Purchasing

Your supplier sells cases. Your forecast counts units.

The number you need is 1,340 and the case holds 24, so the real order is 1,344 or 1,320, and somebody is doing that arithmetic in their head fifty lines per PO. Here is how to do it on purpose. The numbers are invented so every line can be checked.

Demand usually happens in units. A direct-to-consumer customer buys one, or three, rather than a case — though wholesale and B2B buyers order cases routinely, and some products only sell as multipacks. The practical point is about the calculation, not the customer: keep the planning maths in whatever base selling unit your demand actually arrives in, and let the packaging constraint apply afterwards. So the order quantity starts life in units: projected demand over the window, minus what you have, plus a buffer.

Then the purchase order has to be written, and the supplier does not ship 1,340 of anything. They ship cases of 24, or inners of 6 in masters of 48, or whatever their line packs. The unit number has to become a case number, and that conversion is where quantities quietly drift: a few units up on this line, a few down on that one, a mental rounding that is not written down and cannot be reconstructed a month later when the stock does not match the plan.

Three steps make the conversion deliberate. One invented PO, six lines, carried through all three.

Step 1: build the order in units, and keep it in units

The raw need per line is the output of your reorder maths, before any packaging constraint touches it. Here is the invented PO.

Keep this column. It is the only number on the PO that says what you actually need, and once the rounding happens you will want to be able to see how far each line moved from it. If the raw need is overwritten by the rounded quantity, that information is gone.

In SKU Compass the two numbers sit in adjacent columns so neither overwrites the other. When you build a PO from selected SKUs, the Create Purchase Order dialog lists each line with a Need column, described on the page as the raw Need to Order from the Order Estimator before case-pack rounding, a Case column showing units per case, and a separate Qty column, described as the quantity to place on the PO, case-pack rounded by default. Need is what you want. Qty is what you are sending. The gap between them is the rounding, and it is visible per line.

Step 2: round to cases, and choose the direction on purpose

Cases needed is raw need divided by case pack. It is almost never a whole number, so you round, and the direction matters more than it looks.

Rounding up guarantees you cover the need and buys some extra. The extra costs money now and warehouse space until it sells. Rounding down saves that money and leaves you short of the need by up to one case. Which is right depends on what the extra units mean for that SKU, and the way to see it is to convert the extra units into days at that SKU’s rate.

Read the last two columns together: extra cost tells you what the rounding consumes in cash; extra days of cover tells you whether that cash is buying hours or months. On SKU A, rounding up adds four units and about two hours of cover — usually not worth a second thought, though unit cost, perishability, hazmat storage or sheer physical size can make even a small quantity matter. On SKU D, rounding up adds 31 units, which is two months of cover for a SKU that sells one unit every two days, and $186 that will sit on a shelf into next quarter. Same operation, same direction, completely different decision.

The rule that falls out is not “always round up” or “always round down.” It is: round up by default, because on a fast mover the extra units are hours of cover and a shortfall is a stockout, and then look hard at every line where the extra cover is measured in weeks rather than hours. Those lines are slow movers, and slow movers are where rounding up hides cash.

For SKU D the honest options are rounding down to zero cases and ordering next cycle when the need has grown, or accepting the case and treating the extra 31 units as a deliberate buffer. Either is defensible. What is not defensible is not noticing.

In SKU Compass the default matches that rule. The Qty column arrives case-pack rounded, and a Round all to cases button on the dialog does the same for every line at once; its own label says it rounds up to the nearest multiple of the case pack. That handles SKU A, B, E and F without thought. It also rounds SKU D up to a full case, which is exactly the line you should be looking at, and the Need column beside it is what makes the 17-versus-48 gap visible instead of buried.

Rounding is not arithmetic. It is a purchasing decision per line, and the case pack only tells you the size of the step, not which way to take it.

Step 3: send the supplier the number they can ship

Once every line is in whole cases, the PO is in the supplier’s units and two more things become checkable. The extended cost per line is quantity times unit cost, and the PO total is the sum. Both should be computed on the rounded quantity, not the raw need, because the rounded quantity is what you are paying for.

Invented PO, rounded up throughout:

Against a raw need of 2,912 units, the rounded PO is 2,988: 76 extra units and $392.10 of extra cost, of which $186 is one slow-moving SKU. That is the number to know before you sign, and it is a number you can only see if the raw need survived to this step.

In SKU Compass the dialog carries Unit Cost and Extended per line and a running total above the lines, so the rounding’s cost is on the same screen as the rounding, and Create Draft PO is the dialog’s final button. The supplier gets whole cases; the Need column beside them is the record of what you actually needed.

Case pack, MOQ and pallet are three different constraints

They get lumped together as “the supplier’s minimums” and they behave differently. Apply them in this order, after the unit forecast and never inside it:

1. Raw unit need — projected demand minus what you have, plus a buffer.
2. Apply the MOQ floor — raise the quantity if the supplier will not accept less.
3. Round to a valid case multiple — up or down, decided per line.
4. Apply pallet and freight constraints — on the whole PO, not the line.

The MOQ goes before the case rounding, and that ordering is not cosmetic. Round to cases first and the MOQ can push you back off the case grid: you need 17 units, the case pack is 48 and the MOQ is 200. Case-rounding gives 48; applying the MOQ then gives 200 — but 200 is not divisible by 48, so it is not a quantity the supplier can ship, and you have to round a second time to 240. Floor to the MOQ first and the single case-rounding pass lands on 240 directly, with no invalid number in between.

Case pack

The unit of shipment. Quantities must be a multiple of it. It rounds you by at most one case, so its cost is bounded and small on fast movers, large in days on slow ones, as above.

Minimum order quantity

The smallest order the supplier will accept per line or per PO, in units or in money. It does not round; it floors. That per-line versus per-PO distinction matters for the sequence above: a line-level unit MOQ slots straight into step 2, but a supplier-wide dollar minimum is a property of the whole order and cannot be applied SKU by SKU. Check it once the lines exist, and fix it by adding or enlarging lines you actually want. If the MOQ on SKU D is 200 units and you need 17, the case pack question is irrelevant, because the real question is whether you want 200 units, which is over a year of cover at half a unit a day. MOQs are where slow SKUs become inventory problems, and the decision is a purchasing one about whether to carry the SKU at all, not a rounding one.

Pallet and container quantities

Two different things wear this label, and only one of them is a constraint. A required pallet or container multiple is a genuine purchasing rule — the supplier or the mode will not take anything else, and it behaves like a coarser case pack. A freight break is an economic threshold: a part-pallet can ship at the full-pallet rate, so filling it is often the cheaper choice per unit, but nothing stops you shipping short. Treat the first as a hard rounding step and the second as a cost comparison you can decline. Either way, evaluate it on the whole PO after the lines are in cases, because it is the total that fills the pallet.

The order matters. A forecast that already has the MOQ baked into it cannot tell you the MOQ is the problem, and a line rounded to cases before the MOQ check can land on a quantity the supplier cannot ship. Units, then minimums, then cases, then freight, each visible as its own step.

Where rounding interacts with the reorder point

Rounding changes how much you order. It does not change when. The reorder point, daily demand times lead time plus safety stock, is the stock level at which the order must be placed, and it is in units regardless of how the supplier ships. What rounding up does is push the next order slightly further out, because you land with a few extra units; rounding down pulls it slightly in. On a fast mover the shift is hours. On a slow mover with a big case it can be a whole cycle, which is another way of saying the same thing the days-of-cover column said: on slow SKUs, the case pack is a bigger planning input than the forecast. Put a slow SKU’s rate and lead time into a reorder point calculator and compare the answer to one case; if one case covers the reorder point several times over, the case pack is the plan.

What to do this week

1. Keep the raw unit need on the PO next to the rounded quantity. If your tool overwrites one with the other, add a column. In SKU Compass they are the Need and Qty columns on the Create Purchase Order dialog.
2. Round up by default, then convert the extra units on every line into days of cover. Any line where the extra is measured in weeks is a decision, not a rounding. In SKU Compass, Round all to cases does the default pass.
3. Apply MOQ and pallet after the case rounding, as separate checks, and read the PO total against the raw need total before you send it. The difference is the cost of the supplier’s packaging, and you should know it.

Two units of measure, one purchase order. The rounding is where quantities drift, and the fix is not cleverer rounding. It is refusing to let the unit number disappear.