Safety stock is the buffer you hold against two uncertainties: customers wanting more than expected, and suppliers delivering later than expected. Too little buffer means stockouts. Too much means cash on shelves. Two formulas cover most cases: a simple one and a statistical one.
The max-and-average method
Safety stock = (maximum daily usage x maximum lead time) – (average daily usage x average lead time). It covers the worst realistic case: peak demand during the longest delivery wait, minus what you would normally need.
Worked example: maximum daily usage 30 units, maximum lead time 10 days, average daily usage 20 units, average lead time 7 days. Safety stock = (30 x 10) – (20 x 7) = 300 – 140 = 160 units. Hold 160 units above your normal cycle stock.
This method is generous by design, because it stacks two maximums. Use it for critical items where a stockout stops production or loses a key customer. For ordinary items it often over-stocks, which is why the Z method exists.
The Z method
Safety stock = Z x standard deviation of daily demand x square root of lead time. Z is the service factor: 1.65 for a 95% service level (in stock 95% of cycles), 2.33 for 99%. The standard deviation measures how much daily demand varies.
Worked example: Z = 1.65 (95% service), demand standard deviation 5 units, lead time 7 days. Safety stock = 1.65 x 5 x square root of 7 = 1.65 x 5 x 2.646 = 21.8, about 22 units. Far lower than the max-average figure, because statistics says the worst case rarely happens on both dimensions at once.
The Z method needs decent demand history to estimate the standard deviation. With little or erratic history, the max-average method is more honest, since it does not pretend to a precision the data cannot support.
Choosing a service level
Higher service levels cost disproportionately more stock: moving from 95% to 99% raises the Z factor from 1.65 to 2.33, a 41% increase in safety stock for 4 extra points of availability. Reserve 99% for items where a stockout is very expensive. For most items 90 to 95% is the sensible band, and the savings fund the buffer for the critical few.
Service level is per replenishment cycle, not per year. A 95% service level still means roughly one stockout in twenty cycles, which for weekly ordering is two or three stockouts a year.
Keeping it current
Recompute safety stock when demand patterns, lead times, or service targets change. A supplier that slips from 7 to 10 days needs a bigger buffer even if demand is unchanged. Automate the recomputation if you can: safety stock set once and forgotten is usually wrong within a year.
Safety stock across multiple locations
Holding safety stock in several warehouses costs more than holding it centrally, because each location needs its own buffer against its own variability. The square-root law quantifies this: total safety stock for n identical locations is roughly the single-location safety stock multiplied by the square root of n.
Example: one location needs 100 units of safety stock. Four locations need 100 x square root of 4 = 200 units in total, not 400. Centralising stock in one warehouse therefore halves the buffer for the same service level, at the cost of longer delivery times to distant customers. That trade-off, stock cost versus delivery speed, is the central decision in warehouse network design.
Try it faster with the tool
You can follow every step above by hand, but the free Safety Stock Calculator — the Simple Max-and-Average Formula, Plus the Z Method does the arithmetic for you. Enter your numbers and it gives the answer right away, so you can check your hand calculation or skip straight to the result.
Quick tips
- Max-average = (max use x max lead) – (avg use x avg lead).
- Z method = Z x demand deviation x square root of lead time.
- 95% service uses Z = 1.65; 99% uses Z = 2.33.