Two buyers, one parcel: splitting costs fairly

The case is one parcel assembled by two buyers and one argument that lasted longer than the shipping did. Buyer A contributed two pairs of sneakers. Buyer B contributed a hoodie and three small accessories. Goods came to 6.4 kilograms in total, packaging added 1.2 kilograms, and the parcel billed at 7.6 kilograms chargeable. The agreement, reached in a chat thread the week before ordering, was one sentence long: shipping split by weight.

That sentence survived contact with reality for about ten days. The shoe boxes came off at the warehouse at A's request, which cut mass that A had been carrying in the split. The service line turned out to scale with the order subtotal rather than with the head count. Domestic freight arrived as four separate legs from four sellers, three of them A's and one of B's, and none of them had been mentioned in the agreement because neither buyer thought of domestic freight as shipping.

By the time the final bill landed the two buyers were working from two different mental models. A was computing shares from item weights taken off the seller listings. B was computing shares from the number of order lines, on the reasoning that a hoodie and three accessories is more lines than two pairs of sneakers and therefore more work. Both models were internally consistent and neither matched the bill, so the conversation became about the models rather than about the money.

When both models were finally written on one page, the gap between them turned out to be smaller than the argument they caused. Each side had been defending a method whose output differed by less than the currency conversion on the international line, while the two decisions that actually moved the bill went undiscussed: the box removal, which changed what the parcel weighed, and who carried the risk while the payer fronted the whole amount on one card. That is the pattern worth taking from this case. Arguments about method are usually cheap to resolve and expensive to have, and the money hides somewhere neither method looks.

A's error was treating a self-serving change as a shared saving. Removing the boxes reduced the parcel's chargeable weight, which reduced everyone's share, so A assumed the benefit should be shared and the decision should therefore be unilateral. The flaw is that the mass removed belonged to A's items, so the correct accounting is that A's share fell and B's share stayed roughly where it was. A also assumed "weight" meant the weight listed by the seller, which is goods only and excludes the 1.2 kilograms of packaging that everybody pays for.

B's error was treating line count as a proxy for cost. Order lines generate work at the desk, but the fee lines they generate are attributed to whoever owns the line, not spread across the parcel. A four-line contribution does not entitle you to a quarter of the international freight if your goods are a sixth of the chargeable weight. B also assumed the service line was charged per head, when in fact it scales with the order subtotal, which means it should be attributed per buyer's own subtotal.

The shared error was sequence. The rule was agreed before ordering and never revisited at the three points where it could have been applied cheaply: when the boxes were removed, when the freight was quoted, and when the final bill landed. A settlement rule that is only read once is not a rule, it is a hope with a timestamp.

There was a fourth error, and it was the most expensive of the four because it was invisible. Neither buyer distinguished paying from carrying the risk. One account fronted the entire amount, which meant that account absorbed the exchange-rate movement between the quote and the settlement, and it also absorbed the possibility that the other side would settle late or not at all. In a two-person order between people who trust each other, that exposure is usually waved away and occasionally realised. Pricing it is simple: settle the estimate before the payment is made, so the payer is never lending money as well as time.

Three buckets settle a combined parcel without argument. The attributed bucket contains everything that has an owner: each buyer's item cost, each buyer's domestic legs, and any handling line raised for a specific line or item. Nothing in this bucket is divided. It is the largest bucket in most hauls and the one that removes the most disagreement, because most disputes are about attribution rather than about arithmetic.

The prorated bucket contains the one genuinely shared cost, which is the international freight on the consolidated parcel. Divide it by chargeable weight share, measured from the final packed weight rather than from the seller listings, and round each share up to the next hundred grams so the ledger balances in one direction rather than oscillating. Rounding up means the collected total slightly exceeds the bill, and the surplus belongs in the third bucket rather than in anyone's pocket.

The reserved bucket is a hold-back against the things nobody can cost in advance: a late handling line, a currency conversion that settles off the estimate, or a customs charge assessed on the whole parcel. Our rule is one tenth of the estimated total, held by the payer until the final bill lands and then returned in proportion to what each side actually paid. The tenth is a rule we apply, not a measured rate, and its size matters less than the fact that it exists before the argument starts.

Run the three buckets against the actual numbers and the rounding rule shows its work. Goods came to 6.4 kilograms, of which one side's two pairs of sneakers were 5.2. Packaging added 1.2 kilograms, which belongs to nobody in particular and is therefore shared in proportion to goods, giving that side 6.18 kilograms and the other 1.43. Rounding each up to the next hundred grams produces 6.2 and 1.5, which totals 7.7 against a parcel billed at 7.6. The extra hundred grams is not a mistake; it is the mechanism by which the ledger balances without anyone owing a fraction of a kilo.

One payer, one invoice, one private ledger. The payer is the account holder, and the ledger is a shared document with three columns: attributed, prorated, reserved. Every line in the final bill gets assigned to a column before it is paid, and any line that cannot be assigned is a line the two buyers have not discussed yet. Doing that assignment on the day the freight quote arrives, rather than on the day the bill arrives, is what converts the rule from documentation into practice.

Then five operating rules. Weight shares are computed from the final packed weight, never from listed item weights. A change requested by one side is charged to that side, including any handling line it generates, while the weight effect falls where the mass was. Rounding always goes up into the reserve. No international payment is made until every side has funded its estimated share, because funding after the fact is where friendships go. And any charge assessed on the parcel as a whole, such as a clearance line, is prorated by weight share, because it genuinely could not be attributed.

The last rule is about disagreement. Where the two sides cannot agree whether a cost was shared or attributable, the default is attribution, on the reasoning that an unattributed shared cost grows every time it is invoked. That default is arbitrary and it is also written down, which is the only property that matters: a rule that both sides read before the parcel ships will survive a bill, and a rule invented after the bill will not.

For a parcel that has already been paid, there is a recovery protocol worth having in advance. Take the final bill as the only source of truth and ignore the estimates, apply the three buckets retrospectively to the lines on it, and circulate the resulting ledger to both sides before discussing it. Then settle the difference and close the row, even where the difference is trivial. The failure mode to avoid is carrying an unsettled parcel into the next one, because the second order inherits the first argument and neither side can then tell which cost belongs to which decision. Two open rows make a settlement impossible to reason about.