Inventory restock math for Amazon FBA in 2026

Inventory restock used to be a math problem: take your daily sales rate, multiply by your lead time, add a safety buffer, place the order. Three variables, one decision.

In 2026 it’s a multi-constraint optimization problem with at least seven inputs and several hard ceilings you can’t see until you hit them. Most sellers haven’t updated their mental model since 2021, which is why FBA stockouts are at record levels for sellers who were running fine 18 months ago.

This article walks through the math that actually works now — the inputs you need, the formulas that combine them, the failure modes to avoid, and a worked example you can adapt.

What changed since 2022

Three things, each significant on its own:

1. Capacity Limits replaced restock limits. Amazon now grants you a fixed cubic-foot allocation per fulfillment center that includes everything currently in the warehouse plus everything inbound. You can sell out of a SKU and still not be allowed to send replacement units in because your total allocation is full of other SKUs.

2. Inventory Performance Index (IPI) directly controls capacity. Your IPI score (which Amazon recalculates weekly) determines your capacity. Below 400, your allocation tightens. Below 350, you start getting overage fees. The relationship between IPI and capacity used to be advisory; now it’s mechanical.

3. Lead time variance widened. Sea freight from China is more reliable than 2021’s chaos but more variable than pre-pandemic. Customs holds vary by route. Amazon’s receiving time at fulfillment centers ranges from 3 to 21 days depending on facility. A “30-day lead time” is really a 25-to-45-day range.

The implication: you can’t optimize for sales velocity alone anymore. You’re optimizing for three intertwined constraints: never stocking out, never exceeding your cubic-foot allocation, and keeping your IPI high enough to preserve future allocation.

The seven inputs you need

For every SKU, before you can do restock math, you need:

Without these seven inputs you can’t compute correct restock quantities. With them, the math becomes tractable.

The corrected restock formula

The right formula has three components: reorder point, reorder quantity, and capacity check.

Reorder point is the inventory level that triggers a new order:

Reorder Point = (Daily Velocity × Lead Time Avg)
             + (Safety Stock for Velocity Variance)
             + (Safety Stock for Lead Time Variance)

Where the safety stocks are:

Velocity Safety = 1.65 × Daily Velocity StdDev × √Lead Time
Lead Time Safety = 1.65 × Lead Time StdDev × Daily Velocity

(The 1.65 factor gives you 95% service level — i.e., a 5% chance of stockout per cycle. Use 2.05 for 98% if your category has steep stockout penalties.)

Reorder quantity is how much to send when you reorder. This is where capacity becomes the binding constraint:

Target Days of Supply = 60 (typical)
Naive Reorder Qty = 60 × Daily Velocity = 600 units

But capped by:
Available Capacity (in your units) = (Cubic Feet Remaining / Cubic Feet Per Unit)

Actual Reorder = min(Naive Reorder, Available Capacity × 0.8)

The 0.8 multiplier is a buffer — never use your full allocation, because you need room for receiving variance and other SKU growth.

Capacity check is the new step most sellers skip:

The four failure modes

Failure mode 1: Using a 30-day flat average for velocity. As covered in the profit-tracking article, flat averages lag trends. For restock math the consequence is worse than for profit: you under-order trending-up SKUs (stockout) and over-order trending-down SKUs (LTSF + capacity waste).

Failure mode 2: Ignoring inbound in current inventory. Your “current FC inventory” should include inbound shipments that will arrive before your next reorder cycle. Many sellers double-order because they look at FC stock but forget the 800 units already on a truck.

Failure mode 3: Single lead time average. Suppliers, freight forwarders, and Amazon receiving each have their own variance. Treating “30 days total” as a single estimate hides the fact that an unlucky combination (delayed production + customs hold + slow FC receiving) can easily blow out to 60 days. Use the standard deviation of total observed lead times, not just the average.

Failure mode 4: Not adjusting for seasonality. If you’re running a steady-state restock formula on a SKU that does 30% of its annual volume in November and December, you’ll be perpetually short in Q4 and over-stocked in January.

The IPI feedback loop

Here’s the part most sellers miss: your restock decisions affect your future restock capacity through IPI.

IPI is calculated from four signals:

1. Excess inventory percentage (older than 90 days × not selling)
2. Sell-through rate (units sold ÷ average inventory)
3. Stranded inventory percentage (inventory not available to buy)
4. In-stock rate (percentage of time your SKUs are buyable)

The feedback: if you over-order a slow mover, your excess inventory percentage rises, your IPI drops, your capacity allocation tightens, and now your fast-mover SKUs get squeezed too.

The implication for restock math: the right reorder quantity isn’t the one that maximizes this SKU’s sell-through. It’s the one that maximizes total catalog throughput within your capacity allocation.

The 2026 restock cadence we recommend

For most sellers, the right cycle has four pieces:

Weekly (15 minutes)

• Check the Capacity Monitor for current allocation status
• Identify any SKU below its reorder point
• Confirm inbound shipments are progressing on schedule

Biweekly (45 minutes)

• Recalculate per-SKU velocity using last 4 weeks weighted average
• Update lead time averages if recent shipments deviated significantly
• Place supplier POs for any SKU at or near reorder point with capacity available

Monthly (90 minutes)

• Review IPI score and identify which SKUs are dragging it down
• Disposition slow movers (price drops, removal orders, multi-pack bundles)
• Recompute safety stocks if variance has changed

Quarterly (3 hours)

• Full SKU-level profitability review (which SKUs deserve their allocation?)
• Seasonal adjustments for upcoming quarter
• Discontinue or sunset SKUs whose IPI contribution is negative

This isn’t onerous. The total time investment is roughly 10 hours per quarter, against which a stockout-free Q4 is worth orders of magnitude more.

What to do this week

1. Pull your current Capacity Monitor and calculate your remaining allocation in cubic feet
2. Compute true safety stocks for your top 20 SKUs using the formulas above (a spreadsheet is fine)
3. Cross-check inbound against your “current inventory” calculations — if you’ve been double-counting, your reorder points are too high
4. Set IPI alerts for when your score drops below 400

The math itself is mechanical once you have the inputs. The hard part is convincing yourself that velocity, capacity, and IPI all need to be optimized together — not one at a time.

(SellerPulse runs all four checks weekly and surfaces SKUs that are at risk of stockout, at risk of LTSF, or dragging IPI down. Flat $99/mo on the Pro plan; one prevented Q4 stockout typically covers the year.)

The actionable summary

The biggest mindset shift: capacity is now the binding constraint, not velocity. Optimizing for the wrong constraint is why so many sellers stocked out in Q4 2025 while sitting on warehouse-fulls of unmovable SKUs.