Fees for Amazon Sellers in 2026: What Actually Comes Out of Every Sale

How much do fees for Amazon sellers actually cost per sale?

There is no single answer, because the fees are charged on different bases. The mental model that matters is this: only one of your fees is a percentage of revenue. Everything else is a flat per-unit charge, which means your effective take rate moves inversely with your average selling price.

Here is the full stack on one FBA order, using a $100.00 product in a 15% referral category, large standard size. The fulfillment figure is the one number you must pull from your own rate card — it varies by size tier and price band — so it is marked as such.

Marketplace fees on that order come to roughly $22.13, or about 22% — not 15%. And that is before cost of goods, inbound freight, advertising, or the returns you will eat.

Now run the same stack on a $12.00 item. The referral fee drops to $1.80, but the fulfillment fee barely moves, because a box does not get cheaper to pick and ship just because you priced it lower. That is the entire economics of low-ASP Amazon selling in one sentence: flat fees are regressive. A product under roughly $15 spends a far larger share of its price on fulfillment than the category percentage ever suggests.

The two fee floors that catch new sellers

• The $0.30 minimum referral fee. Most categories apply it. On a $1.50 item, that $0.30 is a 20% effective referral rate, not 15%. Sub-$2 products are almost never viable on Amazon for this reason alone.
• Referral is charged on the total sales price. If you sell FBM at $20.00 plus $5.99 shipping, the referral fee is calculated on $25.99, not $20.00. Sellers who price shipping separately to look cheaper are paying referral fees on their own freight.

Amazon referral fees by category: the percentage everyone quotes

The referral fee is Amazon's commission and the single largest line on most orders. Rates run from 5% to 45% depending on category, though the great majority of physical goods sit between 8% and 15%. Amazon publishes the authoritative table on its official selling fees page, and it did not introduce new referral fee types in the US for 2026 — rates have now held steady across two consecutive years.

What generic fee articles miss is that several categories are tiered or conditional, so the headline percentage is often not the rate you pay:

Books, DVD, music, software and video also carry a variable closing fee of $1.80 per item on top of the referral fee. On a $6.99 paperback that single flat charge is more than a quarter of the sale price — which is why media sellers live or die on sourcing cost, not on pricing strategy.

One practical note worth more than any rate table: new sellers can qualify for automatic referral-fee waivers in their first year, worth up to $50,000 in total incentives. If you are modelling year-one unit economics, model them twice — once with the waivers and once without — because the second model is the business you actually have in year two.

The FBA fees that never show up in a calculator

Fulfillment fee calculators answer one question: what does it cost to pick, pack and ship this unit once. That is the fee sellers plan for. The fees that surprise them are the ones tied to inventory behaviour rather than to sales.

The pattern is consistent: Amazon has steadily shifted fees away from the moment of sale and toward the way you operate your inventory. Two sellers with identical products, identical prices and identical sales volume can now have materially different fee profiles purely because one manages weeks-of-cover well and the other does not.

What changed in 2026: the January rate card and the April surcharge

Two separate events moved fees for Amazon sellers this year, and they work very differently.

January 2026: the annual rate card

Effective 15 January 2026, Amazon updated US FBA fees. The headline is that fulfillment fees rose by an average of $0.08 per unit sold — less than 0.5% of the average item's selling price — with adjustments spread across fulfillment, inbound placement, storage, aged inventory, low-inventory-level, removal, disposal and returns processing. Referral fees were left alone, as they were in 2025.

An $0.08 average is genuinely modest. But it is an average, and the 2026 fulfillment schedule prices small standard items in three bands — under $10, $10 to $50, and over $50. If your ASP sits close to a band boundary, your per-unit change can be several times the average in either direction. Check your own tiers; do not plan from the headline.

April 2026: the 3.5% fuel and logistics surcharge

The larger story is the surcharge Amazon announced in April. From 17 April 2026, a 3.5% fuel and logistics-related surcharge applies to fulfillment fees across FBA in the US and Canada, and to Remote Fulfillment with FBA from the US into Canada, Mexico and Brazil. From 2 May 2026 it extends to Buy with Prime in the US and Multi-Channel Fulfillment in the US and Canada. Amazon describes it as temporary and says it will continue to evaluate it as conditions evolve, but has published no end date. The full notice is in Amazon's Seller Central announcement.

The critical detail, and the one most commentary got wrong in the first week: the surcharge is calculated on your fulfillment fees, not on your sale price. That distinction is worth real money. 3.5% of revenue on a $100 order would be $3.50; 3.5% of a $6.50 fulfillment fee is $0.23.

But because it rides on a flat fee, the surcharge inherits the same regressive shape as everything else in the FBA stack:

Substitute your own rate-card figures and the shape holds: high-ASP sellers can absorb this without repricing, while sub-$15 catalogues lose a meaningful slice of an already thin margin. If you sell across both ends of that range, do not apply a blanket price increase — the surcharge does not affect your SKUs equally, and repricing your $100 items to cover a $0.23 charge hands share to competitors for nothing.

How to calculate Amazon selling fees before you source a product

Most Amazon selling fees calculators give you referral plus fulfillment and stop. That produces an optimistic number you cannot run a business on, because it excludes every inventory-behaviour fee and every allowance for returns and advertising.

The model below is the one worth building into a spreadsheet or a sourcing script. It computes contribution per unit, which is the only number that tells you whether a product is worth buying.

def net_per_unit(
    price,              # your list price
    shipping_charged,   # 0 for FBA/Prime
    referral_rate,      # e.g. 0.15
    fulfillment_fee,    # from YOUR rate card, by size tier + price band
    cogs,
    inbound_per_unit,   # freight + prep, amortised per unit
    storage_per_unit,   # monthly storage / expected monthly velocity
    ads_per_unit,       # total ad spend / total units, not just attributed
    return_rate,        # category reality, not your optimism
    return_cost,        # returns processing + unsellable write-off
    closing_fee=0.0,    # $1.80 on media
):
    revenue_base = price + shipping_charged
    referral = max(0.30, referral_rate * revenue_base)
    fulfillment = fulfillment_fee * 1.035   # 3.5% fuel + logistics surcharge
    returns = return_rate * (return_cost + fulfillment)

    return (
        revenue_base
        - referral
        - closing_fee
        - fulfillment
        - storage_per_unit
        - ads_per_unit
        - returns
        - cogs
        - inbound_per_unit
    )

Three modelling decisions separate a useful calculator from a comforting one:

1. Divide total ad spend by total units, not by attributed units. Attributed-sales accounting makes every campaign look profitable. Organic sales subsidised by advertising are still sales you paid for.
2. Charge the fulfillment fee again on the return. A returned unit consumed a fulfillment event that is not refunded, plus a return trip and often a grade-down. Modelling returns at only the processing fee understates them badly in apparel and electronics.
3. Apply the surcharge multiplier inside the model, not as a footnote. It is 1.035 today. Keep it as a variable, because Amazon has called it temporary and it will either lapse or move.

If the output is positive but under roughly 10% of price, treat the product as a cash-flow liability rather than a margin business — one storage season or one return spike puts it underwater.

Where to find your real Amazon Seller Central fees, not an estimate

Every calculator, including Amazon's own fee preview, is an estimate derived from the product dimensions currently on file. The authoritative number is what Amazon actually deducted. Three places hold it, in increasing order of usefulness:

• Fee Preview report (Reports → Fulfillment). Per-SKU estimated fees based on recorded dimensions and current rates. Useful for catalogue-wide scanning, not for reconciliation.
• Settlement reports (Payments → Reports Repository). The ground truth: every fee line Amazon charged in a settlement period, per order. This is what you reconcile against, and it is the only source that catches a fee you were not expecting.
• The SP-API Product Fees API. getMyFeesEstimateForSKU returns a structured fee breakdown for a given price, which lets you price-test a SKU programmatically instead of guessing at band boundaries one product at a time.

The reconciliation that pays for itself is comparing settlement-report fulfillment fees against the size tier you believe a SKU occupies. Amazon measures units on intake, and a re-measurement — a slightly puffier poly bag, a revised carton — can silently move a SKU into a more expensive tier. Nothing alerts you. You see it as margin erosion weeks later, and by then you have been overcharged on every unit shipped since. Where a measurement is genuinely wrong, it can be disputed and overcharged fees reimbursed, but only if you noticed.

If you sell across several marketplaces, the same reconciliation logic applies everywhere and is worth automating rather than repeating in spreadsheets. We cover the wider cost picture in our complete guide to Amazon marketplace costs, which maps the full fee taxonomy alongside the operating costs that sit outside Amazon's invoices.

How to reduce fees for Amazon selling without losing the Buy Box

Very few Amazon fees are negotiable. Nearly all of them are engineerable, because they are triggered by thresholds — and thresholds can be designed around.

1. Shrink the package, not the price. Dropping a size tier is the single highest-leverage change available to most sellers. Reducing a carton by half an inch to clear a tier boundary can beat a year of price optimisation, and it compounds: a lower fulfillment fee also means a smaller 3.5% surcharge and less cubic footage in storage.
2. Price with the fulfillment bands in mind. The 2026 schedule bands small standard items under $10, $10–$50 and over $50. A SKU sitting at $9.85 and a SKU at $10.15 are not in the same cost structure. Test both sides of a boundary before assuming the higher price wins.
3. Check referral tier boundaries before repricing. Split-rate categories change the marginal rate above a threshold — in Electronics Accessories, the portion above $100.00 is charged at 8% rather than 15%, which makes the economics of a bundle materially different from two single units.
4. Manage weeks-of-cover in both directions. Too much inventory triggers storage and aged-inventory charges; too little triggers the low-inventory-level fee. There is now a billable penalty on both sides, so replenishment planning is a fee-reduction activity, not just an ops one.
5. Do the placement-fee arithmetic per shipment. Splitting a shipment across more destinations costs freight; not splitting it costs the inbound placement service fee. Which is cheaper depends on volume, weight and lane — calculate it each time rather than defaulting to one policy.
6. Move the worst offenders off FBA. Large, slow-moving, low-margin SKUs are where FBA's flat fees hurt most. Seller-fulfilled or Seller Fulfilled Prime can be cheaper per unit for those specific products, even while FBA remains right for the rest of the catalogue.
7. Audit reimbursements on a schedule. Lost and damaged units, dimension errors and fee miscalculations are recoverable, but only within Amazon's claim windows. An unaudited account leaves money behind by default.

None of these require permission from Amazon. They require knowing where the thresholds sit and having clean data on which SKUs are near them — which, for most catalogues above a few hundred SKUs, means pulling the numbers through the API rather than reading reports by hand. If that reconciliation work is eating your week, our eCommerce support team builds marketplace API integrations and fee-reconciliation pipelines for exactly this problem.