The order is booked before all its consequences arrive.
The customer returns the bundle. Shipping costs more than expected. A payment fee remains. Another unit has to be inspected, repacked, or written off. Meanwhile, the campaign that generated the order is still being scaled.
This is how a best seller can become an expensive growth strategy.
The first act of our Media demo makes the problem visible. In its default DTC setting, the synthetic hero SKU shows 4.2× reported ROAS and approximately −$624 in contribution. Both figures belong to an illustrative scenario, not a customer result.
Move the return-rate control. Reported ROAS stays fixed in this demonstration while the contribution figure changes. It is a deliberate simplification that isolates the effect of returns. In a real account, the reported metric also depends on the values and adjustments sent to the platform.
The business calculation starts with a defined period and a consistent cost basis:
Contribution after advertising = revenue − refunds − cost of goods − payment fees − fulfillment − advertising spend.
The treatment of returned inventory, recovered product value, and return handling needs to be explicit. Shipping and fulfillment must not be counted twice. Estimated return costs should remain distinguishable from settled actuals.
A missing cost is a missing cost. In the demo, a row with an undeclared unit cost is excluded and named. Filling that gap with an invented margin would make the calculation easier to display and harder to trust.
Contribution is also not net profit or cash flow. Overhead, taxes, financing, inventory purchases, and payment timing can all affect the result the business ultimately experiences. A useful media decision starts with sound unit economics and stays connected to those wider constraints.
That changes the conversation between growth and merchandising. The product generating the most checkout revenue may be a poor candidate for the next advertising dollar. Another product may produce less revenue per order and retain more value after its costs.
The right response could be a different product mix. It could be a repair to the bundle, its description, its fulfillment process, or its return experience. The economics identify the problem; they do not automatically select the remedy.
MIZ OKI's net-yield pricing and bidding remain Preview · in development. Measurement and net-yield writebacks are off. The Shopify page presents the product direction and illustrative commerce scenarios; it should not be read as a claim that every merchant connector or profit-based bidding workflow is generally available.
The decision standard is already clear: cost completeness comes before an optimization recommendation, and a recommendation needs evidence and authority before money moves.
The same discipline applies to spend curves. A modeled relationship within an observed range is not permission to extrapolate indefinitely. Where evidence runs out, “no estimate” can be the most useful answer.
Before your next budget increase, ask which cost could reverse the apparent winner. Then ask whether that cost is in the decision record.
Open Act 1 and move the hero SKU's return-rate control. The Shopify contribution page explains the direction of the commerce offering.