Plan FBA Restocks per FNSKU, Not per Parent ASIN | Inventory Hero
·17 min readInventory Planning
Plan FBA Restocks per FNSKU, Not per Parent ASIN
Parent-level cover hides a thin child variation. How to set reorder points, buffers, and order mix per FNSKU, and where the low inventory level fee may land when you get it wrong.
T. Brian Jones is co-founder and CTO of Inventory Hero. He leads the engineering behind its Amazon data pipeline, demand forecasting, and the AI platform that lets sellers talk to their live inventory, sales, and supplier data in plain language.
Sellers and seller-tool publishers report that Amazon moved the calculation from the parent-ASIN level to the individual seller FNSKU level on January 15, 2026, as part of that year's US FBA fee changes. We could not verify that against a primary Amazon source, because the low-inventory-level fee help page sits behind a Seller Central login, so treat it as reported rather than confirmed and check your own fee preview. The planning answer does not depend on it either way: measure days of supply per child variation, because a well-stocked color already hides a thin one from you regardless of how the fee is assessed.
Can one variation stock out while the rest of the parent looks fine?
Yes, and that is the normal case in apparel, multipack, and multi-color catalogs. Demand almost never follows the ratio you bought, so the fast children draw down to nothing while the slow ones sit. The parent's aggregate days of supply stays healthy the entire time. If the low-inventory-level fee is assessed per FNSKU as reported, that thin child would also carry the extra per-unit charge while its siblings ship at the normal fulfillment fee.
How do I see historical days of supply for each child variation?
Open Inventory then FBA Inventory in Seller Central. Amazon publishes a historical days of supply column per product there, and it is per-SKU, so you are already looking at child-level data. Pair it with Business Reports, Detail Page Sales and Traffic by Child Item, for child-level units ordered so you can compute your own velocity per FNSKU.
What days of supply should I plan to per child SKU?
Amazon's published trigger for the low-inventory-level fee is 28 days of historical supply. That is a floor to stay clear of, not a target to aim at, so plan to an average cover of roughly 30 to 40 days per child, closer to 40 if your replenishment timing slips. Some 2026 write-ups report the trigger moving to 35 days for certain size tiers; we could not confirm that against Amazon's own help page, so verify your own threshold and charges in your FBA fee preview and monthly storage-and-fees report.
Plan restocks per child variation, not per parent ASIN. Every FNSKU has its own velocity, its own lead-time demand, and its own reorder point, and the moment you plan at the parent level you are averaging a thin child together with a deep one and buying for neither. That is true no matter how Amazon prices your fulfillment. This article is about how to actually do it: reorder points, buffers, and order mix, per child.
Because aggregation hides the outlier, and one number for a variation family is an average of children that do not behave alike.
Here is a three-color hoodie parent, all large standard, with 1,185 total units in FBA:
Child (FNSKU)
Units in FBA
Units/day (30d)
Days of supply
Thin?
Black / M
310
17.4
17.8
Yes
Navy / M
480
7.6
63.2
No
Olive / M
395
4.2
94.0
No
Parent (aggregate)
1,185
29.2
40.6
Looks fine
The parent reads 1,185 / 29.2 = 40.6 days of cover, which looks like a business with nothing to do this week. Black reads 310 / 17.4 = 17.8 days and is the only child anyone should be looking at. It is also your best seller, so it is the child shipping the most units, which means every consequence below lands on your highest-volume SKU first.
Four things go wrong when you plan against the 40.6 instead of the 17.8:
Lost sales and rank on the thin child. Black runs out first, and it is the color carrying the listing's velocity and ad performance.
Stranded cash in the siblings. Navy and Olive are holding units that black needed. The parent is not over-invested in aggregate. It is invested in the wrong children.
Aged-inventory exposure on the deep ones. Olive at 94 days of cover is already walking toward the aged-inventory surcharge while you are worrying about a stockout.
Fee exposure per FNSKU, if the reported change holds. Sellers and seller-tool publishers report that as of January 15, 2026 the low-inventory-level fee is assessed per seller FNSKU rather than per parent ASIN, which would put black under the trigger while its siblings ship clean.1
FNSKU
Fulfillment Network Stock Keeping Unit. The barcode Amazon assigns to a specific seller's specific product variation. Two colors of the same shirt are two FNSKUs, and Amazon's inventory reporting is already per-FNSKU even when your planning is not.
Two numbers, used for two different jobs, and confusing them is the most common planning error here:
Threshold: 28 days. Amazon's published trigger for the low-inventory-level fee.2 It is a line to stay clear of.
Target: roughly 30 to 40 days of average cover per child, closer to 40 if your replenishment timing slips or your lead time is noisy. That is our buffer recommendation, not a published Amazon figure.
The rest of this article says "threshold" for the first and "target" for the second, and does not use them interchangeably.
Each FNSKU needs its own velocity and its own safety stock. The lead time is usually shared, since the children ride the same PO and the same container, but nothing else is.
Child reorder point = child daily velocity x lead time + child safety stock
Take the hoodie parent above with a 45-day door-to-Amazon lead time and a 10-day safety floor per child:
Child
Velocity
Lead-time demand
Safety stock (10 days)
Reorder point
Black / M
17.4/day
783
174
957
Navy / M
7.6/day
342
76
418
Olive / M
4.2/day
189
42
231
Black's reorder point (957 units) is more than three times its current on-hand. It should have been reordered a long time ago. Size your own per-child numbers in the reorder point calculator, and size the buffer properly with how much safety stock to hold.
Now the piece specific to the fee. Historical days of supply is an average over each window, not the low point of your inventory sawtooth. A single thin day does not trigger anything; a sustained stretch of thin cover does. So the thing to plan is your average cover, which sits roughly midway between your replenishment peak and your safety floor:
Average cover = (peak days of cover + floor days of cover) / 2
If your floor is 10 days, a peak near 60 days puts the average at (60 + 10) / 2 = 35, inside the target band. That backs directly into an order quantity per child of roughly (peak days minus floor days) x velocity, so black needs about (60 - 10) x 17.4 = 870 units on the next PO. Run the same shape for the whole parent in the FBA restock calculator.
Two caveats before you lean on that midpoint.
First, it assumes your reorder cycle is roughly the same length as the measurement window. If you replenish black every 50 days, the trailing 90-day window sees close to two full sawtooths and the midpoint holds. If you replenish once a quarter or you are inbounding in dribs, the window sees one lopsided slice of the cycle and the average can land well below the midpoint. Check the actual column rather than trusting the model.
Second, and this is where a lot of otherwise sound restock plans trip themselves: a 60-day peak over a 10-day floor will read under the threshold on the trailing 30-day window for part of every cycle. That is expected and it is not, by itself, a problem, because the fee requires both the 30-day and the 90-day windows to sit below the line.2 The 90-day window is the one that smooths your sawtooth, so it is the number to manage to. Watch the 90-day column, treat a dip in the 30-day column as normal mid-cycle behavior, and only panic when both are heading the same direction.
Rebase the mix on observed sell-through per child and stop repeating the ratio you bought last time. This is the real failure mode, and it is worth being precise about where it happens.
No Amazon report is doing the bad aggregation for you. FBA Inventory and the restock reports are per-SKU already. The aggregation happens in your own planning: in the parent-level cover number you glance at, and above all in the PO, where the easiest thing in the world is to re-cut last order's ratio and move on. That is the actual mistake. Demand does not follow the mix you bought, so after two or three cycles the same ratio has produced one child running thin and two children heading toward the aged-inventory surcharge. The mix, not the total, is the failure.
A workable rule of thumb, hedged because your category and seasonality will move it:
Rebase on trailing 90-day child unit share, not on the last PO. Pull units ordered per child, convert to a percentage of parent units, and cut the next PO to that ratio. The 90-day window matches the fee's long window and smooths a single promo spike.
Re-cut whenever a child's share has moved more than about 5 percentage points since your last order. Below that, drift is noise and the MOQ hassle is not worth it. Above it, repeating the old ratio is a decision to strand units.
Skip any child whose cover exceeds about twice your lead time. With a 45-day lead time, a child sitting above 90 days of supply does not go on this PO at all, whatever the size curve says. Olive above fails this test.
Fund the exposed fast movers before the deep slow ones when the PO is cash-capped. The thin child is the one losing sales now and the one shipping the most units, so it gets the constrained dollars first. Segment your catalog so this is a rule rather than a judgment call each cycle.
New children need a different rule. A variation with under 90 days of clean sales history has no trailing 90-day share to rebase on. Use whatever window you do have (28 days minimum, out-of-stock days excluded), seed it with the share of the closest existing sibling, buy shallow, and re-cut every cycle until the child has a real 90-day history.
Your supplier's minimum order quantity per colorway is usually what forces the bad mix in the first place, so here is what it actually looks like.
Trailing 90-day share says black 45 percent, navy 30 percent, olive 25 percent. Cash supports an 800-unit PO. At target mix that is 360 black, 240 navy, 200 olive. Your supplier's per-color MOQ is 500. Every line fails. Your real options:
Buy one color at MOQ and defer the rest. 500 black, nothing else, this cycle. Black is the only child near its reorder point anyway, and this comes in under budget at 500 units, freeing cash. This is usually the right answer and almost nobody picks it, because it feels like an incomplete order.
Stretch the PO to 1,500 to hit MOQ on all three. Nearly double the cash, and it pushes navy toward 130 days of cover and olive well past 200. You would be buying an aged-inventory problem to avoid an awkward conversation.
Negotiate a blended minimum. A 1,500-unit total commitment with flexible per-color splits is a much easier ask than a per-color reduction, and it is the ask that actually solves this for good. Worth doing before the next cycle, not during this one.
Then round for the case pack. Suppliers ship in cartons, often 24s, so a computed 500 units becomes 21 cartons at 504. Trivial at black's velocity. Not trivial at olive's, where a single carton is nearly six days of cover, and rounding two low-velocity children up a carton each is how a tight PO quietly grows 10 percent.
The slow children usually are not the problem, which is the opposite of most sellers' instinct.
Reported exemptions include products shipping fewer than about 20 units in a 7-day period, new-to-FBA products for a window after first inventory received, new professional sellers for a period after their first receipt, SKUs largely replenished from Amazon Warehousing and Distribution (a genuine planning lever if you are already using AWD, since the replenishment source, not just the quantity, can change your exposure), and (as of 2026) the grocery category.3 We could not verify any of these against Amazon's own help page, which is behind a login, so confirm your own status before you plan around them.
If the 20-units-in-7-days exemption is accurate, the picture inverts. A child selling 2 units a day ships 14 units in 7 days and would fall under the low-volume line. A child selling 17 a day ships around 120 and would not. Exposure would concentrate in the middle and top of your velocity range: the children that sell well enough to matter and get under-ordered because the parent looked fine. Triaging a large catalog on that basis (start with children doing more than about 3 units a day whose cover is under your target, and skip the long tail) is a reasonable first pass, but understand what it is: an inference stacked on an unverified exemption. If the exemption is wrong, or if your category is treated differently, your slow tail is exposed too. Spot-check a handful of low-velocity SKUs against your storage-and-fees report before you commit to ignoring them.
Four places, and only one of them is the dashboard most sellers look at:
Inventory > FBA Inventory (Manage FBA Inventory). FNSKU-level on-hand, and Amazon publishes a historical days of supply column per product here. This is the number the fee is assessed against, so it is your primary exposure check, not something to reconstruct yourself.4
Reports > Business Reports > Detail Page Sales and Traffic by Child Item. Child-level units ordered. The parent-item view is the one that lies to you. Use the child view to compute velocity per FNSKU and to build the trailing 90-day mix share for your next PO.
Reports > Fulfillment > Inventory Ledger. Daily on-hand and receipts per FNSKU. Use it to find the days a child was at zero.
The FBA fee preview and your monthly storage-and-fees report. These show whether a given SKU was actually charged and how much. Amazon's per-unit amounts vary by size tier and by how far under the threshold you sit, so pull your own live figure rather than trusting any published table, including this article's sources.
Amazon computes days of supply using units shipped over the whole window, out-of-stock days included. Your own buying math should exclude the days a child sat at zero, because those days measure your supply chain, not your demand. A child that sold 420 units in 30 days but was out of stock for 9 of them is selling 20 a day, not 14.
Both numbers are right, for different jobs:
Amazon's number is your exposure. It is what the fee is assessed against and what your weekly triage sorts on. Do not "correct" it.
Your out-of-stock-adjusted number is what to buy. Use it in the reorder point and the order quantity. Plan on 14 a day for a child that really sells 20 and you will under-order it again next cycle, which is how a SKU stays permanently thin.
Fifteen minutes a week, plus one deeper pass per replenishment cycle:
Weekly: sort FBA Inventory by the 90-day historical days of supply column, ascending, at the child level. The 90-day column is the one to flag on, because raw current cover will look alarming at the bottom of every normal sawtooth and tell you nothing. Anything under your target is on the list. Anything under the threshold is already exposed.
Weekly: recompute velocity per flagged child, out-of-stock days excluded via the Inventory Ledger, for buying purposes.
Weekly: compare each flagged child against its own reorder point, not its parent's. If it is under, it goes on the next PO regardless of what the rest of the parent looks like.
Per replenishment cycle: check mix drift. Trailing 90-day child share versus your last PO ratio, re-cut if any child moved more than about 5 points, then resolve MOQ and case-pack rounding before the order goes out.
Monthly: reconcile against the storage-and-fees report. It is the only place that tells you whether your planning actually worked or whether you are still paying.
This cadence sits inside the wider FBA restock planning rhythm rather than replacing it. The only thing it changes is the altitude: every one of these checks runs at the child level.
Parent-level cover is an average, and averages are exactly the wrong tool for a variation family where one color does half the volume. Give every child its own velocity, its own reorder point, and its own buffer; target average cover comfortably above the threshold rather than at it; manage to the 90-day window; and rebase your order mix on trailing 90-day child share instead of repeating the last PO. Do that and the reported per-FNSKU fee assessment stops mattering to you, because you were never relying on a healthy sibling to cover a thin one. Then go verify your own numbers: the historical days of supply column in FBA Inventory tells you where you stand, and the monthly storage-and-fees report tells you what it has already cost. For what the fee is and why it exists, the low-inventory-level fee explainer and the glossary definition cover it.
Amazon announced its 2026 US referral and FBA fee changes with an effective date of January 15, 2026 (Amazon Selling Partners, "Update to U.S. Referral and Fulfillment by Amazon fees for 2026," sellingpartners.aboutamazon.com). Amazon's own fee-detail pages are behind a Seller Central login and could not be read directly for this article, so we have no primary source for what changed inside them. The specific claims that the low-inventory-level fee moved to per-seller-FNSKU assessment on that date, that small bulky and large bulky products came into scope, and that grocery became exempt come from seller-tool publishers reporting the change: Seller Snap, "Amazon Fee Changes 2026" (sellersnap.io, accessed August 2026) and AMZ Prep, "Amazon Low Inventory Level Fee Explained" (amzprep.com, accessed August 2026). These sources are independent of each other but may share an upstream summary, and note that "FNSKU-level" is also commonly used to describe assessment on your units of an ASIN versus another seller's, which is a different claim from child-versus-parent. We have treated the change as reported rather than established. Confirm against the Seller Central low-inventory-level fee help page in your own account. ↩
Historical days of supply is average daily inventory units divided by average daily units shipped, evaluated over a trailing 30-day (short-term) and a trailing 90-day (long-term) window, with the fee applying only when both are below the threshold. Inventory Hero's reference-data registry carries the 28-day threshold and nothing else from this footnote (fba.threshold.lowInventoryLevelFee.daysOfSupply, effective 2024-04-01); the dual-window mechanic described here comes from the third-party explainers cited below, not from the registry. Some 2026 write-ups report the threshold moving to 35 days for certain size tiers; we could not confirm that against a primary Amazon source and have not treated it as fact. This article's target of roughly 30 to 40 days of average cover is a buffer recommendation, not a published threshold. Sources: Amazon Seller Central low-inventory-level fee help page (reference GV43F6S76Y9DHYRH, login-gated); Red Stag Fulfillment and Scale Insights explainers (accessed August 2026). Amazon does not publish a durable public per-unit rate table we can cite, so no per-unit dollar figures appear in this article. Check your own FBA fee preview and monthly storage-and-fees report. ↩↩
Reported low-inventory-level fee exemptions include products with fewer than roughly 20 units shipped in the trailing 7 days, new-to-FBA products for a period after the first inventory-received date, new professional sellers for a period after their first receipt, SKUs largely replenished from Amazon Warehousing and Distribution, and grocery-category products as of 2026. Sources: AMZ Prep (amzprep.com) and Ecomclips (ecomclips.com), both accessed August 2026. These are third-party summaries of a login-gated Amazon help page; verify your own eligibility in Seller Central before planning around an exemption. ↩
Amazon's 2026 fee communications describe a historical days of supply column calculated per product on the FBA Inventory page, alongside new fee-visibility tooling (Amazon Selling Partners, "Update to U.S. Referral and Fulfillment by Amazon fees for 2026," sellingpartners.aboutamazon.com, accessed August 2026). ↩