Set a useful ROAS threshold from contribution margin, customer mix, repeat behavior, and the uncertainty inside attribution.
There is no universal good ROAS. The same reported return can be healthy for a high-margin repeat product and destructive for a discounted item with expensive fulfillment. The threshold must start with the economics of the order and the quality of the customer acquired.
Key takeaways
- Calculate contribution after product cost, discounts, fees, fulfillment, returns, and variable service.
- Returning customers can make platform efficiency look stronger while adding less incremental value.
- Use observed cohort behavior with a stated window instead of assuming every new buyer reaches an average lifetime value.
Translate margin into an acquisition boundary
Calculate contribution after product cost, discounts, fees, fulfillment, returns, and variable service.
Use the remaining contribution to define an affordable acquisition range rather than borrowing a benchmark.
Separate new and returning customers
Returning customers can make platform efficiency look stronger while adding less incremental value.
Read customer mix and set different expectations for acquisition and reactivation.
Treat repeat value as evidence, not hope
Use observed cohort behavior with a stated window instead of assuming every new buyer reaches an average lifetime value.
Discount uncertain future value and update the assumption as cohorts mature.
Reconcile attribution
Compare platform reporting with orders, analytics, CRM, and the selected attribution window.
Record disagreement instead of forcing unlike systems to match.
Use a decision range
Build scale, hold, and cut conditions from contribution, customer quality, creative health, and capacity.
Review the range when pricing, mix, or repeat behavior changes.
- Translate margin into an acquisition boundary
- Separate new and returning customers
- Treat repeat value as evidence, not hope
- Reconcile attribution
- Use a decision range
