Fulfilment is charged by the parcel and reported by the order, and that single mismatch hides one of the most avoidable costs in ecommerce. An order that leaves in two boxes pays twice for the label, twice for the picking, twice for the box, filler and tape and, if the carrier prices by bracket, twice for whatever band each box falls into. On a store running 1,08 parcels per order that is eight percent on top of every per-order fulfilment cost before anything goes wrong, and none of it appears in a report that counts orders rather than parcels.
- Every parcel carries its own label, picking and packaging; an order is not a unit of cost.
- A per-order average hides splits completely, which is why most stores do not know their own rate.
- Splits are caused by stock location and stock levels far more often than by product size.
- The customer usually experiences it as a problem too, which makes the cost bigger than the invoice.
Why the average hides it
Divide a month of fulfilment spend by a month of orders and you get a figure that is true and useless: it silently blends the orders that shipped in one box with the ones that shipped in three, and it will never tell you which. The number that means something is the parcel count, and the ratio between the two is the whole finding.
Once a store knows its own ratio, the arithmetic is immediate. Multiply your true cost per order by the ratio and the difference is what splits cost every month, which is usually the first time anybody has seen the figure written down.
What actually causes them
Rarely the obvious thing. Oversized products that genuinely cannot travel together are a small minority; the usual causes are a back-ordered line holding up an otherwise complete order, stock split across two locations, and a picking rule that releases what is available rather than waiting. All three are inventory decisions wearing a logistics costume.
That is also why the fix usually lives upstream. Holding the fast-moving lines in one location, and holding enough of them, removes most splits at source, which is one of the quieter arguments for getting stock coverage and reorder points right.
| One parcel | Two parcels | |
|---|---|---|
| Label and carriage | once | twice |
| Picking and packing | once | twice |
| Box, filler, tape | once | twice |
| Packaging fees by weight | one composition | two |
| Customer experience | one delivery | two, and one query |
The carrier's view of two parcels
Splitting also loses whatever benefit a heavier single parcel would have earned. Rate cards step by weight, and two light parcels frequently cost more than one heavier one that stays inside its band, so the split is charged twice at the least efficient point of the card rather than once at a better one.
Where a carrier applies a per-consignment surcharge, remote area fee or fuel percentage, those repeat as well. None of it is dramatic on a single order and all of it compounds across a year of a stable split rate.
The cost nobody invoices
Two deliveries means two chances for something to go missing, two tracking numbers to explain and a returns process that has to be described twice. Support tickets are the cheapest of those and the most measurable: a store that starts counting contacts per split order usually finds the ratio uncomfortable.
Measuring and reducing it
- Get the parcel count from your carrier or warehouse, not from your order count, and divide.
- Price the gap: your true fulfilment cost per order times the ratio, minus the same cost at one parcel.
- Split the rate by cause, back-order against multi-location against picking rule, because the fixes differ.
- Check whether the products involved are always the same handful; if so, this is a stock problem with a name.
- Re-measure after any change to stock levels or picking rules, since the ratio moves quietly in both directions.
What reducing them is worth
Take a store shipping 706 orders a month at 1,08 parcels per order. That is 56 extra parcels, and at even five euros of all-in fulfilment per parcel it is €280 a month, or €3.360 a year, spent on boxes nobody ordered. Halving the split rate recovers half of that without touching a price, a supplier or an ad account.
The recovery is also permanent in a way promotions are not. A split rate improved by better stock placement stays improved until something changes the placement, which makes it one of the few operational fixes that keeps paying after the person who made it has moved on to something else.
When a split is the right answer
Sometimes it is. A customer waiting three weeks for one back-ordered item while the rest of their order sits in a warehouse is a worse outcome than two parcels, and a heavy item shipping separately on a cheaper freight service can genuinely cost less than forcing everything into one oversized box. The point is not to eliminate splits; it is to know the rate, know the cost, and split on purpose rather than by default.