Every order has two numbers. How many. And how many of each.
The first gets debated for weeks. The second gets decided in an afternoon, usually by copying the last order. Then the mediums sell through in a month and the XLs ride out the season on the clearance rail.
A size curve is the percentage split of a buy across sizes. Get the total right and the curve wrong, and you still have the wrong inventory.
A flat run is a guess with good manners.
Two of every size feels fair. It is not a plan.
No membership, no corporate list and no retail customer is shaped like a flat run. Bodies cluster. The middle sizes carry the order. The edges are thin.
A flat run guarantees two things at once. Stockouts in the core sizes. Markdowns on the ends.
Both cost margin. Only one shows up as a sold-out badge.
Start from the people, not the pattern.
The right curve depends on who is wearing the garment.
A men's member-guest skews large. M and L carry most of the field, with real weight in XL and a tail into XXL.
A women's club program centers on S and M. XS sells faster than the numbers suggest. The XS customer has fewer options elsewhere and buys when she finds one.
A corporate gifting list is the easiest and the most ignored. You have names. Ask for sizes before you cut. A size survey costs a week and saves a reorder.
A retail capsule has the most data and uses it least. Last season's size-level sell-through is the curve. Not last season's buy. What sold, by size, at full price.
The fit changes the curve.
The same customer is not the same size in every garment.
A slim knit polo pulls the curve up half a size. Customers who wear a medium in a relaxed fit take a large when the body skims.
A boxy layer pulls it down. Outerwear and half-zips sell more small and medium than the polo from the same program.
Stretch widens the tolerance. A rib with spandex covers more bodies per size than a woven with none. A rigid woven punishes a thin curve.
Women's bottoms run on a different axis entirely. Waist and length both matter. A skirt offered in a regular and a tall length is two curves, not one.
Set the curve per style. Not per program.
Small runs break the math.
On a 400-unit buy, a curve is percentages. On a 48-unit buy, it is whole garments, and rounding decides the outcome.
Twelve percent of 48 is 5.76. Round down and the XL customer is already short. Round up across every size and the total no longer matches the minimum.
Set a floor first. Every size you choose to offer gets at least two or three units. Then distribute the rest by the curve.
If a size cannot carry its floor, drop it from the run and plan a special order for the two people who need it. Offering a size in a quantity of one is not offering it. It is a stockout scheduled in advance.
Minimums live per color, not per style.
Mills and factories quote minimums by color, and sometimes by size within a color. A 300-piece style in three colorways is three 100-piece runs. Each carries its own curve.
The navy will sell differently from the white. Darker, safer colors pull larger. Lighter colors and prints pull smaller.
Split the curve by colorway when the history supports it. When it does not, hold the bulk of the units in the core color and run the fashion colors on a tighter, center-weighted curve.
Plan for the reorder before the first order ships.
A curve is a forecast. It will be wrong somewhere.
Build the reorder path into the program. Hold greige or undyed fabric where the mill allows it. Keep the size set and the PP standard on file. Know the reorder lead time before the first unit sells, not after the medium is gone.
A program that can reorder the core sizes in six weeks can afford a leaner first buy. A program that cannot has to bet on the curve up front.
Read the curve after it sells.
The last step is the one that gets skipped. Pull sell-through by size at the end of the season. Compare it to the buy.
Where a size sold out in week three, the curve was short. Where a size went to markdown, the curve was long. Write the correction down and carry it into the next tech pack.
Two seasons of honest size data beat any industry average.
The curve is not a detail under the order. It is the order, split into the only units a customer can actually buy.