Is Amazon FBA Still Profitable in 2026? A Margin Model
Test Amazon FBA profitability with three US product examples, current fee inputs, refund accounting, advertising sensitivity and fixed-cost break-even math.
FBA can be profitable, but a fee table cannot tell you whether your particular product will make money. The useful test is narrower: after customer refunds, Amazon charges, the cost of the goods and advertising, does each original order leave enough to pay the business's fixed costs? A product can pass that test at one price and fail it after a discount. More orders do not repair a negative contribution per order.
This guide compares three hypothetical Home and Kitchen products sold on Amazon.com for $15, $35 and $80. The published fees come from Amazon; costs, demand and return behaviour are illustrative assumptions. These are not customer accounts or industry averages. All calculations use US dollars and the US marketplace. A translation of the article does not turn the figures into another country's rate card.
The sources are Amazon's selling fee schedule and 2026 US FBA rate cards, checked on October 22, 2026. Use Amazon's Revenue Calculator to obtain a current estimate for a real ASIN. This article explains how to use those inputs without confusing contribution with net profit.
Start with the right fees and the right comparison
All three products use a 15% referral fee. Product A is small standard, more than 4 oz and no more than 6 oz; B is large standard, more than 12 oz and no more than 16 oz; C is large standard, more than 2 lb and no more than 2.25 lb. A and B use the $10–$50 price band. C uses the band above $50. Changing the price band or measured dimensions can change the calculation.
Input
A
B
C
Price
$15
$35
$80
Size and shipping weight
Small standard, 4+–6 oz
Large standard, 12+–16 oz
Large standard, 2+–2.25 lb
Referral fee (15%)
$2.25
$5.25
$12.00
Non-peak fulfillment, with surcharge
$3.57
$4.76
$6.40
Peak fulfillment, with surcharge
$3.78
$5.06
$6.78
Refund administration
$0.45
$1.05
$2.40
The non-peak and peak fulfillment figures both include the 3.5% fuel and logistics surcharge. It took effect for US FBA on April 17, 2026; the peak window runs from October 15, 2026 through January 14, 2027. The non-peak comparison therefore means the 2026 non-peak card with the surcharge applied, not the actual fee charged before April 17. Do not apply the surcharge a second time to these figures. They are rounded to cents before the rest of the model is calculated.
The refund administration amount follows the published rule: the smaller of $5 or 20% of the referral fee. A different fee category, a program discount or a category-specific returns charge can produce different results. Match the actual transaction rather than treating these three examples as quotes for every product.
Keep operating assumptions separate from Amazon's rules
For the worked model, landed product cost is 30% of the selling price, advertising is 12%, and storage plus miscellaneous variable expenses are 2%. Landed cost includes the supplier price, freight, duty and allocated inbound costs. If your inbound placement fee or prep bill is already in that number, do not subtract it again. If it is missing, add it before using the result to place an order.
Advertising means spend divided by total gross sales for the same set of original orders. It is not campaign ACOS, whose denominator is attributed advertising sales. A campaign with 30% ACOS does not imply that 30% of all store revenue goes to advertising. The model uses the assumed $1.80, $4.20 and $9.60 ad cost per original order even when that order is later refunded. It does not assume you recover the click cost.
The examples initially assume no returns and then show a separate 5% full-refund scenario. In that scenario every returned product can be sold again at its original inventory cost. The outbound fulfillment fee stays charged, the referral fee is credited less the refund administration amount, and no additional return transport or processing charge is included. These are explicit modelling assumptions to replace with your settlement data, not promises about reimbursement eligibility.
Calculate contribution before calling anything profit
An order the buyer keeps contributes its price minus the referral fee, fulfillment fee, landed product cost, advertising and the 2% variable-expense allowance. Subscription, other fixed overhead, financing costs and tax have not yet been subtracted. That remaining amount is contribution, not money available to withdraw from the company.
Contribution per original order
A ($15)
B ($35)
C ($80)
No returns, non-peak
$2.58
$9.59
$26.40
No returns, peak
$2.37
$9.29
$26.02
5% full refunds, non-peak
$2.15
$8.58
$24.08
5% full refunds, peak
$1.94
$8.28
$23.70
Peak contribution / original price
12.9%
23.6%
29.6%
The percentage row divides expected peak contribution by the original selling price. It is not a net-profit margin or a margin on revenue after refunds. Keeping that denominator explicit matters when comparing it with a finance report. Results are rounded for display; keep the underlying precision when adding hundreds of orders or calculating a break-even volume.
The small product has the least room because fulfillment takes a larger share of a $15 selling price. That does not prove that expensive products are always better. They may need more working capital, cost more to acquire, sell more slowly or have higher return losses. The comparison holds those other inputs constant only to show how the fee structure behaves.
Account for refunds without charging fulfillment twice
For Product B in the peak window, a retained order contributes $35 − $5.25 − $5.06 − $10.50 − $4.20 − $0.70 = $9.29. A fully refunded order has zero retained sales revenue. Under the recoverable-inventory assumption, its costs are $5.06 outbound fulfillment, $1.05 refund administration, $4.20 advertising and $0.70 other variable expenses. Its contribution is negative $11.01. The $10.50 product returns to inventory; that is an asset recovery, not new revenue.
With 95% retained orders and 5% refunded orders, expected contribution is 0.95 × $9.29 + 0.05 × negative $11.01 = $8.275, displayed as $8.28. This averages two mutually exclusive outcomes. It does not keep $35 revenue for a refunded order, and it does not subtract the outbound fulfillment charge once for every order and then again for returned orders.
If only 80% of the product cost is recoverable on a return, subtract another 0.05 × 20% × $10.50 = $0.105 per original order. Additional return shipping, inspection, processing fees or disposal costs reduce contribution further. Refunds can arrive weeks after the sale, so compare completed order cohorts rather than treating a young sales week as a final result.
Test landed cost and advertising together
The following grid holds Product B's price, peak fulfillment fee, 5% refund assumption and 2% expense allowance constant. Rows change landed cost as a share of price. Columns change advertising spend as a share of original gross sales. Every cell is expected contribution per original order, before fixed overhead and tax.
Landed cost / advertising
8% ads
12% ads
20% ads
25%
$11.34
$9.94
$7.14
35%
$8.01
$6.61
$3.81
45%
$4.69
$3.29
$0.49
At 45% landed cost and 20% advertising, only about $0.49 remains. A small discount, a higher return-processing bill or one packaging problem could consume it. At 25% landed cost and 8% advertising, the same fee card leaves $11.34. Neither cell is a prediction; the point is to show which assumptions deserve measurement before ordering more inventory.
Moving advertising from 12% to 20% costs Product B $2.80 per original order. Moving from the non-peak to the peak fulfillment card, while applying the surcharge to both, costs $0.30. The surcharge and the peak increase are different comparisons: against the non-peak base fee without surcharge, peak fulfillment is $5.06 rather than $4.60, a $0.46 difference. Do not label $0.30 as the combined cost of both changes.
Fixed costs and cash timing still decide the business result
Suppose the Professional subscription is $39.99 a month and other fixed overhead is an illustrative $300. Total fixed costs are $339.99. Divide that once by expected contribution per original order. Using the unrounded peak results, the break-even volumes are 176 original orders for A, 42 for B and 15 for C. Round required orders up, never down. Allocating the subscription to units and also subtracting it as a monthly expense would count it twice.
At 200 monthly original orders, Product B contributes $1,655 before fixed costs. Subtract $339.99 and $1,315.01 remains before tax, financing and any omitted expenses. That figure is conditional on all the assumptions holding. It is not a suggested income target or evidence that a new seller will reach that order volume.
Positive contribution also differs from cash in the bank. You may pay suppliers and freight months before Amazon disburses sales proceeds. Reordering stock can consume more cash than the accounting profit from the previous batch. Track the supplier payment date, receiving delay, payout timing and refund reserve separately. A growing SKU can require additional cash even when every retained order is profitable.
Turn the model into a product decision
Build one row per ASIN and per marketplace. Record the fee estimate date, the packaged dimensions and the cost documents used. Separate current prices from proposed discounts. Then run three versions: the current case, a case with higher ad spend or refunds, and a lower-volume case with the same monthly overhead. This exposes a fragile plan faster than changing every input to the best number you can imagine.
Confirm the fee category and shipping-weight tier against Amazon's estimate and actual settlements.
Reconcile landed cost to invoices, freight, duties and inbound bills; assign each cost once.
Use total advertising spend and total sales from a matching period, not an attributed-sales denominator by accident.
Separate retained orders from full and partial refunds, and record what returned stock is actually worth.
Add aging inventory, storage surcharges, insurance, software and other expenses that the simplified allowance misses.
Compare contribution with fixed overhead and the cash required for the next purchase order.
Use the listing optimization guide when unclear dimensions, compatibility or claims attract the wrong customers. Correcting those facts may improve the economics, but it does not guarantee a lower return rate. Product defects, delivery damage and poor supply terms need operational fixes. Do not buy more traffic simply because a rewritten listing looks cleaner.
Common questions
Is FBA still worth it for low-priced products?
It can be, but the $15 example leaves much less room for error than the others. A cheaper landed cost, smaller package or more organic demand can help; unmeasured assumptions cannot. Use the current Low-Price FBA card separately if the selling price is below $10.
What profit margin should a new seller expect?
These examples cannot answer that. They show contribution under stated assumptions, not representative seller earnings. Obtain real purchase costs and order data, allow for refunds, and subtract all overhead before calling the remaining percentage profit.
Can a higher price solve the problem?
Only if enough customers still buy. Recalculate referral fees, the fulfillment price band and expected demand. A price increase that improves each order but halves the order count may leave less total contribution to cover fixed costs.
What should be fixed before increasing ad spend?
Start with any negative contribution, inaccurate product facts, unprofitable targeting and expensive returns. Then test a limited budget change and wait for the attribution and refund data to mature. Scaling a loss-making order merely makes the loss larger.
