Every Amazon seller who uses Fulfillment by Amazon eventually runs into the same wall: a number in Seller Central decides how much warehouse space the business is allowed to use, and that number is calculated from how well inventory performed in the recent past. The Inventory Performance Index, almost always shortened to IPI, is that number. It is one of the few Amazon metrics that converts directly into a hard operational constraint rather than a soft ranking signal.
The painful part is the lag. A score built from spring and summer behaviour sets the capacity a seller gets in the fourth quarter, which is exactly when extra space is worth the most. Planners who treat IPI as a quarterly scoreboard find out too late. Planners who treat it as a rolling forecast input usually keep their space.
In short
- IPI is a composite score, not a single ratio. Amazon’s published documentation describes it as built from sell-through, stranded inventory, excess inventory and in-stock performance, each measured over a trailing window.
- The score sets capacity, and capacity is measured in cubic feet, not in units. Bulky low-velocity items consume the allowance far faster than their revenue justifies.
- Thresholds and review dates change. Amazon has revised both the qualifying IPI number and the mechanics of how limits are granted more than once, so the only reliable figure is the one showing in the Capacity Monitor inside your own Seller Central account.
- Sell-through is the fastest lever because it responds within weeks, while excess inventory percentage can take a full quarter to unwind if the only tool used is organic demand.
- Stranded inventory is free points. It is usually a listing error rather than a demand problem, and fixing it moves the score without discounting a single unit.
What the inventory performance index measures
IPI is Amazon’s attempt to answer one internal question: is this seller using our warehouse space productively? Amazon owns a finite fulfillment network. Space given to inventory that sits still is space not given to inventory that moves. The index is the rationing mechanism that follows from that constraint.
Amazon’s Seller Central help pages describe the index as a score on a 0 to 1000 scale, updated continuously and snapshotted at defined review points. The score blends several underlying behaviours rather than averaging them in a published formula. Amazon has never released the exact weighting, and sellers should treat any precise weighting circulating in forums as reverse-engineered guesswork rather than documented fact.
What Amazon does publish is the list of inputs. Understanding those inputs matters more than chasing a formula, because each input has a different response time and a different cost to fix. Treating them as interchangeable is the most common planning error we see in seller accounts.
The four published inputs
The first input is excess inventory: the share of your FBA stock that Amazon considers to be more than it expects you to sell in a reasonable forward window. The second is sell-through rate, usually framed as units shipped over a trailing period divided by average units on hand. The third is stranded inventory, meaning units physically in an Amazon warehouse that no active listing can sell. The fourth is in-stock rate, which penalises the opposite failure: running out of a product that has proven demand.
Those four pull against each other on purpose. A seller who strips inventory down to avoid excess will eventually break in-stock rate. A seller who ships in a deep buffer to protect in-stock will drift into excess. The index rewards the narrow band in between, which is simply a formal way of rewarding accurate forecasting.
| Input | What it actually measures | Typical response time | Cost to fix |
|---|---|---|---|
| Sell-through rate | Units shipped against average units held over a trailing window | 2–6 weeks | Medium: usually margin given up on price or ads |
| Excess inventory | Share of stock above Amazon’s forward demand estimate | 4–12 weeks | High if cleared by discount, low if cleared by removal |
| Stranded inventory | Units in the network with no sellable active listing | Days | Near zero: mostly listing hygiene |
| In-stock rate | Availability of products with demonstrated demand | 3–8 weeks | Medium: working capital tied up in buffer stock |
Read that table as a priority queue rather than a reference list. Stranded inventory sits at the top because it costs almost nothing and resolves in days. Excess inventory sits at the bottom because it is slow and expensive. Sellers who start at the bottom, which usually means launching a panic discount, spend money to move a metric that would have moved anyway.
The wider marketplace context matters too. Capacity rationing is not unique to Amazon, and the same productivity logic shows up in different wrappers on other platforms, a pattern covered in our complete guide to selling on global e-commerce marketplaces. Sellers running three or four channels should expect each one to ration something, whether that is space, placement or payout timing.
How storage limits are calculated and reviewed
This is the part of the system that has changed most, and where outdated advice does the most damage. For several years Amazon ran a simple gate: clear a published IPI threshold at the quarterly checkpoint and receive unlimited standard storage, or fall below it and receive a capped allowance. That model was straightforward and widely documented, which is precisely why so much stale content still describes it.
Amazon subsequently moved US FBA sellers onto a capacity management model, where the allowance is expressed in cubic feet and granted on a rolling basis rather than as a pass or fail outcome. Under that model IPI remains an input, but it is one of several, alongside shipment defect history and forecast demand. Amazon’s own capacity documentation is the authoritative description of how the current model works, and the figures displayed in the Capacity Monitor are account specific.
Two practical consequences follow. First, a seller cannot infer their own allowance from a competitor’s or from a forum post, because the inputs differ per account. Second, the question “what IPI do I need?” has become less useful than the question “what is my current allowance and how much of it am I consuming?” The second question has an answer visible in the account today.
Cubic feet, not units
The shift from a unit mindset to a volume mindset catches out sellers in bulky categories. A pallet of compact high-turn accessories and a pallet of oversized seasonal goods can carry identical unit counts and wildly different cubic consumption. Under a volume-based allowance, the oversized line is quietly expensive even when it is profitable on a per-unit basis.
The planning fix is to carry a cubic-feet-per-week-of-cover figure next to the usual margin and velocity columns. A product that looks acceptable on contribution margin can look unacceptable once it is charged for the share of a scarce allowance that it occupies. That is not an argument for dropping bulky products. It is an argument for pricing the space they consume into the decision.
The review cadence trap
Because capacity is reviewed on a schedule and demand is not, there is always a window where the allowance reflects conditions that have already passed. A seller who cleans up a messy catalogue in late October may still be operating under an allowance set from a messy September. The score improves first and the allowance follows later.
Planners should therefore work backwards from the review point rather than from the peak. If capacity for the holiday quarter is informed by late summer and early autumn behaviour, then the cleanup work belongs in August and September, not in November when the shortage becomes visible.
| Model | How the allowance was expressed | What IPI did | Planning implication |
|---|---|---|---|
| Legacy IPI threshold model | Unlimited standard storage, or a capped quantity | Acted as a pass or fail gate at a published number | Hit the number by the checkpoint date and stop worrying |
| Capacity management model | A rolling allowance in cubic feet per account | Acts as one weighted input among several | Manage consumption continuously; there is no single date to clear |
Sellers weighing whether the whole FBA arrangement still suits their catalogue should read that table alongside our breakdown of Amazon FBA versus FBM fulfillment models. For some bulky, slow-moving or highly seasonal lines, merchant fulfillment removes the capacity question entirely rather than optimising around it.
Excess inventory and the aged surcharge trap
Excess inventory is the input most sellers understand least, partly because Amazon’s definition is forward looking. It is not “stock older than X days”. It is stock above what Amazon’s demand model expects you to sell within a defined forward horizon. A product with collapsing demand can tip into excess while its physical age barely changes.
That distinction explains a frustrating pattern. A seller cuts the price, sells a burst of units, and watches the excess figure barely improve. The burst raised recent velocity, which raised the forward forecast, which raised the threshold for what counts as excess. The underlying problem, too much stock relative to sustainable demand, has not been solved.
Aged inventory surcharges compound the problem
Separately from the index, Amazon applies surcharges to inventory that has been in the network beyond defined age bands. Amazon’s fee schedule sets out the bands and the rates, and both have been revised in successive fee updates, so the current published schedule in Seller Central is the only figure worth planning against. Treat any rate quoted in a blog post, including this one, as indicative rather than current.
The compounding effect is what hurts. Slow stock drags the index down, which tightens capacity, which makes it harder to bring in the fast-moving replenishment that would restore the index. Meanwhile the slow stock itself accrues surcharges every month it sits. The cost of indecision rises on two axes at once.
Removal versus liquidation versus discount
There are three standard exits and they are not equivalent. Removal orders return units to the seller or a third-party address and stop the surcharge clock immediately, at the cost of removal fees and the logistics of handling the goods somewhere else. Liquidation hands the units to Amazon’s liquidation partners for a fraction of value with minimal effort. Discounting keeps the units in the network and attempts to sell through.
The right choice depends on how much capacity pressure the account is under. Under severe pressure, removal usually wins even at a worse headline recovery rate, because it frees cubic feet immediately and stops the age clock. With plenty of headroom, a measured discount often recovers more value. The mistake is applying one policy to the whole catalogue.
| Exit route | Speed of capacity relief | Typical value recovered | Best used when |
|---|---|---|---|
| Removal order | Fast, limited by outbound processing time | Full goods value minus removal fees and onward handling | Capacity is the binding constraint and you have somewhere to put the stock |
| Liquidation | Fast, handled entirely by Amazon partners | Low, a small share of retail value | The product is genuinely dead and handling it yourself costs more than it returns |
| Price-led sell-through | Slow and uncertain | Highest when it works | Demand exists at a lower price and the surcharge clock is not close to the next band |
| Bundling or multipacks | Medium | Moderate, protects headline price on the single unit | You want volume out without resetting buyer price expectations |
Bundling deserves a note because it is underused. Combining a slow unit with a fast one moves cubic feet without publicly cutting the price of either, which protects the price history that drives future buyer expectations. It does require new listings and new prep, so it is a planned move rather than an emergency one.
Sell-through rate: the lever with the fastest payback
Sell-through is the input worth optimising first among the slow movers, because it responds within weeks rather than months. It is a ratio, which means it can be improved from either side: ship more units out, or hold fewer units on average. Most sellers reach instinctively for the first and ignore the second.
Holding fewer units is often the cheaper path. Shifting from quarterly replenishment to monthly replenishment on stable lines cuts average on-hand inventory substantially while shipping the same annual volume. The ratio improves without a single price change. The trade is more frequent inbound shipments and more prep work, which is a real cost but a predictable one.
Where advertising actually helps
Advertising raises the numerator, and on products with genuine demand it works quickly. The discipline is to point the spend at the units causing the problem rather than at the account’s best performers. It is extremely common to find ad budget concentrated on the fast-moving hero product, which is already selling through well, while the capacity-consuming laggards get nothing.
Run the exercise as a capacity audit rather than a profitability audit. Rank products by cubic feet consumed against units shipped, then ask which lines would respond to visibility. Some will. Some are slow because nobody wants them at that price, and no amount of spend fixes that.
Organic visibility is part of the same system
Products that rank well sell through faster, which protects capacity, which allows deeper stock on the products that deserve it. That feedback loop is why listing quality and ranking work belong in a capacity conversation at all. The mechanics of how placement is earned are covered in detail in our analysis of how Amazon really ranks products in 2026, and the short version is that conversion-weighted relevance does more for sell-through than most price cuts.
There is also a defensive angle. A product that loses visibility keeps consuming space while shipping fewer units, so a ranking slide shows up in the capacity numbers a few weeks later. Treating a ranking drop as purely a revenue problem understates it.
Stranded inventory and fixable listing errors
Stranded inventory is the cheapest win available in this entire system, and it is routinely ignored because it is boring. These are units sitting in an Amazon warehouse that no live listing can sell. They consume space, they age, they drag the index, and they generate zero revenue. The cause is almost always administrative.
Common causes include a listing deleted or closed while stock remained in the network, a suppressed listing missing a required attribute, a pricing error that triggered a listing deactivation, a category or compliance flag, and SKU mismatches introduced during a catalogue migration. Each has a fix that takes minutes per SKU.
Building a weekly hygiene routine
The Fix Stranded Inventory report in Seller Central lists affected SKUs with a stated reason. A weekly fifteen-minute pass through that report prevents the slow accumulation that turns into a three-figure SKU cleanup at the worst possible moment. Set a recurring calendar entry rather than relying on noticing it.
Amazon also applies automatic actions to stranded units after a set period if the seller takes none. Those defaults are documented in Seller Central and can be configured, so a seller who would rather have stock returned than disposed of should check the setting before it matters rather than after.
The catalogue migration failure mode
The single largest stranding events we see follow catalogue work: a bulk flat-file upload with a column misaligned, a SKU renaming exercise, a move between seller accounts, or an ERP integration that writes a different identifier than the one Amazon holds. Hundreds of SKUs can strand in one afternoon.
The mitigation is to treat any bulk catalogue change as a deployment. Take a before snapshot of active listings, make the change on a small subset first, verify, then roll out. Check the stranded report the following day rather than the following month. This is unglamorous process work and it prevents the worst capacity emergencies.
Recovery plan for a store already over the limit
When an account is already consuming its full allowance and inbound shipments are being restricted, the sequence matters more than the individual tactics. Doing the right things in the wrong order wastes weeks that the account does not have.
Week one: free space without spending money
Start with stranded inventory, because it costs nothing and resolves fastest. Work the full Fix Stranded Inventory report, relist what can be relisted, and raise removal orders for anything that cannot. Then pull the aged inventory report and identify every unit already in or approaching a surcharge band.
Next, raise removal orders on the clearly dead tail. The instinct to wait for a better recovery rate is usually wrong under capacity pressure, because every week of waiting is a week of surcharge accrual plus a week of blocked replenishment on products that actually sell.
Week two: fix the inbound mix
Stop sending bulky low-velocity stock into the network. If a product takes more than a defined number of weeks to sell through, route it to merchant fulfillment or to a third-party warehouse for the time being. The allowance should be spent on the products with the best units-shipped-per-cubic-foot ratio.
At the same time, shorten replenishment cycles on the strong lines. Smaller, more frequent shipments hold the same service level on less average inventory. This is the change that holds the improvement in place after the emergency clears.
Weeks three to six: demand work on the middle tier
The middle tier is the group that is neither dead nor fast: products with real but soft demand. This is where advertising, bundling and modest price work earn their keep. Avoid a blanket discount across the catalogue, which gives away margin on products that would have sold anyway.
Competitive positioning belongs in this phase too, because a product losing the featured offer position sells far fewer units at the same inventory level. The tactics for holding that position without surrendering margin are set out in our guide to winning the Amazon buy box without slashing your margin, and they tend to be cheaper than the price cuts sellers reach for under pressure.
Ongoing: measure the right thing
After the emergency, the metric to institutionalise is capacity consumption as a percentage of allowance, tracked weekly, alongside a simple forward projection. A seller who knows they will hit ninety percent of allowance in five weeks has five weeks to act. A seller who only looks at the index finds out when the inbound restriction appears.
Planning inventory around peak season caps
Peak season inverts the usual logic. For most of the year, excess inventory is the enemy. In the run-up to a major sales event, being out of stock on a product with demonstrated demand costs far more than the surcharge on holding it. The index punishes both, but the revenue consequences are asymmetric.
The resolution is to earn headroom early. Capacity available in the fourth quarter is a function of behaviour in the preceding months, so the cleanup described above has to happen in late summer rather than in the peak itself. Sellers who clear the dead tail in August buy themselves room in November.
Event calendars are capacity events
Major platform events compress months of demand into days, which means the inventory decision is made weeks before the event and cannot be revised during it. The scale of that compression is visible in coverage of events such as Amazon’s Prime Big Deal Days across 22 countries, where the selling window is measured in hours rather than weeks.
Build the capacity plan around the shipping deadline for the event, not the event date. Working backwards from the deadline through carrier transit time, Amazon receiving time and prep time typically moves the real decision point four to six weeks earlier than sellers expect.
The post-peak hangover
The weeks after a peak event are where index damage is actually done. Unsold seasonal stock becomes excess inventory almost immediately, because the forward demand forecast collapses the moment the season ends. A product that was correctly stocked in November is in excess by early January without a single unit changing.
Plan the exit before the entry. Decide in advance what happens to leftover seasonal units: removal to a third-party warehouse for next year, liquidation, or a January clearance. Having the decision pre-made avoids the two months of hesitation that turn a manageable overhang into a surcharge problem and a capacity restriction heading into spring.
The broader discipline here is the same one that applies across every marketplace a brand sells on, and it is worth revisiting the channel-level view in our guide to global marketplace selling when deciding how much of a seasonal bet to place inside any single platform’s warehouse network.
A simple pre-peak checklist
Eight to ten weeks out, run the stranded report and clear it completely. Six to eight weeks out, remove or liquidate the dead tail and confirm current allowance consumption. Four to six weeks out, place the peak inbound shipments with transit buffer. Two weeks out, stop sending anything that is not a proven seller.
Through the event itself, monitor sell-through daily rather than weekly, because the forecast that drives post-event excess classification is being written in real time. Immediately after, execute the pre-made exit decision rather than reopening the debate.
What to verify before acting on any of this
Amazon changes fee schedules, capacity mechanics and reporting surfaces regularly, and it does so with limited notice. Several of the structural points in this article have already been revised at least once since the index was introduced. Everything here is general information about how the system works, not a current rate card.
Before making an inventory commitment, confirm the current numbers in three places inside your own account: the Capacity Monitor for the live allowance and consumption, the FBA fee schedule for current storage and surcharge rates, and the inventory performance dashboard for the score inputs and their current values. Amazon’s own Seller Central documentation is the authoritative source, and it supersedes any third-party summary including this one.
It is also worth separating the index from the underlying business question. IPI is a proxy for something real, which is whether working capital is tied up in stock that moves. The general principle behind it, often discussed as inventory turnover, predates Amazon by a century and applies whether or not a platform is scoring you on it. A seller who runs clean turns rarely has an IPI problem to solve.
FAQ on IPI and FBA limits
What IPI score do I need to avoid storage limits?
There is no single number that holds across time or across accounts. Amazon has published and revised qualifying thresholds under the legacy model, and under the current capacity management model the allowance is calculated from several inputs rather than a single pass mark. Check the Capacity Monitor in your own Seller Central account for the figure that applies to you right now.
How often is the IPI score updated?
The score itself moves continuously as the underlying inputs change, while capacity decisions are made at defined review points. That gap is why improvements can appear in the score before they appear in the allowance. Amazon documents the current review cadence in Seller Central, and it has changed more than once.
Does removing inventory hurt my score?
Removals reduce the units on hand, which generally helps both the excess inventory input and the sell-through ratio. The risk is removing too deeply on products with real demand and then breaking in-stock rate, which is itself a scored input. Remove the dead tail aggressively and the proven sellers cautiously.
Why did my excess inventory percentage barely move after a big sale?
Because the excess definition is forward looking. A sales burst raises the recent velocity that feeds Amazon’s forward demand estimate, which raises the amount of stock considered reasonable to hold. The percentage improves only when holdings fall relative to sustainable demand, not relative to a one-off spike.
Is stranded inventory really worth chasing?
Yes, and it is the first thing to fix. Stranded units consume cubic feet, accrue age, and drag the index while generating no revenue at all. Most cases are listing errors that take a few minutes each to resolve, which makes it the highest return per hour of any action in this article.
Do capacity limits apply to every FBA seller?
Capacity management applies to FBA sellers in the markets where Amazon has rolled it out, and the mechanics differ by region. European and North American programmes have not always moved in lockstep. Sellers operating in multiple regions should check each marketplace separately rather than assuming the home market’s rules apply everywhere.
Can I buy additional capacity?
Amazon has operated mechanisms allowing sellers to request or bid for additional capacity under the capacity management model. Availability, cost and eligibility have varied, and any such mechanism is documented in Seller Central rather than guaranteed. Treat purchased capacity as a contingency, not as the plan.
Should I switch to merchant fulfillment to escape the problem?
For bulky, slow-moving or highly seasonal lines, merchant fulfillment or a third-party warehouse removes the capacity constraint and the surcharge exposure in one move. The trade is losing Prime-badge advantages and taking on fulfillment operations yourself. A hybrid split, with fast lines in FBA and slow or bulky lines fulfilled elsewhere, is the most common resolution.
How far ahead should I plan for the holiday quarter?
Work backwards from the inbound shipping deadline rather than the event date, and add Amazon receiving time plus carrier transit plus prep. In practice that puts the real decision point four to six weeks before the event, and the capacity cleanup that enables it another four to six weeks before that.