Amazon fines you for holding too much and for holding too little. The gap between them is your reorder point.

Key takeaways

• Two fees push in opposite directions: the low-inventory-level fee charges per unit sold below Amazon's 28-day threshold, and storage plus the IPI score penalize excess above roughly 45 days of cover.
• The corridor between them is narrower than a typical replenishment lead time, which is why this is an ordering-frequency problem rather than an ordering-quantity problem.
• Total lead time is longer than most sellers count: production, freight, customs, prep, Amazon receiving of a 1 to 3 business days, and your own decision lag.
• Reorder point = daily sales × total lead time in days, plus safety stock. Everything else in forecasting is refinement on those two terms.
• Three constraints sit outside the arithmetic: capacity limits you may not be able to exceed, DD+7 cash timing, and Q4 where every input gets slower and more expensive.

Your lead time is longer than the window Amazon leaves you to hold stock in. That sentence is the whole problem, and almost every guide on this site ends up here. The storage fees, the low-inventory fee, the IPI score, the capacity limit, the cash cycle and the peak season all converge on one question: how much, how often.

This is the arithmetic, assembled from the constraints the rest of the site documents.

01The corridor

Below a reported 28 days of cover, the low-inventory-level fee charges you per unit sold — a fee for selling while thin; the threshold is 28 or 35 days.

Above roughly 45 days, two things start working against you. Reported IPI guidance targets 45-day days-of-inventory to improve excess inventory percentage and sell-through simultaneously — see the IPI guide. And the storage stack begins compounding, with a utilization surcharge reported above 22 weeks of cover and an aged-inventory trigger at a 180 days.

So the target corridor is roughly 30 to 45 days of cover.

And that is the problem. A corridor of about 15 days, against a replenishment cycle measured in months.

02Total lead time, counted honestly

Most sellers count production and freight. The full chain:

Stage	Typical reported range
Your decision lag	0–14 days
Production	Supplier-specific
Origin prep and consolidation	2–7 days
Freight — ocean LCL door to FBA	35–55 days
Freight — ocean FCL door to FBA	30–45 days
Freight — air door to FBA	8–14 days
Customs clearance	Included above; a hold adds days to weeks
Amazon receiving	1–3 business days after delivery

See the freight guide for the mode figures and the customs holds guide for what a hold does to the timeline.

Two stages sellers routinely omit:

Your own decision lag. The days between the report saying reorder and the purchase order being sent. It is real, it is often the longest controllable stage, and it is invisible because nobody measures it.

Amazon receiving. Delivered is not sellable. One to three business days, longer in Q4.

03The calculation

daily sales   = units sold over the period ÷ days in the period

reorder point  = daily sales × total lead time in days
         + safety stock

order quantity  = daily sales × target cover
         (30–45 days, per the corridor above)

Safety stock covers variability, not average demand. A reasonable starting method: take your worst recent fortnight of daily sales rather than the average, and cover the difference across the lead time.

Two adjustments that matter more than precision in the average:

Order more often rather than more. If your lead time is 60 days and your target cover is 45, you cannot hold 45 days and reorder only when you hit the reorder point — the corridor is narrower than the cycle. The answer is overlapping orders: place the next order before the last one lands. That is uncomfortable and it is what the fee structure requires.

Use the worst month, not the average month. For anything seasonal, an average conceals both the stockout and the overstock.

04The three constraints outside the arithmetic

Capacity. You may not be permitted to send what the calculation says. Reported capacity is roughly five months of forecast sales, with ASIN-level restock limits back in play, set from a quarterly evaluation — so a score that dropped in spring restricts your autumn. See the IPI guide.

Cash. Under DD+7, reported order-to-bank is 14 to 27 days for FBA. A reorder point calculation that ignores whether the money exists on the day is a wish. See the DD+7 guide and the cash flow planner.

Q4. Every input degrades simultaneously: freight slower, receiving slower, storage roughly triple, peak fulfillment fees from October 15, and reported guidance to hold eight to ten weeks of cover rather than ninety days. See the peak season guide.

05What breaks a forecast

A launch. No sales history, so no daily sales figure. The freight guide’s split shipment — a small air batch while the ocean bulk travels — exists for exactly this.

A promotion. A deal compresses weeks of demand into days and leaves you thin afterwards, at the low-inventory threshold. Model the deal quantity separately from baseline demand — see the deals guide.

A subscription base. Subscribe & Save demand is predictable, which helps — but a stockout cancels subscriptions and those subscribers do not return. Subscribed units deserve their own safety stock.

A price change. Moving price changes conversion and therefore velocity, so the daily sales figure that fed the reorder point is stale the moment you reprice.

Seasonality read from an average. The most common forecasting failure and the most expensive, because the correction happens in Q4.

06The routine

Weekly, ten minutes: check days of cover per SKU against the 28-day floor. Anything approaching it is a reorder decision now, not next week.

Monthly, thirty minutes: recalculate daily sales on the last 30 and 90 days. Divergence between the two is a trend — up or down — and the 30-day figure is what the reorder point should use.

Quarterly: review actual lead times against assumed ones, per supplier and per freight mode. Most forecasts fail on a lead time assumption that was true two years ago.

Annually, in August: decide the Q4 position. Reporting is consistent that the decision window closes on September 30.

07What to do

Measure your real total lead time, including your own decision lag and Amazon receiving. Most sellers under-count by two to three weeks.

Target 30 to 45 days of cover and accept that this means ordering more often.

Place overlapping orders where lead time exceeds target cover. It is the only arrangement that fits inside the corridor.

Set safety stock from your worst fortnight, not your average.

Check capacity before ordering, not after the shipment is booked.

Check cash timing before committing. The purchase order and the payout do not arrive on the same schedule.

Treat launches, promotions and seasonal SKUs as separate calculations, not adjustments to the baseline.

Write your total lead time down as one number, including your own decision lag. If it is bigger than 45, you already know why the corridor keeps closing on you — and the fix is order frequency, not order size.

Frequently asked

How many days of inventory should I hold?

Roughly 30 to 45 days of cover. Below a 28-day threshold the low-inventory-level fee applies; above about 45 days the IPI score and the storage stack begin working against you.

How do I calculate a reorder point?

Daily sales multiplied by total lead time in days, plus safety stock. Total lead time includes your own decision lag, production, prep, freight, customs and Amazon receiving of 1 to 3 business days.

What if my lead time is longer than my target cover?

It usually is. The answer is overlapping orders — placing the next order before the previous one lands — rather than holding more inventory, because the fee structure penalizes the higher cover.

How much safety stock do I need?

Enough to cover variability rather than average demand. A reasonable starting method is taking your worst recent fortnight of daily sales instead of the average and covering the difference across the lead time.

Why did my forecast fail in Q4?

Because every input degrades at once: freight and receiving slow down, storage roughly triples, peak fulfillment fees start on October 15, and reported guidance is eight to ten weeks of cover rather than ninety days. The decision window closes on September 30.

What is the most common forecasting mistake?

Under-counting lead time — omitting the decision lag and Amazon receiving — and reading a seasonal product from an average. The first causes stockouts; the second is discovered in Q4 when correcting it is most expensive.

Sources

1. Low-inventory-level fee further explained (28-day threshold on both the 30-day and 90-day windows), Amazon Seller Forums, official announcement Primary source accessed 2026-09-04
2. Amazon IPI score 2026: how to improve it (45-day days-of-inventory target improving excess inventory percentage and sell-through simultaneously; stranded listings and dead SKUs), Nova Analytics Secondary accessed 2026-09-04
3. Amazon FBA restock limits 2026 (quarterly capacity evaluation restricting Q3 inbound before Black Friday; ASIN-level restock limits), Nova Analytics Secondary accessed 2026-09-04
4. Amazon FBA storage fees 2026: monthly, aged and Q4 peak rates (storage utilization surcharge above 22 weeks of supply; four-layer storage structure), ConversionPerk Secondary accessed 2026-09-04
5. Amazon FBA fees breakdown 2026 (low-inventory-level fee below 28 days of historical sales; eight-to-ten-week peak cover guidance), Nventory Secondary accessed 2026-09-04
6. Amazon FBA freight forwarder China: complete guide (transit times by mode; export and import customs; Amazon receiving of 1–3 business days; peak-season delays), FBA Freight Secondary accessed 2026-09-04
7. Shipping from China to Amazon FBA (2026 guide) (35–55 days door-to-FBA for LCL and 30–45 for FCL), WWS Cargo Secondary accessed 2026-09-04
8. Amazon’s new DD+7 payout policy (14–27 days from order to bank deposit for FBA), SlopePay Secondary accessed 2026-09-04
9. Amazon Subscribe & Save data deep-dive 2026 (stockouts causing subscription cancellations, with subscribers not returning), Velocity Sellers Secondary accessed 2026-09-04