What is Contribution margin?

A A$120 order: A$54 cost of goods, A$11 fulfilment and shipping, A$2.40 in payment fees, and A$6 of expected returns cost across the category. Variable costs come to A$73.40, so contribution is A$46.60, a contribution margin of 38.8%. The headline gross margin on the same order is 55%. An acquisition ceiling set from 55% would be A$66 against a true A$46.60 — it credits every order with A$19.40 it does not have, 42% more than it actually leaves.

Contribution margin is what is left of a sale after every cost that scales with it. It answers the question that gross margin only approximates: how much money did this order generate that can be used to pay for anything else.

The distinction sounds academic until the two numbers are put side by side. Gross margin subtracts the cost of goods. Contribution margin subtracts the goods, the shipping out, the shipping back on returns, the packaging, the payment processing, the marketplace commission, and the handling cost of the returns themselves. In a physical-goods business the gap between the two is routinely substantial.

That gap matters because every acquisition decision is made against one of these numbers, and using the wrong one makes every campaign look about a sixth better than it is.

The test is counterfactual: would this cost exist if the order had not been placed.

In. Cost of goods at landed cost, including inbound freight and duty where they are allocated per unit. Outbound shipping and fulfilment, including the labour if it is paid per order. Packaging. Payment processing and gateway fees. Platform or marketplace commission. Expected returns, modelled as the revenue lost plus the cost of getting the item back and dealing with it. Any per-order subsidy, including free shipping and free gifts.

Out. Rent, salaries, software, the advertising team, warehouse leases and anything else that continues regardless. These are covered by the contribution rather than subtracted inside it.

Advertising itself is the ambiguous case. It is variable in the sense that it stops when spending stops, and the convention that makes the metric useful for advertising decisions is to leave it out — so that contribution is the pool the advertising is bidding against rather than a figure that already has the answer inside it.

Of the variable costs, returns are the most consistently modelled badly, and the error is usually in the same direction.

A 25% return rate does not reduce contribution by 25%. The returned order loses its revenue, and it also costs the outbound shipping already paid, the inbound shipping, the handling and inspection labour, and in many categories a share of the goods that come back unsellable. The effective cost of a return is frequently more than the contribution the sale would have made, which means a high-return category can be contributing less than the reports suggest even at a healthy gross margin.

Returns also vary by category far more than most other variable costs, which is the reason a store-wide rate is a poor input. Apparel and hard goods are not in the same business, and averaging them misprices both.

The chain is short and it is worth stating in full, because it is the reasoning behind every acquisition target in an account.

A sale produces some contribution. That contribution is the only money available to pay for the advertising that produced the sale, the fixed costs of the business, and profit. Therefore the most an order can cost to acquire, before that particular transaction loses money, is its contribution. And the most it should cost is materially less, because two of those three claims on the money have not been paid yet.

Where customers return, the same logic applies over the relationship instead of the transaction, which raises the ceiling by however much repeat contribution can be relied on. That is the entire mechanism behind businesses that appear to acquire customers unprofitably and are not, and the discipline it requires is a retention figure somebody is willing to defend.

One contribution margin for a whole catalogue produces one break-even threshold for products that break even at very different points.

A store selling own-brand goods at 55% contribution alongside resold third-party goods at 18% has a blended figure that describes neither. Managed to the blend, the account over-invests in the thin range — where the true break-even is much higher than the target — and under-invests in the profitable range, where it could afford to pay considerably more and is not.

Two resolutions work. The structural one is to split campaigns by margin band and give each its own target, which is straightforward in search and shopping and clumsy in feed-based campaigns. The better one is to get product-level cost into the conversion value so that the platform's own bidder weighs each order by its contribution, which makes the split unnecessary and is the reason profit-valued bidding is worth the engineering.

Contribution margin decays quietly. Every input to it moves without anyone informing the advertising team.

Supplier prices change. Freight rates change, sometimes sharply. Currency moves against a business importing goods. Payment processing tiers step at volume thresholds. A change to the returns policy shifts the returns cost within weeks. Product mix shifts on its own as campaigns find what sells.

A margin model built once and never revisited will still produce a confident number a year later, and every threshold derived from it will be wrong in whichever direction costs moved. The remedy is ownership: somebody in finance or operations owns the inputs, and there is a route from them to whoever owns the bids. In most companies that route does not exist, which is why acquisition targets are so often defended on a margin figure nobody has checked.

Contribution margin is not a metric to report weekly. It is the input to the ones that are.

Break-even return on ad spend is its reciprocal. Profit on ad spend is the version of return on ad spend that has contribution already inside it. The acquisition ceiling and every target cost per acquisition come out of it. Lifetime value is contribution repeated across the relationship rather than revenue repeated across it.

Get this number wrong and every one of those is wrong by the same proportion, in the same direction, and none of them will show any sign of it.

• List the variable costs explicitly and get each one from a system rather than from memory. The ones that get forgotten — payment fees, returns processing, packaging — are individually small and collectively worth several points of margin.
• Compute it per product or per category, not store-wide. A blended figure over-values the thin part of the catalogue and under-values the profitable part, and the advertising will act on the error.
• Model returns as lost revenue plus a handling cost, not as a discount. A returned order costs the outbound shipping, the inbound shipping and the handling, and sometimes the goods.
• Treat a shipping subsidy as a margin decision and price it. Free shipping over a threshold is one of the largest variable costs in most stores and one of the least often costed.
• Recompute when freight, supplier pricing or currency moves. The acquisition ceiling built on this number moves with it, and a stale ceiling sanctions campaigns that no longer pay.