Chargeable weight in practice: nine parcels, three divisors

The test design: nine parcels, three divisors, one rounding step

The run had two arms and one purpose. The purpose was to find out which of the inputs to a chargeable-weight calculation actually moves the figure, and the two arms separated the inputs that belong to the parcel from the inputs that belong to the line.

Arm A was nine footwear parcels measured twice: once as they arrived at the warehouse with the retail box in place, and once after the box was removed. Mass was taken on the same scale and the three outer dimensions with the same tape, so the only thing that changed between the two readings was the packaging.

Arm B was the wider sample of 88 parcels whose quotes were shared with this desk. Those parcels were not re-packed; they were re-calculated. Each one was run through the standard formula at every divisor in the set, and then again with the half-kilo rounding step applied, to see how often the two line-side inputs changed the answer on real parcels.

1. Arm A sample: nine footwear parcels, one warehouse, one scale, one tape, two readings each.
2. Arm A variables: packed volume and mass, before and after removal of the retail box.
3. Arm B sample: 88 parcels taken from quotes readers shared, re-calculated rather than re-measured.
4. Arm B variables: the four divisors, and the half-kilo rounding step, applied and not applied.

Parcel selection and why nine is the honest number

Nine is not a large sample and it was not chosen to be one. The nine were the footwear parcels that passed through the warehouse in one week with a packing request attached, which makes them a convenience sample rather than a random one. Saying so is part of the record, because a convenience sample of nine can show you a mechanism and cannot give you a rate.

The parcels were also similar to each other in a way that flatters the result: every one was a rigid box with a pair inside, every one came from the same category, and every one was built by the same small team. A run of nine mixed parcels would have been more informative and less convenient, which is exactly why it has not happened yet.

The wider arm exists to compensate for the small detailed sample. Re-calculating 88 existing quotes costs nothing and covers categories the nine do not, at the price of relying on figures the desk did not measure itself.

The weighing procedure, including the two mistakes we made

Each parcel was weighed on the warehouse scale with the carton sealed, and the three dimensions were taken at the widest point of each side with the tape flat against the carton. Volume was calculated from those three figures, and the volume weight from the divisor in play. Nothing about the procedure is unusual; the interest is in the two mistakes.

The first mistake was measuring a soft parcel while it was compressed under a stack. The tape read a smaller height than the parcel would hold once lifted, which produced a volume weight below the mass weight and briefly made the parcel look as though it was billed on mass. Re-measuring it upright moved the volume figure by enough to change which of the two drove the chargeable weight.

The second mistake was reading the tape at the carton flap rather than the body. A flap that has been taped shut sits proud of the side by a few millimetres, which is invisible on one dimension and compounds across three. The correction is mundane: measure at the body of the carton, at the widest point, with the tape flat rather than hooked over an edge.

1. Seal the carton before weighing, because an open flap changes both mass and shape.
2. Measure the carton body rather than the flap, and take each dimension at its widest point.
3. Never measure compressible goods under load; lift the parcel clear and let it recover first.
4. Record mass and volume on the same line, because a volume figure without a mass figure cannot be interpreted.

Results at divisors 5000, 6000, 7000 and 8000

The nine parcels were quoted at three divisors, 5000, 6000 and 7000, and 8000 was run as a control rather than as a mode any of them could realistically use. The result that matters is not a per-parcel table but a direction: the divisor is a line-side input, so it changes every parcel at once, while the packing change is a parcel-side input, so it changes only the parcels where volume was driving the figure.

That distinction explains the headline number from the run. Removing the retail box changed the chargeable figure on seven of the nine parcels, but it changed nothing at all on the two where mass was already the greater of the two weights, because shrinking volume cannot help a parcel that is billed on what it weighs.

The wider arm supplies the scale of the line-side effect. Across 88 sampled parcels, the divisor set and the half-kilo rounding step between them changed the chargeable figure on 31 of them. Two inputs that cost nothing to get right were decisive on more than a third of real parcels.

Where the half-kilo rounding step changed the answer

Rounding is the most under-read line in a chargeable-weight calculation because it happens after the interesting part. The step is applied to the greater of mass and volume weight, upward, so a parcel that calculates to 9.2 kilograms can be billed as 9.5 or as 10 depending on the step the line uses and on how the line applies it.

The arithmetic is worth doing once by hand. Take a carton measuring 60 by 40 by 30 centimetres. Its volume is 72 litres, which implies 14.4 kilograms at a divisor of 5000 and 12.0 at 6000. If the parcel weighs 9.0 kilograms, volume drives it in both cases, and the half-kilo step turns 14.4 into 14.5 while leaving 12.0 alone.

That example is chosen to be round, and it shows the mechanism rather than any measured parcel. The finding from the run is only that this second-order step is not second-order in effect: it participates in enough of the 31 changed figures to be worth asking about, and a line that cannot say which step it applies cannot be checked.

Two regimes are worth telling apart before any of this is applied. Where measured mass and volume weight sit close together, the rounding step decides the answer and the divisor hardly matters. Where they sit far apart, the divisor decides and the step is noise. A parcel can be assigned to one regime or the other in a single calculation, and that assignment tells you where to spend effort, because packing work and mode choice are not substitutes for one another.

1. Ask which divisor the line applied before asking anything about the rate.
2. Ask whether the step is applied to the chargeable figure or to each dimension of the volume weight, because the two differ.
3. Compare your own figure with the quoted figure at the same divisor and the same step, or the comparison means nothing.

Bias, drift and what a nine-parcel run cannot carry

Three limitations belong in the record. The nine parcels were selected by convenience, so they over-represent footwear and under-represent everything soft, bulky or fragile. The second arm inherits a different bias: quotes get shared when they surprise somebody, so a re-calculated sample of shared quotes is tilted towards the expensive end of the distribution.

There is also instrument drift to consider. One tape and one scale were used for the detailed arm, which removes disagreement between instruments but does not remove the possibility that both drifted together. A scale that reads 1 per cent high would move every parcel in the same direction and the run would never notice.

Finally, the run measured parcels, not lines. Nothing in it establishes what any particular line charges, which is the point: the record is about the calculation, and the calculation is the only part of the quote you can check without asking anybody.

One more limitation affects how these figures should be used. The nine parcels were measured at one warehouse in one week, and the wider arm was collected over months from quotes that arrived unsolicited. Neither arm is a random sample of what people ship, so the run describes a mechanism dependably and a distribution only roughly, and any percentage read from it should be treated as a count of parcels rather than as a rate.

What the run does not license you to conclude

The run does not support a claim about typical shipping cost, because it did not measure rates at all. It does not support a claim that a particular divisor is correct for a particular line; it supports only the statement that the divisor belongs to the mode and should be readable on the quote.

It also does not support the common conclusion that packing requests are pointless. Two of the nine parcels were unaffected because mass was driving them, and the same request applied to a soft parcel would have moved nothing for a different reason. What the record supports is narrower and more useful: know which input is driving your figure before you spend effort on the other one.

Three statements are supported by the run and nothing wider. The divisor belongs to the mode rather than to the parcel. The rounding step is applied after the calculation rather than inside it. And packing work moves only those parcels whose figure is driven by volume, which is a condition you can test in one calculation rather than guess at.

The data point behind this note

Own parcel test, KB-2026-02 and KB-2026-04: the nine-parcel arm re-measured nine footwear parcels before and after retail-box removal (7 of 9 changed figure); the wider arm re-ran 88 sampled parcels at the 5000/6000/7000/8000 divisors with the 0.5 kg step applied and not applied, changing the figure on 31.

◇KB-2026-02· 2026-09-30 Batch ledger · Challenge a figure

Notes in the same cluster

Related entries and gauges

• How is chargeable weight calculated on a consolidated parcel?
• Parcel weight bench

  Which of volume and mass is actually paying for the flight.