How to Calculate SKU Profitability

Every catalogue contains products that lose money, and almost every store is surprised by which ones.

The reason is structural. Gross margin is easy to calculate and gets calculated. Contribution margin per unit requires allocating costs that were never incurred per unit, so it gets skipped — and the gap between the two is where the losing products hide.

Why gross margin ranks the catalogue wrongly

Gross margin per product is straightforward: price minus cost of goods. It is also blind to everything about the product that affects what it costs to sell.

A lightweight accessory and a heavy homeware item can both carry a 63% gross margin. Ship them and the resemblance ends. The accessory adds almost nothing to parcel weight and rides along inside a box that was already going. The homeware item drives the box size, pushes the parcel into a higher dimensional band, and attracts an oversize surcharge on remote deliveries.

By the time both have paid for their own delivery, one is at 55% contribution margin and the other is at 18%. Nothing in the gross margin column suggested a difference.

The allocation problem, stated honestly

Shipping, packaging and payment fees are incurred per order, not per unit. Pushing them down to SKU level requires an allocation, and every allocation is a judgement.

Three bases are defensible, and they are not interchangeable:

Even split per unit. Simple, and wrong in the specific way that matters. It charges every unit the same delivery cost regardless of what it contributed to the parcel, which systematically flatters heavy products.

By weight or volume share. The best default for shipping and packaging. A unit that is 60% of a parcel’s billable weight carries 60% of the label.

By revenue share. Correct for payment fees, which genuinely do scale with order value, and wrong for shipping, which does not care what anything cost.

The important discipline is not picking the perfect basis. It is picking one, writing it down, and using it consistently — because the value of these numbers is comparative. A ranking produced by a consistent method is useful even if any individual figure carries an error bar.

Variant level, not product level

Run this on variants rather than parent products.

Sizes differ in cost and in shipping weight. Colours can differ in supplier price and in return rate. A parent product showing a healthy blended margin routinely contains one variant that loses money on every unit and another that carries the whole line.

Apparel is the clearest case: the largest sizes often cost more to make, weigh more to ship, and return at higher rates. Averaged into a parent product, all three effects disappear.

Two numbers, not one

Contribution margin per unit answers whether the item earns its place. It does not tell you how much the store depends on it.

Multiply by units sold and you get total margin contribution — how much of the store’s contribution margin this SKU is responsible for. The two together drive different decisions.

High margin per unit, low total contribution: a good product nobody knows about. A merchandising and promotion problem.

Low margin per unit, high total contribution: the product carrying your revenue at a rate you cannot really afford. Usually the most dangerous position on the list, because volume disguises it and any correction touches the store’s biggest line.

High on both: protect the stock position. This is the product you cannot afford to run out of.

Low on both: discontinue, unless basket effects justify it.

Basket effects, and when not to cut

The one thing per-unit analysis cannot see is what a product does for the orders it appears in.

An accessory with thin margin that appears in a third of orders may be lifting average order value across the whole store. Removing it improves the SKU-level table and reduces total margin — a decision that looks correct in the spreadsheet that prompted it and is wrong in the accounts three months later.

Before cutting a weak SKU, check two things: what proportion of its units ship alone, and what the average order containing it is worth compared to the average order without it. If it rarely ships alone and the orders containing it are larger, it is doing a job the per-unit figure was never going to show.

Where the numbers usually break

Missing landed cost. Supplier price recorded, inbound freight and duty ignored. On imported goods this understates cost materially, and it does so unevenly across the catalogue.

Bundles. A bundle SKU with no cost of its own reports as pure margin. Every bundle must explode into components before any of this works.

List price instead of realised price. Any product regularly discounted sells below list, and the difference lands entirely in the margin figure. This one compounds, because the products that get discounted most are usually the ones already under margin pressure.

Stale costs. A supplier price rise applied retroactively recosts historical orders and changes last quarter’s rankings. Cost changes need effective dates.

Returns left out. A product returning at twenty percent carries the full cost of those returns, and none of it appears in a per-unit calculation built from sales alone. On apparel and anything with fit or colour expectations, this is frequently the largest single correction — and it moves the ranking most on the products that look best before it is applied.