What Is Safety Stock? Definition, Formula, and a Worked Example

Safety Stock, Defined

Safety stock is the extra units you hold above what you expect to sell, sized to cover the difference between your forecast and reality — a late shipment, a demand spike, a forecasting miss. It's not a guess and it's not padding; it's a number calculated from how much your demand and lead time actually vary.

It's easy to confuse with two neighbors. Cycle stock is the inventory you expect to sell through in a normal replenishment cycle — the stuff you planned for. Safety stock sits on top of that, covering the demand cycle stock doesn't. The reorder point is different again: it's the inventory level that triggers a new order, and it's built from cycle stock plus safety stock, not a substitute for either.

The Formula, and What Each Input Actually Means

Every safety stock formula, from the simple fixed-days method to Greasley's standard-deviation formula, pulls from the same handful of inputs. Get one wrong — usually lead time — and the rest of the math doesn't matter.

• Service level (Z-score): the stockout risk you're willing to accept. Higher service level, more safety stock.
• Average demand: units sold per day, averaged over a period long enough to be representative.
• Demand variability: the standard deviation of that daily demand — how much it actually swings, not how much you wish it didn't.
• Lead time: days from placing the order to units being sellable, not the supplier's quoted transit time.
• Lead time variability: how much that lead time actually moves shipment to shipment, taken from your own purchase order history.

The table below shows where an Amazon seller actually finds each number, because "check your forecast" isn't an answer.

A Worked Example With Real Numbers

Take a real-shaped example. A SKU sells an average of 40 units a day, with a daily standard deviation of 12 units — some days 25, some days 60. Lead time from PO to sellable FBA inventory runs 30 days. You want a 95% service level, which is a Z-score of 1.65.

Using the standard formula for variable demand with a fixed lead time — Safety Stock = Z × σD × √L — that's 1.65 × 12 × √30, or 1.65 × 12 × 5.48, which comes out to roughly 108 units.

Add that to average demand during the lead time (40 × 30 = 1,200 units) and your reorder point is 1,308 units: the level at which you place the next order, not the level at which you run out.

Change one input — push the service level from 95% to 99% (Z = 2.33) — and safety stock jumps to roughly 153 units. That's the point: safety stock isn't a fixed cushion, it's a function of how much risk you're accepting today.

The Mistake Everyone Makes (Including Us)

The most common mistake isn't a wrong formula. It's a formula that was right once. Lead time and demand variability both change — a supplier gets slower after a factory move, a SKU's demand smooths out post-launch — and the safety stock number that was correct in March is either starving the reorder point or quietly building aged inventory by September.

We've made this mistake ourselves. Across the more than $500M in managed revenue we've managed across 100+ brands, the pattern repeats: a service level gets set to protect a Q4 spike, nobody revisits it in Q2, and the brand pays carrying cost on safety stock sized for a demand curve that no longer exists.

The fix isn't a better formula. It's a cadence — recomputing demand and lead time variability on a schedule, not when someone finally notices a problem.

When the Number Is Wrong: Signals and Fixes

Two signals tell you the number is wrong, and they point in opposite directions.

• You're stocking out despite a calculated safety stock. Usually the lead time input is stale — you used the supplier's quoted lead time instead of your actual PO-to-sellable history, which includes customs delays, Amazon check-in queues, and the supplier's own missed dates.
• Safety stock is turning into aged inventory. The service level or demand average is set higher than current reality supports — often left over from a seasonal push or a launch forecast that never repeated.

Either way, the fix starts with the same question: when did you last recompute this using what actually happened, not what you planned? If the answer is "at launch" or "last year," the number is decorative, not functional.

Where Dr. Stock Fits

None of this requires a vendor. The formulas above work with a spreadsheet and honest inputs. Where it gets hard on Amazon specifically is keeping lead time and demand numbers current across hundreds of SKUs while watching for the stockouts that hit rank mid-campaign and the aged inventory building storage fees behind them — that's what Dr. Stock, built by Full Circle (which has managed more than $500M in revenue across 100+ brands), is built to catch, with a human still approving every purchasing decision. If the problem you're chasing is in the ad account rather than the warehouse, that's Dr. PPC's territory, not ours.