Amazon fees

The 2026 Amazon fees that changed your reorder math

Four of the fee mechanics hitting an FBA account are driven by how much you send and when. Those are the two things a purchase order decides, which makes them a planning input, not a billing surprise.

The short version

Low inventory level, aged inventory and inbound placement are fees; capacity is a limit that behaves like one. All four are functions of quantity and timing. Change the reorder quantity, change the fee.
Running thin is priced and running deep is priced, so the output of planning is a band, not a number.
Capacity is spent in cubic feet, one pool per storage type, so a bulky slow mover charges rent to every SKU that shares its storage type, your best seller included.
Look up your own fee schedule and date it. This post prints no rates on purpose.

It is easy to read Amazon’s fees the way you read a utility bill. The number turns up after the month is over, you make a face, and you file it. Fees are accounting, at the far end of the spreadsheet.

That gets expensive for one reason. Several of the 2026 fee mechanics are not charged on what you sold. They are charged on how much you sent and how long it sat, which are the two variables a reorder decision sets. The fee is not something that happens to your purchase order. It is something your purchase order chooses.

First: there are no fee rates in this post

You will not find a dollar figure, a threshold or a percentage below, and that is deliberate. Amazon’s schedule moves and this page does not, and a stale rate is worse than none, because people do not read it and move on, they build a spreadsheet on it. Where I write “the published threshold” or “the first age band”, I mean the number in your own Seller Central fee schedule today, which you should look up and date.

A fee is a price on behaviour, not a punishment for bad luck

Amazon is not fining you. Amazon runs a warehouse network with finite space and a promise about delivery speed, and each of these fees buys a behaviour change with your money. The fee is what you pay to keep doing the old thing.

Read a fee as bad luck and you file it. Read it as a price and you ask what it is buying, and whether you want to sell it. Then the planning response falls out, because each fee pulls one lever.

The mechanic	The behaviour it prices	Your lever
Low inventory level fee	Holding too little against your sales rate	The floor: days of cover you never breach
Monthly storage	Occupying volume every month, sold or not	The ceiling: how much you send in one go
Aged inventory surcharge	Holding the same physical unit past a date	Quantity on slow movers, and when you stop holding
Inbound placement	Sending everything to one place	How you split a shipment, and how early you commit
Capacity limits	Total volume per storage type, not per SKU	Which SKUs get the cubic feet

Every lever in that right hand column is pulled when you decide a quantity and a date, not in the settlement report. By then the decision is eight weeks old and on a boat.

A fee that moves with the quantity you send is not a cost. It is a price on a decision you are already making, every time you approve a purchase order.

Thin is priced, deep is priced, so the answer is a band

This is the part I would keep if you threw the rest away.

The low inventory level fee prices running thin. It is assessed on units you ship while both the 30-day and 90-day historical days of supply for that SKU sit below the published threshold, so it reads a trailing window rather than today’s snapshot. If that is how it works on your account, a dip is not a one day event. You keep paying through the recovery, because the history the measure reads still contains the hole.

Storage and the aged surcharge price the opposite behaviour. Every extra unit is volume rented monthly, and the ones at the back of the pile age into a surcharge. Two costs pulling opposite ways along one axis, days of cover: expensive on the left, expensive on the right, and a wide flat floor in the middle where moving quantity by a few days just trades a little of one fee for a little of another. The honest output is a band, and anywhere inside it is fine.

Working out the floor

The floor is the level you plan never to go below. Here is the arithmetic on a SKU I will call Brand A’s core unit. It sells about 12 a day. I order on a 30 day cycle. Lead time from purchase order to sellable is 46 days: 30 to produce, 12 in transit, 4 from delivery to check in. Across my last eight receipts, the slowest ran 9 days longer than the median. So the inventory position I have to cover, counting on hand plus everything on the water, is:

Review interval: 30 days
Lead time to sellable: 46 days
Observed spread: 9 days
Sub total: 85 days of cover, before the threshold is considered at all

Add the published threshold underneath and that is your target. That target is this article’s planning translation, not an Amazon rule: Amazon measures the fee on historical days of supply, and the target is one way to stay clear of it. The threshold is Amazon’s and is not negotiable. The 85 days is yours, and it is the part I would work on first. Halve the review cycle and 15 days come out. Cut check in time and more comes out. Every day removed from that 85 is a day of cover you never fund again, on every SKU.

Working out the ceiling

The ceiling is not a cost calculation. It is a date lookup. Take the quantity you are about to order and ask when its last unit sells, not at your forecast but at the pessimistic rate a soft quarter produces. Say the supplier minimum pushes me to 900 units. At 12 a day the last unit sells on day 75. At 9 a day, 900 divided by 9, it sells on day 100. The real question is not whether 900 feels like a lot. It is whether day 100 sits on the wrong side of an age band. The surcharge is assessed monthly against how long a unit has sat in a fulfilment centre, in bands that step up as the age increases, which makes it a step function, not a slope. Day 100 either crosses a line or it does not, and you can look that up before ordering instead of discovering it five months later. Cut to 600 and the pessimistic tail clears on day 67.

The gap between what the review cycle wants, 30 days of demand or 360 units, and what the minimum forces, 900, is where the aged inventory in this example is born. Not carelessness. Minimums and container economics.

When the band is empty

Sometimes the floor lands above the ceiling: the threshold plus 85 days is more cover than you can hold without the tail crossing a band. That is not a spreadsheet error, it is a finding. The SKU cannot be planned profitably as configured, and there are four honest answers.

Shorten the lead time. Air a portion, dual source, move production closer. The only answer that fixes the floor rather than dodging it.
Shorten the review cycle. Smaller quantities, more often, paid for in inbound freight and handling.
Move the holding upstream. Buy the minimum but take delivery in tranches, so the supplier or your 3PL holds the balance where space is cheaper.
Take the SKU out of FBA, or out of the catalogue. Some products are not worth the shelf space they demand. That is a conclusion, not a failure.

What you should not do is average the two and order the middle. When the band is inverted, the middle is the worst of both.

Aged inventory: the surcharge is the visible part

When a unit crosses into a surcharge band, the line item on the report is the smallest cost in the room. It is just the one with a number next to it. The bigger cost is capital, and the way to feel it is to convert it into units rather than money. A pallet that will not clear for another hundred days is a specific number of units of your best seller that you did not buy. Say that out loud, in units, and the argument about whether to liquidate gets much shorter.

Inventory also ages because a forecast was wrong, and a wrong forecast does not correct itself while you wait. So make the hold or clear call forward looking only. What you paid is gone. The question is whether holding one more month, meaning storage plus surcharge plus the volume denied to something faster, costs more than what you would recover now.

Because the bands step rather than slope, this is a calendar job. A fourth quarter build also starts a large share of the catalogue’s age clock in the same fortnight, so it crosses every later band in the same fortnight too.

Inbound placement: you are buying convenience either way

Placement exists because the network would rather receive your inventory spread across several locations, which puts units near more customers before an order is placed. One destination is easier for you and worse for them, so there is a price on it. The fee is charged per unit when you send to a single location; Amazon’s partial-split option, for bulky items only, reduces it, and its Amazon-optimized split option removes it when your shipment meets the packing requirements. Check the options on your own shipment, because this mechanic changed again in January 2026.

Here is what gets missed. Pricing this as a fee against freight is incomplete, because splitting also changes your lead time distribution. One destination has one check in event. Four destinations have four, and they do not land together. Inventory stops becoming sellable on a date and starts becoming sellable across a range, with the slowest destination setting the range. Variance in lead time is a first-order input to safety stock, and if you have not seen how sharply that bends, walk through the safety stock calculation with your own receipt history.

So the cheapest placement option can be the expensive one. It pays for itself out of your safety stock, and safety stock pushes days of cover toward the ceiling. The fee is visible and the safety stock is not, which is why the fee wins arguments it should lose.

Two rules of thumb, to check against your own receipt history rather than assume:

Fast movers you replenish often: split. Sending every two or three weeks means variance averages out, and one late check in is covered by the last shipment still selling through.
Slow movers you send twice a year: think hard. Nothing to average against, so one late destination is a real hole.
Ask the supplier to pre split. Cartons palletised by destination at origin remove most of the operational pain, and the decision becomes a straight freight comparison.

Capacity is volume, and your slow movers are spending it

Capacity is granted to the account monthly, one limit per storage type, measured in cubic feet rather than units, and the amount is influenced by your IPI score and Amazon’s own forecasts rather than being fixed. The consequence holds whatever the numbers are: your SKUs do not each hold their own allowance. Every SKU in a storage type drinks from that storage type’s pool, and the pool is measured in space. So the right productivity metric is not units sold or margin per unit. It is sales per cubic foot of held inventory.

Two standard-size SKUs, so they share one pool:

SKU A is 0.2 cubic feet per unit and sells 300 a month, 10 a day. At 45 days of cover that is 450 units, times 0.2, so 90 cubic feet producing 300 sales. Call it 3.3 sales per cubic foot per month.
SKU B is 1.1 cubic feet per unit and sells 40 a month. Slow, so the supplier minimum leaves it on 180 days of cover: 240 units, times 1.1, so 264 cubic feet producing 40 sales. Call it 0.15 sales per cubic foot per month.

SKU B occupies nearly three times the space of SKU A, 264 against 90, to produce roughly a seventh of the sales, 40 against 300. No per unit margin report shows that, because per unit SKU B may look healthy. The cost surfaces only when the pool runs out and something has to be cut, and what gets cut is usually whatever someone was looking at that week.

Run it the useful direction. Take SKU B from 180 days of cover to 90: 120 units instead of 240, so 132 cubic feet instead of 264, releasing 132. At SKU A’s density, that is 660 more units of A, and at 10 a day, 66 extra days of cover on your best seller, funded by not overstocking a slow one.

The same shelf space is either half a year of your slowest mover or sixty-six extra days of your best one. Capacity does not care which you chose. Your cash does.

Rank the catalogue by sales per cubic foot quarterly. The bottom of that list is where your headroom went.

The monthly routine

None of this needs to be a project. It needs to be a monthly routine.

Open the fee schedule and date it. The low inventory threshold, the age band boundaries, the placement options, your granted capacity. Monthly, not annually.
Recompute the floor per supplier, because spread belongs to the supplier and the route. It is the arithmetic behind a reorder point, with the threshold added as a floor you have chosen not to breach.
Recompute the ceiling on anything you are about to order. At the pessimistic rate, when does the last unit sell, and does that date cross a band? If so, the quantity comes down or the delivery gets staged.
Flag every SKU where the floor exceeds the ceiling. Those need a lead time fix, an upstream holding arrangement, or a conversation about whether they belong in FBA.
Rank by sales per cubic foot. Take the bottom ten and ask what that space would buy if it went to the top ten. Answer in units.
List every unit crossing an age band next month. A decision made a week early beats a perfect one made a week late.
Price placement against lead time to sellable, not the fee alone, then check whether that moves safety stock on those SKUs.
Write down what the schedule changed this month. If the answer is nothing, three months running, you are not really reading it.

What this does not fix

Fee aware planning is a discipline, not a solution, and I would rather say where it stops. It does not fix demand: every calculation above starts with a daily rate and a pessimistic daily rate, and if both are guesses then the floor and the ceiling are guesses wearing a suit. Nor does it fix a supplier minimum. If the factory will not go below 900 and 900 crosses a band, planning moves that to a commercial conversation about minimums or staged delivery, but it does not win that conversation for you.

And it does not remove judgement. Whether a soft quarter is a blip or a trend is the call that decides if the last unit sells on day 75 or day 100. SKU Compass keeps days of cover, lead time by supplier and a suggested reorder quantity across Amazon, Shopify and Walmart in one place, so the inputs to the floor and ceiling arithmetic are sitting there rather than being rebuilt each month. It does not carry Amazon’s fee schedule, and it will not tell you whether a soft month is noise. You still open the schedule yourself, and you still make the call.

These fees are not weather. They are prices attached to specific behaviours, each behaviour is set by a quantity and a date, and you choose both.

Questions people actually ask

Which Amazon fees can I actually change by reordering differently?

The ones driven by quantity and timing rather than by the sale: the low inventory level fee, monthly storage, the aged inventory surcharge and inbound placement, plus capacity, which is a constraint rather than a fee but behaves the same way. Referral and fulfilment fees are set by the product, so reordering does not move them.

How many days of cover do I need to avoid the low inventory level fee?

Take the threshold published on your own fee schedule, then add your review interval, your lead time to sellable, and the spread between your median receipt and your slowest one. In the example above that was 85 days on top of the threshold. I have not printed the threshold, because Amazon’s schedule moves and a stale number ends up inside somebody’s spreadsheet.

Is it cheaper to split an FBA shipment or pay to keep it in one place?

Compare three things, not two: the placement fee, the inbound freight, and the change to your lead time to sellable. More destinations means more check in events, and the slowest sets when inventory is really available, which can raise safety stock if it widens your measured lead-time variability. On a fast mover replenished fortnightly, one late destination is covered by the last shipment still selling through, so splitting is easier to justify. On a slow mover sent twice a year there is nothing to average against, so consolidation is easier to justify. Measure it on your own receipts rather than assume either.

Should I liquidate aged inventory or hold it and wait?

Decide forward only. What you paid is gone. Compare the cost of holding one more month, meaning storage plus any surcharge plus the volume denied to a faster SKU, against what you would recover by discounting, bundling or removing now. Because the surcharge steps rather than slopes, decide before the unit crosses the next band.

Does inventory software solve this?

It solves the bookkeeping half: days of cover and lead time by supplier in one place, with cubic feet per unit kept by you alongside them. SKU Compass does that part across Amazon, Shopify and Walmart. It does not solve the judgement half. SKU Compass does not carry Amazon’s fee schedule, so the threshold and the age bands still come from Seller Central by hand. Keeping those two current is your job either way.