How to calculate an Amazon FBA return reserve per sale
Last updated
An FBA return reserve is the expected net loss from returns allocated to each unit sold. Multiply the return rate by the net loss per return, not by the sale price.
The formula
Return reserve per sale = expected return rate × expected net loss per return
Enter the resulting dollar amount in the returns field of the FBA profit calculator. Keep actual refunds and the reserve from covering the same loss twice - a reserve is a forward-looking estimate, not an addition to a loss that has already posted.
Return rate alone is not return cost
A 10% return rate does not mean the seller loses 10% of revenue. Some returned units go back into sellable inventory. Others become damaged, require repackaging, incur processing or removal costs, or receive a reimbursement. The reserve needs two separate inputs: how often a delivered unit becomes a return, and how much money one return costs after recoveries. Two products with the same return rate can need different reserves - a sealed item returned in resellable condition may lose a few dollars, while a used or damaged item may lose its full landed cost plus fees and disposal or removal cost.
Building the loss for each return outcome
Resellable return. The unit goes back to available inventory, so the seller recovers the product's inventory value. The remaining loss can include nonrefundable charges, repackaging, relabeling, a discount needed to resell, and the cost of extra time in storage.
Unsellable return. The seller can lose the landed product cost and add removal, disposal, inspection, or refurbishment expense. Subtract salvage proceeds, insurance recovery, supplier credit, or Amazon reimbursement when those recoveries are supported by records.
Returnless refund or missing return. The seller can lose both the refund and the inventory when the unit never comes back. Any later reimbursement changes the final loss, so keep the claim as a separate transaction until it is matched to the original order.
Estimate the loss per return outcome:
Net loss per return = refund and return costs + lost inventory value − fee credits − reimbursements − recovered value
When resellable and unsellable returns are tracked separately:
Expected loss per return = (resellable share × resellable-return loss) + (unsellable share × unsellable-return loss)
Worked example: an 18% return rate creates a $2.10 reserve
A seller models a $39.99 product with a high expected return rate. Based on similar products and early data, the seller uses these assumptions: an 18% return rate, a 65% resellable share, a $4.50 net loss on a resellable return, and a $25.00 net loss on an unsellable return. The loss amounts are modeled net costs after expected fee credits, inventory recovery, and reimbursements - not Amazon fee quotes.
| Step | Calculation | Result |
|---|---|---|
| Expected loss per return | (65% × $4.50) + (35% × $25.00) | $11.68 |
| Return reserve per sale | 18% × $11.68 | $2.10 |
| Profit after reserve | $8.40 pre-return profit − $2.10 | $6.30 |
Multiplying the 18% return rate straight by the $39.99 sale price would create a $7.20 reserve and cut profit to $1.20. That shortcut treats every return as a complete loss of revenue and ignores resellable inventory, fee credits, and reimbursements - use matched net loss per return rather than a flat percentage of price.
Illustrative reserves by category
These rows are modeled scenarios for comparison, not Amazon-quoted rates or category averages. Replace them with a product's own mature return data as soon as it exists.
| Category profile | Return rate | Dominant loss driver | Modeled reserve per sale |
|---|---|---|---|
| Sized apparel | 25% | High resellable share after inspection, steaming, or relabeling | $1.80 on a $32 item |
| Small electronics accessory | 9% | Low resellable share, high unsellable loss from testing or defect claims | $2.35 on a $28 item |
| Consumable or single-use item | 2% | Mostly returnless refunds on damaged-in-transit claims | $0.14 on a $15 item |
| Oversized or furniture-adjacent item | 7% | Removal or disposal cost on damaged freight, some returnless refunds | $6.40 on a $180 item |
| Mid-price kitchen tool | 6% | Balanced resellable and unsellable mix | $0.78 on a $34 item |
Calculate from a mature cohort
For a completed historical cohort, the same result can be calculated in one step:
Realized return loss per sale = total matched net return loss / cohort units sold
Suppose 1,000 delivered units have had enough time for returns and follow-up adjustments to finish posting: 50 resellable returns average $5.00 net loss each, and 30 unsellable returns average $22.00 net loss each, for $910 total net loss. The observed return rate is 80 returns / 1,000 units = 8%. The average net loss per return is $910 / 80 = $11.38. The realized return loss per sale is 8% × $11.38 = $0.91, which matches the direct calculation of $910 / 1,000 units. Use unrounded values in the worksheet - the displayed percentages and averages can create a cent of rounding difference.
A historical result is useful only after the cohort has had enough time to produce returns, condition updates, fee credits, and reimbursements. A recent sales week can look unusually clean because many returns have not posted yet.
New products need a scenario range
A launch has no product-specific return history. Use a low, base, and high case rather than copying one marketplace-wide return-rate claim.
| Scenario | Return rate | Net loss per return | Reserve per sale | Profit after reserve, if pre-return profit is $6.50 |
|---|---|---|---|---|
| Low | 4% | $12.00 | $0.48 | $6.02 |
| Base | 8% | $16.00 | $1.28 | $5.22 |
| High | 12% | $20.00 | $2.40 | $4.10 |
The range forces the buying decision to carry uncertainty. If the product only reaches the seller's target margin in the low case, the launch depends on unusually favorable returns. Use evidence from the closest product type, price, customer use, and packaging, then replace the starting assumption once the product's own mature cohort is large enough to be useful.
Do not count the same return twice
Forward-looking and historical models use different treatments. Product screening uses an expected reserve per future sale. An open recent cohort uses actual losses posted so far plus a reserve for unresolved exposure. A mature historical cohort uses matched realized losses, with no duplicate reserve for the same returns. If a payment report already contains a $25 net return loss and the profit worksheet also subtracts a $2 reserve for that same closed sale, profit is understated. Keep a column that labels each number as actual, estimated, or unresolved. The true profit per SKU guide covers the wider cost stack this reserve feeds into.
Where the reserve comes from in Seller Central
FBA Customer Returns reports the return reason and product condition. Inventory Ledger records customer returns and inventory adjustments. Reimbursements lists seller-requested and automatically generated inventory reimbursements. Removal reports show removal status and fees. Payments Transaction View adds refunds, refunded fees, service charges, and credits. Match these sources by order, SKU, and transaction date where the data allows it - a raw refund total is not the same number as the net loss the reserve needs.
Questions
What is a return reserve in an Amazon FBA profit calculator?
It is the expected net return loss assigned to each unit sold. It keeps forward-looking profit from assuming that every sale remains final and every returned unit recovers full value.
It is a single dollar figure per sale, built from two inputs multiplied together: how often a delivered unit becomes a return, and how much money one return costs after inventory recovery, fee credits, and reimbursements.
How do I calculate an FBA return reserve?
Multiply the expected return rate by the expected net loss per
return: reserve per sale = return rate × net loss per
return. Estimate net loss after resellable inventory, fee
credits, reimbursements, salvage, and any return-related
charges.
On a product with an 18% return rate and an $11.68 expected net loss per return, the reserve is 18% × $11.68 = $2.10 per sale.
Should I enter return rate or dollars into an FBA profit calculator?
Enter the expected dollar loss per sale in the returns field. Calculate that amount outside the field from return rate and loss severity, then test a high case before committing to a purchase order.
A calculator field that only accepts one number cannot separate return probability from loss severity, so doing that split in a worksheet first is what keeps the single dollar figure honest.
Is return rate multiplied by sale price an accurate return reserve?
Only when a return loses the full sale price with no recovery or extra cost, which is an unusual modeling assumption for most products. Use outcome-based net loss instead.
On the $39.99 worked example above, return rate times price gives a $7.20 reserve and a $1.20 profit; the outcome-based calculation gives a $2.10 reserve and a $6.30 profit. The gap is the resellable inventory, fee credits, and reimbursements the flat-percentage shortcut ignores.
What return reserve should a sized apparel product model?
Apparel with size-dependent fit tends to carry a high return rate but a high resellable share, since most returns come back unworn and go straight back to inventory after inspection or a light repackaging pass. A modeled scenario of a 25% return rate with most returns resellable can still land near $1.80 per sale on a $32 item, well under what the raw return rate suggests.
This is a modeled illustration, not a category-wide rate. Model size-related apparel returns separately from an unrelated household product, since blending them into one account-level rate hides the SKUs that actually need the reserve.
What return reserve should a small electronics accessory model?
Small electronics tend to carry a lower return rate than apparel but a higher loss per return, because a unit returned as defective, incompatible, or "did not work as expected" is often not resellable as new and may need testing, inspection, or disposal. A modeled scenario of a 9% return rate with a low resellable share can land near $2.35 per sale on a $28 item, close to or above what a much higher apparel return rate produces.
The lesson is not "electronics is worse than apparel" - it is that return rate alone ranks products incorrectly. Loss severity per return outcome is what actually drives the reserve.
Does a low-return consumable product still need a return reserve?
Yes, but the reserve can be very small. A single-use or consumable item with a 2% return rate, mostly returnless refunds on damaged-in-transit claims, can model to roughly $0.14 per sale on a $15 item - small enough that some sellers round it to a nominal placeholder rather than zero.
Zero is only valid when the stated evidence supports no expected net loss at all. For a new consumable without return history, use a small low-base-high range instead of assuming zero, since a returnless-refund claim still costs the full unit even at a low rate.
How does a returnless refund change the return reserve?
A returnless refund means the seller loses both the sale revenue and the unit, since Amazon refunds the customer without requiring the item back. There is no resellable-inventory recovery to offset the loss, so the net loss per return is closer to full landed cost plus fees than a standard resellable return.
Model returnless-refund-heavy categories - typically bulky, low-value, or shipping-damage-prone items - with their own loss severity rather than the blended resellable/unsellable split used for a normal return, and keep any later reimbursement as a separate transaction until it is matched to the original refund.
What return reserve should an oversized or furniture-adjacent item model?
Oversized items carry real removal and disposal cost on top of a comparatively modest return rate, because freight damage and returnless refunds on bulky products are expensive per incident even when they are infrequent. A modeled 7% return rate with removal and disposal costs weighted in can land near $6.40 per sale on a $180 item - a small percentage of return rate producing a large per-sale dollar reserve.
For this profile, loss severity per incident matters more than return frequency. Underestimating disposal and removal cost is the most common way an oversized SKU's reserve is modeled too low.
How do I work out a return reserve when resellable and unsellable returns are mixed?
Split the return population into a resellable share and an
unsellable share, price the net loss for each outcome separately,
then weight them by their share of total returns: expected
loss per return = (resellable share × resellable-return loss)
+ (unsellable share × unsellable-return loss).
On the worked example, a 65% resellable share at $4.50 loss and a 35% unsellable share at $25.00 loss blend to $11.68 expected loss per return - a figure neither share alone would produce.
How do I calculate a return reserve from a mature return cohort?
Divide total matched net return loss for the cohort by the
number of units delivered in that cohort: realized return
loss per sale = total matched net return loss / cohort units
sold. On 1,000 delivered units with $910 in total matched net
return loss, the realized reserve is $0.91 per sale.
Only use a cohort once it is mature enough that returns, condition updates, fee credits, and reimbursements have finished posting - a recent sales week looks artificially clean because many of its eventual returns have not arrived yet.
How often should I update an FBA return reserve?
Review it after a mature cohort becomes available, and after any material product, packaging, listing, price, or policy change. High-volume SKUs can support more frequent review than low-volume products, since they reach cohort maturity faster.
A packaging fix that reduces damage-related returns, a price change that shifts customer mix, or a listing edit that changes buyer expectations can all move the reserve - recalculate rather than carrying a number from before the change.
What mistakes distort an FBA return reserve the most?
Four show up repeatedly: treating the refunded amount as the net loss instead of accounting for recoveries; measuring a cohort before its returns have finished posting; assuming every return is resellable instead of measuring the sellable share; and using one return rate for an entire catalog instead of calculating it per SKU or product group.
All four push the reserve in the same direction - too low - which is what makes a return-heavy SKU look more profitable on paper than it actually is.
Run your own numbers: the free FBA profit calculator models fees, storage, and cost of capital per unit - no signup, nothing leaves your browser.