A successful-order estimate is only one outcome
A normal sale calculator usually describes an order that stays sold. Returns introduce other outcomes: revenue may be refunded, shipping may already have been paid, inventory may or may not be resalable, and some original fees may remain. A profitable completed order can therefore overstate the average contribution across all orders placed.
Write down the retained-order and returned-order outcomes separately. Use your provider’s refund policy to determine which charges reverse. Do not assume every payment or marketplace fee is refunded. Stripe’s standard pricing page states that original transaction processing, Connect and currency-conversion fees are not returned on refunds; other arrangements need their own checks.
Use a two-outcome example
Suppose a retained order contributes $18 after its included costs. A returned order instead produces a $12 loss after the refund, unrecovered shipping, retained fees and handling. Assume 8% of placed orders are returned. Expected contribution per placed order is 0.92 × $18 + 0.08 × −$12 = $15.60.
The difference from the all-retained estimate is $2.40 per order. It is not merely 8% of the $12 loss. A return both removes the positive contribution that a retained order would have produced and introduces the loss. Relative to the $18 baseline, each returned order changes the result by $30.
Avoid counting the inventory loss twice
If returned stock can be sold again, the first return may not consume its entire product cost permanently. Inspection, repackaging, damage and markdowns still matter. If the item cannot be resold, include that loss in the returned-order outcome. Use a consistent inventory treatment across both outcomes.
For an allowance based on these assumptions, you could add $2.40 per order to allocated overhead while leaving the calculator’s normal-sale inputs unchanged. Label it as an expected return allowance. Do not also lower revenue or add the full returned-order loss elsewhere for the same allowance.
Replace assumptions with your own history
Record the number of returned orders, not just the dollar value of refunds. Separate categories with different return behavior, and allow enough time for recent orders to complete their return window. A month with many new sales and few completed returns can make the rate look artificially low.
Test a higher-return scenario before relying on the average. This allowance is a pricing model rather than a cash-flow forecast: a cluster of refunds can require cash even if expected contribution remains positive. Keep the underlying assumptions available so you can update the estimate when delivery costs or refund terms change.
Sources
Official sources checked October 29, 2026. Confirm current terms for your account before using a published rate.