a different system at a different granularity, and turning those components into a figure per unit takes a run of allocation choices that nobody writes down. Those choices move the answer by more than most sourcing decisions do.

The stack, and where the build usually stops early

A complete build runs from the supplier's invoice to the moment the unit leaves your control. In most businesses it contains:

Most builds stop at the first warehouse door. That is where the phrase landed cost traditionally ends, and it is also where roughly a third of the cost of getting a unit to a customer still lies ahead of you. Calling the result delivered cost while stopping at the inbound dock is the first and largest error, and it is usually invisible because the outbound half sits in a logistics cost centre that never gets pushed back to the item.

One container, four allocation rules

Here is the part that decides the answer. Take a single forty foot container carrying two items.

Item A is dense: 3,000 units, 6.0 kg each for 18,000 kg, 0.006 cubic metres each for 18 cubic metres, purchase price 12.00 a unit for 36,000. Item B is light and bulky: 2,000 units, 1.5 kg each for 3,000 kg, 0.0225 cubic metres each for 45 cubic metres, purchase price 30.00 a unit for 60,000. The container carries 21,000 kg and 63 cubic metres of goods worth 96,000, and the freight, surcharges, clearance and inland leg come to 6,300.

Now allocate that 6,300.

By weight, item A takes 18,000 of 21,000, which is 85.7 percent, or 5,400. That is 1.80 a unit. Item B takes 900, or 0.45 a unit.

By cube, item A takes 18 of 63, which is 28.6 percent, or 1,800. That is 0.60 a unit. Item B takes 4,500, or 2.25 a unit.

By purchase value, item A takes 36 of 96, which is 37.5 percent, or 2,362, giving 0.79 a unit. Item B takes 3,937, or 1.97 a unit.

By unit count, both take 1.26 a unit.

So item A's inbound logistics cost is 0.60, 0.79, 1.26 or 1.80 a unit depending on a rule nobody in the business chose deliberately. Its delivered cost at the inbound dock is 12.60, 12.79, 13.26 or 13.80. If it sells ex-warehouse at 15.40, its gross margin is 18.2 percent under cube allocation and 10.4 percent under weight allocation. Same container, same money, same physical facts, two very different verdicts on whether the item deserves shelf space.

There is a defensible answer, and it comes from asking which constraint filled the container. A forty foot high cube gives you around 68 usable cubic metres and a payload near 26,000 kg. This load used 63 of the cubic metres and 21,000 of the kilograms, so cube was binding and weight had slack. The cost was incurred to move volume, which makes volume the causal driver and cube allocation the closest thing to a true answer. Load a container of dense goods that hits the weight limit with cube to spare and the rule flips. Which means the allocation basis wants deciding per shipment by whichever constraint binds, rather than set once in a costing configuration screen and left there for eleven years.

Carriers already work this way. Ocean and air freight charge on the greater of actual and volumetric weight, a pricing rule that encodes exactly this logic. Your internal allocation can borrow it: allocate on chargeable weight, computed the way the carrier computed it, and the internal number tracks the external invoice for a reason.

The money that is inventory

The second large omission is the cost of the time a unit spends as your asset. This one is easy to compute and almost always left out because it lives in treasury rather than in logistics.

Take item A. Transit is 42 days, and average dwell in the network before sale is another 35 days. At a 14 percent cost of capital on a 12.00 unit, the in-transit carrying cost is 12.00 times 0.14 times 42 over 365, which is 0.193 a unit. The dwell adds 0.161. Together that is 0.354 a unit, larger than the entire difference between value allocation and cube allocation for this item.

The reason this matters beyond bookkeeping is that it changes sourcing arithmetic. Suppose a second supplier offers item A at 11.60, forty cents cheaper, on a 63 day transit rather than 42. Annual volume is 26,000 units, so the price saving is 10,400 a year. Against that, the extra 21 days adds about 1,500 units of pipeline stock, and the longer lead time raises the buffer as well. At 500 units a week with a demand standard deviation of 150, moving from a six week to a nine week lead time takes safety stock from 1.65 times 150 times the square root of 6, which is 606 units, to 1.65 times 150 times the square root of 9, which is 742 units. The network therefore carries about 1,636 more units at roughly 13.00 delivered, which is 21,270 of extra capital costing 2,978 a year.

The cheaper supplier still wins here, and the useful output is the crossover. Divide 2,978 by 26,000 and you get 0.115. Any price saving below about twelve cents a unit gets consumed entirely by the extra pipeline and buffer that the longer lead time forces. That figure belongs in the buyer's brief before the negotiation rather than in a review afterwards, and computing it takes five minutes.

The rate the cost was struck at

Where the purchase is in one currency and the sale is in another, a third definitional choice sits underneath everything above, and it moves the answer as far as allocation does.

Item A at 12.00 in the supplier's currency has a local cost that depends entirely on which rate you apply. At the budget rate of 33.5 it is 402.0. At the rate on the day the purchase order was raised, 34.2, it is 410.4. At the rate on the day of customs clearance, 35.8, it is 429.6. At the rate the supplier invoice was actually settled, 36.9, it is 442.8. At the hedge rate of 35.0 it is 420.0. That is a spread of just over 10 percent on one shipment, wider than the entire allocation argument.

The defensible practice is two figures with different jobs. For reported actuals, use the rate at which the liability was settled, because that is the cash that left. For forward decisions, use the hedge rate where the exposure is hedged and a planning rate where it is open, and state which. Costing everything at a budget rate held constant for a year is the version that causes damage, because it makes every item look equally profitable while the real cash cost of the import-heavy ones drifts away from the plan. W2 covers the timing decision that follows from all this.

What people do when they are measured on it

Give a buyer a delivered cost per unit target and watch what moves. The purchase price falls, because that is the component the buyer controls directly and the one that shows up immediately. Everything downstream of the invoice is somebody else's cost centre.

The specific failure is a supplier switch that saves 0.30 a unit at the factory gate on a carton 18 percent larger. Fewer units fit the container, per unit freight rises, and if freight is allocated by value rather than by cube the increase lands partly on other items in the same container, which makes it invisible on the item that caused it. That is the predictable consequence of allocating cost on a basis unrelated to what causes it while measuring people on the result.

The second behaviour is timing. Where delivered cost is computed on a standard and variances go to a purchase price variance account, a buyer under pressure at quarter end can pull forward a container to capture a price break. The delivered cost per unit improves, the working capital position worsens, and the two numbers sit with different owners.

What to pair it with

Delivered cost per unit gets gamed by anyone who can move cost out of the item and into a period. Three pairings close most of that.

Pair it with a lead time, always reported together. The cheapest delivered cost in any category is a slow boat, and a delivered cost quoted without the transit time that produced it is half a sentence.

Pair it with the working capital consequence, expressed as the pipeline plus buffer units the sourcing choice implies. The crossover arithmetic above turns that into one figure a buyer can carry into a negotiation.

Pair it with a service measure at the point the unit is delivered, because the last cost anyone counts is the expedite that a thin buffer forces. Air freight recovery on 8 percent of volume will erase a 5 percent saving on the other 92 percent, and it lands in a different account.

Where this stops

No allocation is true. Allocation distributes a cost incurred jointly, and the moment you ask what changes if this item goes away, the allocation stops being useful. Drop item A from the container and the ship still sails, the clearance still happens, and most of that 6,300 reappears on item B. Fully absorbed delivered cost is the right instrument for a pricing floor or a standard cost, and the wrong instrument for a delisting decision, where the question is incremental and the answer is usually that far less cost disappears than the costing model implies.

The same limit applies at customer level, where the allocation choices multiply and the incremental question gets harder still. That belongs to cost to serve analysis (W4), and the honest position is that a delivered cost model built for standard costing will mislead you if you point it at a customer profitability question without rebuilding the drivers.

There is also a data limit worth naming. Surcharges, demurrage and duty adjustments arrive weeks after the goods, so a delivered cost computed at receipt is provisional. Businesses that recompute at month end get a better number and a worse decision cycle, because the buyer has already placed the next order. Running a provisional figure at receipt and a true up quarterly is the usual compromise, and the size of the true up is itself a decent measure of how well the model is built.

Pull last quarter's inbound containers, compute the weight fill and the cube fill of each, and see which constraint actually bound before you defend your current allocation rule.

More on this in metrics on the DemandLabs platform.

See what DemandLabs could do for your plan

Request demo

More planning notes

View all

Nov 13, 2026

The same line ran at 37 percent and at 84 percent last week

A capital request for a second filling line reaches the investment committee with a chart showing utilisation at 84 percent and a note saying the site is running out of headroom. A finance analyst pulls a different report, which shows the same line at 37 percent, and asks why a business would buy a second one.

Nov 12, 2026

Your bias number is near zero and the warehouse still filled up

The quarterly pack has two numbers on the forecast slide. Accuracy at 78 percent, flat on last quarter. Bias at 0.2 percent, with a footnote saying it sits inside a tolerance of plus or minus two. Nobody asks about either one, because both look fine.

Nov 11, 2026

Service went up after go-live and so did four other things

The quarter three review has the chart on it. Line fill was 94.1 percent in the six months before go-live and 97.3 percent in the six months after. That is 3.2 points, roughly a million and a half of recovered margin on volume that would otherwise have been missed, and the programme is declared a success on the second slide.