Net sales minus the variable costs of fulfilling an order, layered as CM1 (after COGS), CM2 (after fulfilment, fees, shipping, returns), and CM3 (after acquisition cost).
Contribution margin is the dollars left from a sale after you subtract the variable costs of producing that sale. It is the foundation of every profit-true metric on this page, because revenue and ROAS both ignore what an order actually costs to fulfil. For DTC and ecommerce the standard practice is to layer it so you can see exactly where profit leaks: CM1, CM2, then CM3.
CM1 is net sales minus COGS, the raw product profit. CM2 takes CM1 and removes the variable cost of getting the order out the door: payment processing, pick and pack, outbound shipping, and a returns reserve. CM3 then subtracts the customer acquisition cost attributed to that order. Polar Analytics frames CM3 as 'the real DTC number' and notes that most Shopify reports stop at CM1, so gross margin overstates profit (polaranalytics.com). Saras Analytics describes the same layering as a way to measure profitability as variable-cost complexity grows (sarasanalytics.com).
Worked example (Demonstrative, from Polar's published figures): a A$72 net-sales apparel order. Minus A$22 COGS gives CM1 = A$50.00, or 69%. Minus A$2.30 payment, A$3.50 pick and pack, A$6.00 shipping, and A$2.50 returns reserve gives CM2 = A$35.70, or 50%. Minus A$14.00 CAC gives CM3 = A$21.70, or 30%. The order looked like a 69% margin at CM1 and is really a 30% margin once everything variable is counted.
Blufire computes every downstream number (break-even ROAS, POAS, LTV:CAC, CAC payback) from the operator's real contribution margin rather than a generic gross-margin proxy, then attaches the highest-value move and its dollar impact. That is the whole thesis: analytics built on profit, not revenue.