Meta Ads ROAS After Refunds: Are You Still Profitable?
Before I increase an ecommerce budget, I want to know what the acquired orders leave after reversals and variable costs. An early revenue-to-spend ratio can look healthy while returns are still on their way. Scaling that number can buy more of a margin problem.
For a brand spending at least $15,000 a month on Meta, I would keep two views: the calendar report that shows what happened this week, and the order cohort that follows purchases from the same acquisition period. They answer different questions. You need both to make the next spend decision.
Keep the original order date attached
Shopify's current sales-report documentation uses sales reversals for negative order adjustments, including more than physical returns. It records sales and reversals on their respective processing dates. Sales reports also differ from reports of payments moving between you and customers.
That means a September refund for an August order can affect the September calendar view. It does not establish that September's newly acquired customers caused the problem. Shopify's discrepancy guide explains why timing and report definitions need reconciliation.
Build an order-level working table with original order ID, order date, acquisition assignment under your stated model, product variant, net merchandise amount, subsequent adjustments and observation date. Keep physical returns, cash refunds, cancellations and goodwill credits distinguishable. One broad negative number is not a diagnosis.
Build the contribution bridge
This example is hypothetical and excludes taxes and shipping revenue for simplicity. A cohort of 100 orders initially produces $10,000 after discounts. Twenty orders are fully refunded. The returned goods are recovered in sellable condition; the final product cost below covers the 80 retained orders. Other costs are supplied assumptions.
| Cohort item | Amount | Treatment |
|---|---|---|
| Initial merchandise sales | $10,000 | After discounts |
| Sales refunded | −$2,000 | Linked to these orders |
| Net product cost | −$2,400 | 80 retained orders at $30 |
| Outbound fulfillment | −$800 | All 100 orders at $8 |
| Payment costs retained | −$300 | Illustrative actual net fees |
| Return handling | −$300 | 20 returns at $15 |
| Acquisition media | −$3,000 | Assigned consistently to this cohort |
| Contribution after media | $1,200 | Before fixed costs and other acquisition costs |
The initial sales-to-media ratio is 3.33. After refunds, retained merchandise sales divided by the same media spend is 2.67. Neither ratio is profit. The bridge shows $1,200 left under the stated assumptions, before management fees, creative costs and fixed overhead.
If the returned stock is unsellable, include the additional loss. Do not both remove its full product cost and claim a separate inventory recovery benefit. Equally, do not assume payment fees are all refunded: use your actual net fees. The purpose is a consistent contribution calculation, not a cosmetically better ratio.
Compare cohorts at the same age
A seven-day-old cohort has had less time to return than a forty-day-old cohort. Choose observation ages that fit your delivery and return behavior, then compare like with like. Keep the order window and the observation cutoff visible.
For a current cohort, show observed reversals separately from an estimated allowance for returns still to come. Build that allowance from comparable mature orders, then vary it. Product mix, a new offer, delivery delays or a different market can make yesterday's return pattern a poor estimate.
For example, another $500 of net return-related cost would reduce the illustrative $1,200 contribution to $700. That is a sensitivity calculation, not a forecast. It tells you how much uncertainty the budget is carrying. Do not present the allowance as an observed refund.
Find what created the return
Inspect return reasons alongside the exact ad, product page and variant. An ad can set the wrong expectation about size or use. A warehouse error needs a fulfillment fix. A defective batch needs a product response. The return code alone does not prove which one caused the loss.
Keep mixed baskets and exchanges in view. A partial refund does not make the entire customer worthless, and an exchange may preserve revenue while adding handling cost. Connect the adjustment to its order rather than subtracting every refund from whichever campaign happens to be running today.
Make the budget decision from what remains
I would compare contribution after the return allowance with the profit reserve and the cost of acquiring more customers. If the weakest product or promise drives the loss, restrict the relevant promotion and fix that issue. If the whole cohort is unprofitable at realistic return levels, a higher account ROAS target alone does not repair the offer.
Use my new-customer economics guide for acquisition definitions and the cash-flow worksheet for payment timing. My BFCM case shows the account work behind $544,397.42 in Meta spend for one US jewelry brand; it is separate from this hypothetical returns example.
If your ecommerce account already spends $15,000 or more a month, bring the order economics with the ad report. See my ecommerce management service and schedule a call. The budget needs to fit the money your orders actually leave.