What an order really costs: getting contribution margin right in an online shop

Ask a shop owner what an order costs and you will usually get the product cost. It is the number on the supplier invoice, so it feels like the answer. It is roughly half of it.

What actually matters is contribution margin: what is left from a sale after every cost that scales with it. That is the cash each order contributes toward rent, salaries, software and profit. Two shops with identical revenue can have completely different bank balances, and this is the number that explains it.

Only variable costs belong here. Rent, salaries and software do not rise with the next order; they are fixed costs, covered by the margin your orders contribute, not subtracted from each one.

The stack, in full

Here is a real shop's order, with every line that scales. The €50 at the top is the average order value Shopify reports — already net of discounts, which is where most of the surprises start.

The rest of this article is those lines, one at a time: what each really is, and how to model it so it moves with your forecast instead of sitting there as a typed number.

Product cost

The easy one — and the only one most shops get right. Either take it per SKU and weight it by mix, or use a blended cost ratio: if your goods cost 30% of what you sell them for, product cost is revenue × 30% and it scales on its own.

The subtlety is returns. If eight orders in a hundred come back sellable, you only bear the product cost of ninety-two — which is why the line above is €13.80 rather than €15.00. If your returns come back unsellable (fashion in the wrong size is fine; opened cosmetics are not), that adjustment does not apply and the cost stays at €15.00. Know which you are.

Shipping and fulfilment

How you model this depends entirely on who packs the box.

If you ship it yourself, the cost is a label per order plus packaging: orders × average label cost. Find the real average by dividing last quarter's carrier invoices by the orders in the same quarter — not by looking at a price list, which will always be optimistic about weight and zones.

If a 3PL packs for you, the bill splits in two: a monthly base and storage fee that barely moves, and a pick-and-pack charge that scales per order. Model them separately, because only the second one belongs in contribution margin — the base fee is a fixed cost.

Payment fees

Individually tiny, annually real, and almost perfectly predictable because they move with revenue. The fast version is a flat rate times revenue, and it will drift, because you take money through more than one provider and they do not charge the same. Blend them instead, weighted by how much revenue flows through each:

Then the monthly line is revenue × blended rate, and it updates itself whenever the revenue forecast changes.

Returns

A return is the line most shops model as zero, and it is rarely small. It costs you three times: the revenue goes back, the outbound shipping is spent, and you pay for the label coming home. Whether you also lose the goods depends on your category.

Model it as one driver — a return rate — and let it act on revenue and on shipping. The rate is worth watching monthly: it moves with your product mix, with a sizing change, and sharply after a discount campaign.

Discounts

A discount does not come out of revenue evenly. It comes out of margin entirely — and because margin is the smaller number, the percentages are brutal.

A 20% code halves your margin. That can still be the right call — for clearing stock, for a first order that becomes a repeat customer — but it has to be a decision, not a habit, and it has to be made against the margin number rather than the revenue number.

Storage and holding

Stock that sits in a warehouse costs rent, insurance and handling, and it ties up cash you cannot spend on anything else. If a 3PL bills you for storage, that amount is already on your invoice. If you hold the stock yourself, take the warehouse's monthly cost, divide it by orders, and put it on the line — it is small per order and material per year.

The bigger cost is the cash itself. Money in boxes is not money in the bank, which is why a restock is a dated outflow in a forecast and not an afterthought — see planning a restock for how that lands on a specific day.

What the margin then decides

Once the stack is real, two things fall straight out of it.

Your ad ceiling. Break-even ROAS is 1 ÷ contribution margin ratio. At €24.70 on €50 invoiced, that is a 49% margin and a break-even ROAS of about 2.0×. If you had used the naive €29.40, you would have believed 1.7× was enough — and every campaign between the two numbers loses money on every order while looking profitable in the ad account.

Whether growth helps. A shop at 50% margin keeps real cash from each order and can fund its own growth. A shop at 8% needs perfect timing to stay level, and doubling its orders mostly doubles the size of the timing problem. Revenue tells you how big you are; contribution margin tells you whether being bigger is an improvement.

How this looks in Foreqast

Each of these is its own driver row under revenue, fed from the tool that knows it: order value and returns from Shopify, fee rates from your processors, shipping from your own average. Contribution margin is not a side calculation — it is what remains when the model subtracts them, and it moves the moment any driver does.