Amazon Stock Management System: The Math, the Mistakes, and What to Check When It's Wrong

What it actually is (and what it isn't)

An Amazon stock management system is the layer that sits on top of your FBA or FBM inventory data and tells you three things: what you have, where it is, and when to order more. It is not FBA itself — FBA stores and ships; it doesn't decide when you should reorder or flag that a SKU is about to age into a storage surcharge. Amazon's own Seller Central dashboards (Restock recommendations, the IPI score, the Manage Inventory Health report) do some of this, but loosely and on Amazon's schedule, not yours.

The core of any real system is a reorder point calculation: average daily sales × lead time, plus a safety stock buffer. Everything else — barcode scanning, warehouse automation, dropshipping, ABC categorization — is either a fulfillment technique or a way to feed better numbers into that one calculation. If you remember nothing else, remember this: the system is only as good as the three inputs going into it.

The math, worked with real numbers

Say a SKU sells 38 units a day on average. Lead time — production, ocean or air transit, and FBA check-in — runs 52 days. You want 12 days of safety stock to cover demand spikes and a supplier that occasionally ships late.

Reorder point = (38 × 52) + (38 × 12) = 1,976 + 456 = 2,432 units. That's the inventory level at which you place your next order, not the level at which you run out. If you wait until you're at 500 units to reorder, you're already about 30 days behind, and at 38 units a day that's roughly 1,150 units of demand you can't cover — a stockout, not a close call.

Across the $500M+ in managed revenue we've managed across 100+ brands, the single biggest driver of an avoidable stockout is a reorder point that skipped the safety stock term entirely, or set it as a flat percentage instead of tying it to how much that specific SKU's lead time actually varies.

Where this breaks — including a mistake we've made

The formula is simple. The inputs are where sellers, and frankly agencies, get it wrong.

• Trailing sales during a stockout look like demand loss. We've flagged a reorder recommendation that was quietly wrong for exactly this reason: 30-day velocity looked flat on a growing SKU, but five of those days were an actual stockout suppressing the sales data. Exclude stockout days from the average or you'll under-order again next cycle.
• Lead time measured from PO date, not sellable date. If you clock lead time from when the PO is placed to when the shipment leaves the factory, you're missing ocean or air transit and the FBA check-in queue — often 5 to 15 days you didn't budget for.
• Safety stock as a flat 10% or 15%. A SKU with erratic demand and a supplier with a history of late shipments needs more buffer than a stable seller with a reliable factory. One number for every SKU is a shortcut, not a calculation.
• Chasing the IPI score by liquidating a profitable slow mover. The score rewards inventory efficiency, not profitability. Liquidating a SKU that turns slowly but earns well just to improve a number is trading real margin for a metric.

When the answer is bad news

The reorder point tells you a number. What you do when that number turns out to be wrong, or the system was already supposed to be handling it, matters more than the formula.

• The reorder point looks wrong. Check whether the trailing sales window includes stockout days, whether lead time is measured to "shipped" or to "sellable," and whether safety stock reflects that SKU's actual variability or a default setting nobody revisited.
• Automatic reordering was already on and you still stocked out. Check whether the trigger fired on time but the supplier confirmation lagged, or whether a spend or unit guardrail capped the order below what the reorder point actually called for.
• You raised the reorder point and it still isn't working. Confirm the stockout is a supply problem and not a demand-side one — a suppressed listing or a lost buy box will drop sales velocity too, and no amount of extra stock fixes that.

The four inputs, and where each one usually goes wrong

If you're building this yourself before buying anything, these are the four numbers that matter. Get these right and the tool underneath — spreadsheet, software, or a managed system — is almost incidental.