Enter revenue and cost of goods sold. You get gross profit, gross margin, what survives ad spend, and the break-even ROAS that margin implies.
Two numbers, one subtraction. The difficulty is never the arithmetic. It is deciding what belongs in the second term, and that decision moves the answer more than any pricing change you are likely to make.
The uncontested items are the ones that scale one for one with a unit sold: the landed cost of the product, the packaging it ships in, and the labour that touches it. Nobody argues about those.
The arguable items are where two businesses with identical bank balances end up reporting margins eight points apart.
Payment processing. Roughly two to three percent of every order, charged only when an order happens. It behaves exactly like a unit cost, so it belongs in cost of goods sold. Many operators park it in overheads and quietly overstate their margin.
Outbound shipping. If you ship it, it costs you per order, and it belongs in. If the customer pays it in full, it nets out. If you offer free shipping over a threshold, part of it is a discount and part is a cost, and the honest treatment is to average the real cost per order and include it.
Returns and refunds. A ten percent return rate on a physical product is not a rounding error. Deduct returns from revenue and add the unrecoverable cost back to the numerator.
Warehouse rent and salaried staff. These do not move when you sell one more unit, so they sit below gross profit. Including them gives you contribution or operating margin, which is a different and also useful number, but it is not gross margin and it will not answer the question this page exists to answer.
The test is mechanical. Ask whether the cost would still exist if you sold one fewer unit this month. If the answer is no, it is a cost of goods sold.
| Cost of goods on a $100 sale | Gross profit | Gross margin | Break-even ROAS |
|---|---|---|---|
| $10 | $90 | 90% | 1.11x |
| $20 | $80 | 80% | 1.25x |
| $30 | $70 | 70% | 1.43x |
| $40 | $60 | 60% | 1.67x |
| $50 | $50 | 50% | 2.00x |
| $60 | $40 | 40% | 2.50x |
| $70 | $30 | 30% | 3.33x |
| $80 | $20 | 20% | 5.00x |
| $90 | $10 | 10% | 10.00x |
Read the right-hand column downwards and the shape of the problem appears. Between 90 and 60 percent margin the required return barely moves. Below 30 percent it doubles, then doubles again. A business at 20 percent margin needs five dollars back for every dollar spent, which is not a bidding problem or a creative problem. It is a pricing and sourcing problem wearing an advertising costume.
Gross profit is the last figure in the accounts that still moves with volume. Everything below it, rent, salaries, software, is roughly the same whether you sell four hundred units this month or six hundred. So gross profit is the pool advertising is allowed to draw from, and the size of that pool is the honest ceiling on spend.
Take the worked example on this page. Revenue of $37,410 against $21,698 of goods leaves $15,712 of gross profit at a 42 percent margin. Ad spend of $10,000 leaves $5,712 before a single overhead is paid. That $5,712 is what the rent, the salaries and the profit have to come out of.
Run the same campaign at 25 percent margin and the gross profit falls to $9,353. The same $10,000 of spend now puts the month underwater by $647, with no overheads paid at all. Nothing about the advertising changed.
Gross profit is not profit. It is the budget that profit has to be found inside.
The word profit in the name does most of the damage. A founder sees a healthy gross margin, concludes the business is making money, and scales spend against it. Six months later revenue has doubled, gross profit has doubled, and the bank balance has not moved, because fixed costs were never in the calculation and the extra volume brought its own working capital demands.
The second version of the same error is subtler. Gross margin gets calculated once, from a spreadsheet built in a quieter year, and then treated as a constant. Supplier prices moved. The freight surcharge became permanent. Discount codes that were a one-week experiment are now on forty percent of orders. The margin in the model says 42 percent and the margin in the bank says 34, and every ROAS target built on the first number is quietly wrong by a third.
Recalculate it from last quarter’s actual numbers, not from the price list. If the two disagree, the bank is right.