Contribution margin is what a product line earns after the costs that vary with each unit sold, and it is the number that tells a seller whether selling more of something helps or hurts. It is not gross margin and it is not net profit. Gross margin stops at cost of goods sold. Net profit subtracts everything, including rent and salaries that would exist whether or not a particular SKU sold a single unit. Contribution margin sits between them, and for product line decisions it is the more useful of the three.
The definition, precisely
Contribution margin equals revenue minus variable costs. Variable costs are the ones that scale with volume: the unit cost of the goods, inbound freight and duty allocated per unit, the marketplace referral fee, fulfillment and pick and pack fees, payment processing, returns and refunds, and any advertising spend tied directly to that product.
What stays out is anything that does not move with the next unit sold. Warehouse lease, software subscriptions, the bookkeeper, the founder’s salary, and brand level marketing are period costs. They are real and they have to be covered, but they are covered by total contribution margin across all products rather than charged to any one of them.
The line between variable and fixed is where most of the judgment lives. Amazon storage fees are a common argument: they scale with inventory held rather than units sold, which makes them semi variable. A defensible treatment is to keep them out of the per unit calculation and cover them from total contribution, while watching them separately for any SKU that sits too long.
A worked example
Take a kitchen product selling at $39.99 on Amazon.
Referral fee at 15 percent is $6.00. Amazon’s published seller pricing puts most categories at 15 percent, with rates ranging from 5 percent to 45 percent by category and a $0.30 minimum per item, according to Amazon’s pricing page as of the 2026 schedule. Fulfillment for a standard size unit of this weight runs $5.80. Landed unit cost, including freight and duty, is $9.40. Payment and miscellaneous transaction costs are negligible on Amazon because the referral fee absorbs them. Returns run 4 percent, which at this price is $1.60 per unit sold across the line. Advertising attributable to the product averages $3.20 per unit.
Revenue of $39.99 less $6.00, $5.80, $9.40, $1.60, and $3.20 leaves $14.00 in contribution margin, or about 35 percent of revenue.
That figure answers a specific question: every additional unit sold contributes $14.00 toward fixed costs and profit. At $42,000 of monthly fixed costs, this product alone would have to move 3,000 units to cover them. Most sellers run several lines, so the real question is what each one contributes to the same pool.
Why it changes decisions that gross margin does not
Consider a second product in the same catalog at $24.99 with a landed cost of $6.00. Its gross margin percentage looks better than the first product’s. But if it carries a 9 percent return rate, a heavier fulfillment tier because of dimensional weight, and $4.10 in advertising to hold its position, its contribution per unit may be under $4.00.
Ranked by gross margin, the second product wins. Ranked by contribution, the first product is more than three times as valuable per unit sold. A seller optimizing the catalog on gross margin will push inventory dollars and ad budget toward the weaker product, and the error compounds every reorder cycle.
The same logic applies to a product with negative contribution. Any unit sold makes the business worse, and no amount of volume fixes it. That sounds obvious stated plainly and is hard to see when the only reporting available is a channel level profit and loss statement.
Where the inputs come from
The arithmetic is straightforward. Getting trustworthy inputs is the hard part, and it is why many sellers never calculate this at all.
Landed cost requires supplier invoices, freight bills, and duty allocated across units in a shipment, not the price on the purchase order. Marketplace fees require settlement reports rather than deposit totals, because the deposit has already netted everything together. Returns have to be matched back to the original sale, which marketplaces do not always make easy. Advertising needs product level attribution, a reporting exercise in its own right.
This is the work that ecommerce accounting tools are built to do. ConnectBooks, for instance, computes cost of goods sold automatically and reports profit and loss at SKU level for sellers running Amazon, Shopify, Walmart, TikTok Shop, and eBay into QuickBooks or Xero. A seller can assemble the same picture in a spreadsheet, and plenty do. Estimating the inputs is what fails, because contribution margin is sensitive enough that a 10 percent error in landed cost can flip a decision.
How often to recalculate
Quarterly is a reasonable baseline, with two triggers for doing it sooner: a change in supplier pricing or freight rates, and a change in marketplace fees. Both move the variable cost base directly.
Fee schedules change on their own timetable, which is why any contribution figure should carry the date of the fee schedule it was built on. A contribution margin calculated on last year’s fulfillment rates is a historical document, not a decision tool.
The limit of the measure
Contribution margin is a decision tool, not a reporting standard. It does not appear on a tax return or an audited financial statement, and a seller should not expect their accountant to produce it as part of a close. Financial statements follow a different logic, and the cost classifications that govern them are set out in IRS Publication 538 for method purposes and in general accounting standards for presentation.
It also says nothing about whether the fixed cost base is reasonable. A catalog can be full of healthy contribution margins and still lose money because overhead is too large for the revenue it supports. Contribution margin tells a seller which products to push and which to cut. It does not tell them whether the business as a whole is the right size, and treating it as though it does is the most common way the measure gets misused.
