Amazon inventory management is the practice of keeping enough stock in the network to sell without
holding so much that storage fees, surcharges, and capacity limits start working against you. It is a
balancing act with a penalty on both sides, which is what makes it harder than it sounds.
The two failure modes cost differently. Running out costs you sales and rank, and rank is slower to
recover than stock. Holding too much costs you storage fees, aged-inventory surcharges, and capacity
you cannot use for products that would sell. In our experience sellers fear the first and quietly bleed
from the second.
The metrics that actually matter
Four numbers. Everything else is downstream of these.
Inventory Performance Index
IPI is Amazon’s composite score for how well you manage inventory, and it feeds into the storage
capacity you are granted. It reflects excess inventory, sell-through, stranded listings, and in-stock
rates rather than any single one of them.
Amazon has changed how capacity is granted more than once, moving toward capacity limits measured in
cubic feet with IPI as an input rather than the sole gate. Check your current limit and its basis in
Seller Central rather than trusting a threshold quoted in an article, including this one, because the
mechanism has been revised repeatedly.
Sell-through rate
Units sold over a period against average units held. The single most honest measure of whether stock is working. It is the first number I look at on any account, ahead of revenue, ahead of IPI, and ahead of whatever the dashboard has decided to put at the top, because a high sell-through rate on modest volume describes a healthy operation while the reverse describes capital slowly turning into storage fees. Simple to calculate. Rarely looked at.
Low sell-through on a SKU means capital sitting still while incurring fees. It is also the number that
tells you a product is finished before your instincts do.
Weeks of cover
How many weeks your current stock lasts at current velocity. Somewhere between 6 and 10 weeks is a
common working target, and the right figure for you depends entirely on your lead time.
The arithmetic that matters: weeks of cover must exceed your total lead time plus a safety buffer.
If manufacturing takes 30 days and freight takes 40, a 6-week cover is already too thin, and no amount
of careful reordering fixes a structural gap.
Stranded inventory
Units in a fulfillment center attached to a listing that cannot sell: suppressed, inactive, or broken.
This is the cheapest problem on the list to fix and the most commonly ignored, because nothing
alerts you loudly.
Check it weekly. Our guide to
fixing inactive or suppressed listings
covers the causes, and I’ve found most stranded inventory traces to something that took 5 minutes to
repair once somebody looked.
The fees that punish bad inventory management
Worth naming, because they arrive quietly on a statement rather than as a warning. Nobody emails you about them.
Monthly storage fees, charged per cubic foot and higher in the final quarter of the year. Slow stock
costs more in Q4 precisely when space is scarcest.
Aged inventory surcharges on units held beyond defined age thresholds. These escalate with age and
are the mechanism that makes hoarding genuinely expensive.
Low-inventory-level fees on products held too thin relative to demand, which is the counterweight
Amazon added so that the answer to storage fees is not simply holding nothing.
Removal and disposal fees when you take stock out, which is why “just remove it” is a decision with
a cost rather than a free escape.
The trap is that the first two push you to hold less and the third pushes you to hold more. There
is no setting that avoids all of them. There is only a range, and finding your range is the actual work.
A working routine
What I’d do rather than what a dashboard suggests.
Weekly: check stranded inventory, check restock recommendations against your own lead times rather
than accepting them, and check for listings that have gone inactive.
Monthly: review sell-through by SKU and identify the bottom decile. Decide something about each one.
Deciding nothing is the default and it is what produces aged inventory.
Quarterly: review lead times against reality rather than against what your supplier says, and adjust
reorder points. In our experience supplier lead times drift longer without anyone announcing it.
Before Q4: get slow stock out before storage fees rise and capacity tightens. The decision is
cheaper in September than in November, and it is the same decision.
Reorder points, worked through once
The one calculation worth doing by hand before trusting any tool with it.
Reorder point = (daily velocity x total lead time in days) + safety stock.
Take a product selling 20 units a day. Manufacturing runs 30 days, sea freight 40, and receiving at the
fulfillment center takes another 7. Total lead time: 77 days. That is 1,540 units consumed before a new
order lands. Add a safety buffer of, say, 3 weeks at current velocity, another 420 units, and your
reorder point sits near 1,960 units.
Now the uncomfortable part. If you reorder at 800 units because that felt like plenty, you will stock
out for roughly 8 weeks, and no dashboard warned you because the dashboard was watching stock rather
than watching lead time.
Three refinements worth applying:
- Use trailing 30-day velocity, not lifetime average. A product’s past is not its present.
- Measure lead time from your own records, not from what the supplier quotes. In our experience the
gap runs weeks rather than days. - Widen the buffer before a known peak, and narrow it afterward rather than leaving it wide.
I’d redo this calculation quarterly per significant SKU. It takes minutes. When I have found a stockout
that nobody could explain, a stale lead-time assumption was behind it more often than a demand spike.
What a stockout actually costs
Worth quantifying, because sellers treat it as lost sales and it is worse than that.
You lose the sales, obviously. You also lose sales velocity, which feeds organic rank, and rank
recovers more slowly than stock arrives. A product out for 3 weeks does not resume at its previous
position the day inventory lands.
You lose advertising momentum too, since campaigns stop serving and the history they were building
pauses. And competitors who took your position during the gap now have reviews and rank they did not
have before.
That asymmetry is the argument for a slightly generous buffer, and it is why the cheapest-looking
inventory policy is rarely the cheapest one. Storage fees are visible and modest. Lost rank is invisible
and expensive.
What to do with stock that is not moving
Four options, in the order I’d consider them.
Discount and advertise through it. Cheapest if it works. It frequently does on a product with genuine demand and a pricing problem, and in my experience that describes more slow SKUs than sellers assume.
Create a removal order and take the units back. Sensible when you have another channel or the goods
are worth handling.
Use FBA Liquidations, where Amazon sells through its liquidation partners and returns a share of the
value. Less than you paid, more than disposal, and no handling. Our guide to
getting Amazon unsold items covers where that inventory
ends up.
Dispose of it. Sometimes correct, and the option delayed longest because it feels like accepting a
loss that has already happened.
Enable Grade and Resell separately, so eligible customer returns can be regraded and sold as used
rather than written off. In our experience it is the least-used setting with the clearest payback.
The upstream problem nobody wants to name
Inventory problems are usually purchasing problems wearing a logistics costume.
If you are liquidating repeatedly, the decision that needs fixing happened at the purchase order.
Forecasts built on optimism, minimum order quantities accepted because the unit price looked good, and
seasonal products ordered on last year’s peak all produce the same result several months later.
Two habits that help more than any tool. Forecast on trailing velocity rather than on ambition, and
treat a supplier’s minimum order quantity as a cost rather than a constraint, because that is what
it is when it forces you to hold 8 months of stock.
If you would rather have restock planning, capacity, and account health monitored than discovered, that
sits inside our
account management service. We are
an Amazon Ads partner and an Amazon SPN Verified Partner.
Software, briefly
Tools help with visibility and forecasting. None of them fixes a purchasing decision, and I’d be suspicious of any that claims to.
Seller Central’s own reports are sufficient for a small catalog, and I’d exhaust them before paying
for anything. Third-party tools earn their cost when you have enough SKUs that reading the reports by
hand stops being realistic, which our guide to
the best Amazon inventory management software
covers in detail.
When I have seen tools disappoint, it was because they were bought to solve a discipline problem rather
than a data problem.
FAQ
What is Amazon inventory management?
Keeping enough stock in Amazon’s network to sell without holding so much that storage fees,
aged-inventory surcharges, and capacity limits work against you. Both failure modes cost money, which is
what makes it a balancing act rather than a target.
What is a good Inventory Performance Index score?
Amazon has revised how IPI feeds into storage capacity more than once, moving toward limits measured in
cubic feet with IPI as an input. Check your current limit and its basis in Seller Central rather than
relying on a threshold quoted in an article.
How many weeks of cover should I hold?
Commonly somewhere between 6 and 10 weeks, but the figure only works if it exceeds your total lead time
plus a safety buffer. If manufacturing takes 30 days and freight 40, six weeks of cover is already too
thin regardless of any general guidance.
What is stranded inventory on Amazon?
Units sitting in a fulfillment center attached to a listing that cannot sell, usually because the
listing is suppressed, inactive, or broken. It is the cheapest inventory problem to fix and the most
frequently missed, because nothing alerts you loudly.
Which Amazon fees punish poor inventory management?
Monthly storage fees charged per cubic foot and higher in Q4, aged-inventory surcharges that escalate
with age, low-inventory-level fees for holding too thin, and removal or disposal fees when you take
stock out. The first two push you to hold less and the third pushes you to hold more.
What should I do with Amazon inventory that is not selling?
Discount and advertise through it if demand exists, create a removal order if you have another channel,
use FBA Liquidations to recover part of the value, or dispose of it. Enabling Grade and Resell
separately recovers value on returns that would otherwise be written off.
Last updated: August 31, 2026. Amazon’s IPI thresholds, capacity limit mechanism, storage fee rates,
aged-inventory surcharges, and low-inventory-level fees all change, and the capacity system has been
revised repeatedly. Seller Central carries the current figures and rules for your account.
