How to Determine Safety Stock: The Formula, a Worked Example, and What to Do When It's Wrong
Safety stock equals a service-level factor (Z-score) times the combined variability of demand and lead time. For low-data SKUs, use the simpler average-max method instead. Either way, you need 6-12 months of sales and lead-time history before the number means anything.
The team behind Dr. Stock
The Formula That Actually Works: Demand and Lead Time Together
Most stockouts aren't caused by bad luck. They're caused by using an average when you should have used a range. Demand moves, lead time moves, and if you only plan for the average of both, you're planning to be short about half the time.
The statistical formula accounts for both kinds of variability at once: SS = Z × √[(Lead Time × Demand Variance) + (Average Demand² × Lead Time Variance)]. Z is your service-level factor — 1.65 for 95% service, 1.96 for 97.5%, 2.33 for 99%. Higher Z means fewer stockouts and more cash sitting in inventory.
Worked example: a SKU sells an average of 40 units a day with a standard deviation of 12 units. Average lead time is 30 days with a standard deviation of 4 days. Targeting 95% service (Z = 1.65):
- Demand-side term: 30 × 12² = 4,320
- Lead-time-side term: 40² × 4² = 25,600
- Sum and square root: √29,920 ≈ 173
- Safety stock: 1.65 × 173 ≈ 285 units
- Reorder point: 285 + (40 × 30) = 1,485 units
That's the number that goes into your reorder point, not a number you hold forever. Recalculate it when demand, lead time, or your target service level changes — which for most Amazon sellers is more often than they think.
The Fast Method for Low-Data SKUs: Average-Max
The statistical formula needs enough order history to calculate a real standard deviation. New SKUs, seasonal one-offs, and anything under six months old don't have that yet. For those, use the average-max method: SS = (Max Lead Time × Max Daily Sales) − (Average Lead Time × Average Daily Sales).
It's cruder — it protects against the single worst combination you've seen rather than a statistical distribution — which means it tends to run higher than the statistical answer and costs more in holding fees. It's still better than guessing, and it's the right tool for C-tier SKUs where a fully worked formula isn't worth the analyst time.
Exclude genuine one-off outliers before you calculate — a single freak six-week shipping delay from a customs hold shouldn't set your safety stock for the next year. But don't exclude a slow lead time just because it's inconvenient; if it happened once without a special cause, it can happen again.
Amazon-Specific Inputs the Generic Formula Skips
The formula is generic supply chain math. Amazon adds inputs that most safety stock explainers never mention, because they were written for warehouses that ship their own trucks.
Your lead time isn't just supplier production time — it's production plus transit plus the time a shipment sits in an Amazon inbound queue before it's sellable. That queue time varies by fulfillment center, by season, and by how clean your shipping plan is, and it belongs in your lead-time variance, not your demand variance. We've watched this pattern repeat across managed accounts: Full Circle has managed more than $500M in revenue across 100+ brands, and the lead-time swings that blow up a safety stock calculation almost always trace back to receiving time, not the supplier's factory floor.
Other Amazon-only inputs worth building into your inputs, not your gut: restock limits that cap how much you can even send in, IPI score effects on those limits, and the removal-versus-liquidation math if your safety stock ends up higher than your storage allowance can absorb cheaply.
When the Number Comes Back Wrong
If the formula gives you a number that's clearly too big, check for a demand-lead-time correlation problem before you cut it. During peak season, demand and lead time often spike together — more orders and slower receiving at the same time — and the independent version of the formula (treating them as unrelated) understates the real risk. There's a dependent version of the formula for exactly this case, and if you're only running the independent one going into Q4, you're under-covered when it matters most. That's a mistake we've made on client accounts before catching it in a peak-season review — the fix was switching to the dependent formula for Q3-Q4 planning, not lowering the target service level.
If the number is too small and stockouts keep happening anyway, check whether the volatility is coming from your side or the demand side. A safety stock calculation can't fix demand that's spiking because of an ad campaign you didn't plan inventory around — that's a forecasting problem in the ad account, not the warehouse, and it belongs with whoever runs your PPC, not your inventory formula. If that's the actual leak, Dr. PPC is the right place to look, not a bigger buffer.
The Mistake Most Sellers Make
The most common mistake isn't a formula error. It's using safety stock as a permanent patch for a problem that has a root cause: bad forecast accuracy, a supplier who won't commit to a lead time, or an internal process that doesn't communicate demand changes to purchasing fast enough. Stacking safety stock on top of those problems hides them and pays storage fees for the privilege.
Before you accept a high safety stock number as the answer, ask whether the variance driving it is fixable. A supplier whose lead time swings 10 days because they batch small orders with bigger customers is a negotiation problem, not a safety stock problem. Solve that and the whole calculation shrinks.
| Input | What It Captures | Where to Pull It From (Amazon Sellers) |
|---|---|---|
| Average daily demand | Typical units sold per day | Business Reports or Brand Analytics, 6-12 months |
| Demand variability (std dev) | How much daily sales actually swing | Daily unit sales history, same window |
| Average lead time | Days from PO placed to unit sellable | PO dates plus FBA receiving time, not just ship date |
| Lead time variability (std dev) | How much that lead time swings order to order | PO history including customs and inbound queue delays |
| Target service level (Z) | How often you'll accept a stockout risk | Business decision — commonly 95%, 97.5%, or 99% |
Which one you should actually pick
The statistical formula suits sellers with 6-12 months of clean demand and lead-time data and the patience to recalculate it. The average-max method suits new or low-volume SKUs where that history doesn't exist yet. Neither fixes a demand spike that's actually coming from an ad campaign — that's a PPC forecasting problem wearing an inventory costume.
Shortlist on the job, not the feature grid. Total three numbers first: storage and aged-inventory surcharges for the last twelve months, lost sales on days your best sellers were out of stock, and cash sitting in SKUs that have not moved in 180 days. Then ask each vendor what they would do about those three in week one.
Common questions
What service level should Amazon sellers actually use?
Most sellers land between 95% and 99% depending on the SKU. A core, review-generating ASIN that would lose rank in a stockout justifies 97.5-99%. A slow-moving accessory doesn't need the same protection and doesn't need the storage cost that comes with it. Set service level by SKU importance, not one number for the whole catalog.
How often should I recalculate safety stock?
Monthly for anything volatile or newly launched, quarterly for stable SKUs, and always before Q4 regardless of category. Lead time and demand both drift, and a safety stock number calculated in March on last year's data is not protecting you in November.
Does Amazon's own Restock Recommendations calculate this for me?
It runs its own model and gives you a number, but it doesn't show you the demand variance, lead time variance, or service level it assumed to get there — so you can't audit it or adjust it for a supplier issue you know about and Amazon doesn't. Treat it as a sanity check against your own calculation, not a replacement for it.
What if my safety stock number is bigger than I can afford to store?
That's a signal, not a reason to quietly ignore the number. It usually means either your lead time variance needs fixing at the supplier level, or the SKU doesn't justify the service level you set. Cutting the number without fixing the cause just moves the stockout risk back in, and holding it as-is means running the removal-versus-liquidation math on whatever else is tying up space.
Is safety stock the same as buffer stock or reorder point?
Safety stock and buffer stock are the same thing, usually used interchangeably. Reorder point is different — it's safety stock plus expected demand during lead time, and it's the trigger level at which you place a new order, not the cushion itself.
Dr. Stock runs Amazon inventory and supply chain — reorder timing, stockout risk, storage and aged-inventory fees, FBA fee errors and dimensional-weight misclassification, shipment discrepancies and reimbursement recovery — with operators from a $500M+ Amazon team supervising. Purchasing decisions always come to a human. Orbit is included. First 30 days free, priced on the call.
Book a Dr. Stock demoRead next
- Extensiv Pricing 2026: No Rate Card, and What Moves ItPricing · extensiv pricing
- Best Amazon Reimbursement Service: How to Choose in 2026Buyer's guide · best amazon reimbursement service
- Sellerboard Pricing 2026: Plans, Order Bands, LimitsPricing · sellerboard pricing
- Cin7 Pricing 2026: The Order Bands Are the Real PricePricing · cin7 pricing
- Helium 10 Pricing 2026: Every Tier and the Ads FeePricing · helium 10 pricing
- ShipBob Pricing 2026: The Five Lines in Every QuotePricing · shipbob pricing
Part of
- Orbit — the software, included freeInventory, finance, ASIN profitability and the fee, price, BSR and buy box trackers
- Dr. PPCWhen the leak is in the ad account rather than the warehouse