Safety stock formula: four ways to calculate it, and which to use

Safety stock is the stock you hold so that a bad week doesn't become an empty shelf. There are several ways to calculate it, and they give very different answers. This guide runs the four common formulas on the same product, so you can see what each one does, what it costs, and when to use it.

What safety stock protects you from

Two things go wrong between placing an order and receiving it:

• You sell more than average. Demand varies from day to day and month to month.
• The supplier is slower than average. Lead times vary too — usually more than suppliers admit.

Each formula below covers one or both of these. The more completely a formula describes them, the less stock it needs to give the same protection.

The example product

SKU-0012, a 10 mm fuel filter, from the sample data in our free reorder point calculator:

Method 1 — Max minus average

The rule of thumb most people start with, because it needs no statistics. Its weakness is that it plans for your best sales month and your slowest delivery happening at the same time — which is rare. One unusual month or one stuck container sets the buffer for the whole year. It reliably overstocks.

Use it when you have very little history, or you need a number today and will refine it later.

Method 2 — Demand variability only

The textbook formula. It assumes the lead time never changes, and covers only the chance of selling more than average while you wait. The √LT is there because good and bad days partly cancel out over a long lead time.

Use it when the supplier is genuinely reliable — local, or delivering to the day, every time.

Method 3 — Lead time variability only

The mirror image: demand is treated as steady, and the buffer covers the supplier arriving late.

Use it when demand is very stable but the supplier isn't — a consumable that sells the same every week, shipped from overseas.

Method 4 — Demand and lead time combined

This covers both risks at once. Notice it is not 45 + 46 = 91. Selling more than usual and a late supplier are mostly independent, so they rarely hit at full strength together; the square root adds them the way independent risks actually combine. Adding methods 2 and 3 instead is a common mistake that overstocks by about 40%.

Use it as your default for anything bought from a supplier with a real lead time. It is the formula our free calculator uses.

The four methods side by side

Choosing the service level (Z)

The service level here is the probability of not running out during one replenishment cycle. Z is the matching value from the normal distribution: =NORM.S.INV(0.95) returns 1.645.

Each step up costs more than the last. Going from 95% to 99% adds 26 units; going from 99% to 99.9% adds another 29 to remove a risk that was already one in a hundred. Spend the high service levels where a stockout is genuinely expensive, not across the whole catalogue.

Monthly data, daily formula

Most sales reports are monthly; lead times are in days. Convert demand to days:

The √30.4 assumes one day's sales don't depend on the day before. If you sell in occasional bulk orders, that isn't true and daily σ comes out too small — in that case, work with weekly data instead. You can also do the whole calculation in months (lead time 21 ÷ 30.4 = 0.69 months); as long as every input is in the same unit, the answer is the same.

Safety stock formulas in Excel

With average daily demand in R2, daily SD in S2, lead time in B2, its SD in C2, the slowest delivery in X2, monthly sales in F2:Q2 and the service level in Z1:

Method 1  max − average    =ROUNDUP(MAX(F2:Q2)/30.4*X2-R2*B2,0)
Method 2  demand only      =ROUNDUP(NORM.S.INV($Z$1)*S2*SQRT(B2),0)
Method 3  lead time only   =ROUNDUP(NORM.S.INV($Z$1)*R2*C2,0)
Method 4  combined         =ROUNDUP(NORM.S.INV($Z$1)*SQRT(B2*S2^2+R2^2*C2^2),0)

To get σLT, list your last ten or more deliveries from that supplier with the order date and the date received, and take =STDEV.S() of the days between them. If you only have the supplier's quoted lead time, you don't know σLT yet — start with method 2 plus a few days of extra lead time, and start recording deliveries.

Which formula should you use?

Common safety stock mistakes

1. Trusting the quoted lead time. Measured lead times are almost always longer and more variable. The variation (σLT) is often the biggest term in the formula.
2. Treating safety stock as spare stock. If you sell it down every cycle, it isn't a buffer; your reorder point is too low.
3. Counting stockout months as low demand. A month you were empty for two weeks shows half the real demand and shrinks both d and σd.
4. Never recalculating. A buffer set last year is protecting last year's demand.
5. One service level for every product. The cheapest way to cut stock without more stockouts is to lower the service level on items where running out costs little.

Frequently asked questions