Some of your best sellers are losing money.
A style can hold 75% gross margin and still cost you money on every sale, once you count what you paid to acquire the customer, what came back, and what it cost to ship both ways. The P&L never shows it, because the loss is spread across marketing and logistics lines nobody attributes to product.
Gross margin
75.0%
What the P&L shows
Contribution
−$6.40
Per unit kept
Where the margin went
Who acts
Exclude from feeds — break-even is $14 acquisition, you are paying $28
Do not reorder. 72% sell-through, negative contribution
Bundle above $80 or reprice
Approve · defer · override
$180/week bleeding
Gross margin was never the number that mattered.
It was the only one you could get. Every system holds a piece of the true cost, and none of them holds all of it.
The cost sits in four places
Product cost is in your range tool. Acquisition is in your ad accounts. Fulfilment is in a 3PL invoice. Returns are in the returns platform. Nothing joins them at style level, so nobody can say what a garment actually earned.
Returns are counted as revenue, not cost
A refund shows up as revenue lost. What it doesn't show is the outbound freight you already paid, the return leg, the handling, and the payment fee you never got back. That lands in operating costs, attributed to nothing.
So the wrong styles get repeated
Sell-through says the style is working. Gross margin says it earns 75%. Every signal you can see says reorder — and the one signal that says don't is the one nothing produces.
What it computes
Contribution per unit kept. Not per unit sold.
A unit that came back isn't a sale. It's two freight legs, a handling cost and a payment fee you didn't recover. Contribution divides by what the customer actually kept.
Loss-making at full price
Contribution below zero on undiscounted sales. Not a markdown problem — the style never earned its cost.
Acquisition-dependent
Profitable organically, loss-making with paid attributed. The style works; the bid doesn't.
Returns erosion
Gross margin above target, contribution below it, and returns the whole difference.
Discount-dependent
Contribution positive at full price only — on a style that almost never sells at full price.
Category exposure
A whole category below the portfolio average, which is a range conversation rather than a style one.
Price architecture
An entire price band loss-making after acquisition and returns. The answer isn't a markdown, it's the range.
One finding, four owners
A loss-making style isn't only the buyer's problem.
It is the only agent in the platform whose recommendations leave the buying team — because the fastest fix is usually to stop paying to sell it.
Digital and paid
Exclude the style from your product feeds, or cap the bid at break-even. The recommendation carries the acquisition cost at which it turns profitable, not just a stop.
Buying
Gate the reorder. Sell-through and margin both say repeat; contribution says the repeat loses money.
Trading
Reprice, bundle above break-even, or move it behind a minimum spend.
The board
Contribution by price band and category. If everything under $80 loses money, that is a range decision.
Constrained by design
It will not guess at a cost.
A style with no cost held produces no contribution figure. Below a meaningful volume of units kept, none is asserted. Where landed cost is derived rather than measured, the basis is recorded against the style so a weaker figure can be told from a stronger one.
Every cost assumption is stated on the output, and the report says who confirmed them.
This number gets used to pull ad spend and cancel repeats. A missing figure is a question. A wrong one is an expensive decision.
Across the suite
True Margin changes what the other agents recommend.
True Margin