The safety stock formula sets the buffer inventory you hold to cover uncertainty in demand and supplier lead time. With a target service level, you multiply a statistical factor by the variation in demand over the lead time. The result is the extra units that keep stockouts at the chance you accept.
What is the safety stock formula?
When both demand and lead time vary, safety stock equals the service level factor Z times the square root of the combined variance. If lead time is fixed, the formula shortens.
Safety stock = Z x square root of ( L x sigma_d squared + d squared x sigma_L squared )
Fixed lead time: Safety stock = Z x sigma_d x square root of L
- Z: service level factor from the normal distribution. 90% = 1.28, 95% = 1.645, 97.5% = 1.96, 99% = 2.33, 99.9% = 3.09.
- L: average lead time in days.
- d: average daily demand, and sigma_d its standard deviation.
- sigma_L: standard deviation of lead time in days.
The reorder point is then average daily demand x lead time + safety stock. You can run the numbers in the safety stock calculator, which uses the same definitions.
What does service level mean here?
Service level is the probability of not running out of stock during a replenishment cycle. A 95% target means roughly one cycle in twenty ends in a stockout, not that 95% of units are filled.
That distinction trips people up. This cycle service level differs from fill rate, which measures the share of demand met from stock. Two items with the same cycle service level can have very different fill rates, depending on order size. Choose the target by item class: higher for items where a stockout is costly or customers will not wait, lower for slow, cheap or easily substituted items.
Worked example with fixed lead time
This is an illustration. An item sells 40 units a day on average, with a standard deviation of 10 units. Lead time is 16 days and does not vary. The target service level is 95%, so Z = 1.645.
- Square root of lead time = square root of 16 = 4.
- Safety stock = 1.645 x 10 x 4 = 65.8, about 66 units.
- Demand during lead time = 40 x 16 = 640 units.
- Reorder point = 640 + 65.8 = about 706 units.
The reorder point is mostly expected demand. The buffer is a small slice of it, which is typical when demand is fairly steady.
What changes when lead time also varies?
Supplier variability often dominates. Keep the same item but let lead time vary by a standard deviation of 2 days around its 16-day average.
- Demand term: L x sigma_d squared = 16 x 100 = 1,600.
- Lead time term: d squared x sigma_L squared = 1,600 x 4 = 6,400.
- Square root of (1,600 + 6,400) = square root of 8,000 = about 89.4.
- Safety stock = 1.645 x 89.4 = about 147 units.
Safety stock more than doubles, from 66 to 147 units, even though demand did not change. The lesson: working with suppliers to make delivery more reliable often saves more inventory than refining the demand forecast.
How does the service level choice affect the buffer?
Each step up in service level costs more than the last, because Z grows faster as you approach certainty.
| Target service level | Z factor | Buffer relative to 95% |
|---|---|---|
| 90% | 1.28 | Lower, about three quarters |
| 95% | 1.645 | Baseline |
| 99% | 2.33 | Higher, roughly 1.4 times |
| 99.9% | 3.09 | Nearly double |
Moving from 95% to 99.9% raises safety stock by almost 90% for the same item. Reserve the highest targets for items where the cost of a stockout justifies that holding cost.
What data should you gather first?
The formula is only as good as its inputs. Collect these before calculating.
- Demand history at the same time bucket as lead time, usually daily or weekly, with promotions and one-off orders flagged.
- Actual lead times from order date to usable stock, including supplier processing, transit, customs and receiving, not the quoted time.
- A consistent standard deviation: measure it on the same unit and period as the average.
- The target service level agreed with sales or customer service, by item class.
What are the limits of the method?
The formula assumes demand is roughly normally distributed and that the past predicts the future. It works best for items with steady, frequent demand.
For intermittent items, strong seasonality, new products, or after a sudden supply change, it can give a buffer that is too low or too high. Check results against actual stockouts and adjust. If you use forecast error rather than raw demand variation, the buffer reflects how well you predict, which is usually the better measure when a forecast drives replenishment. Analytics tools can estimate these distributions and flag items whose variability has changed, but they need clean history and a person to challenge odd results. For related methods, see the inventory and warehousing guides and the reorder point calculator.
Common questions
How do you calculate safety stock?
Multiply the service level factor Z by the standard deviation of demand over the lead time. With fixed lead time, that is Z x sigma_d x square root of L. If lead time varies too, use the combined formula with both variances.
What is the difference between safety stock and reorder point?
Safety stock is the buffer for uncertainty. The reorder point is the trigger level for placing an order, and it equals expected demand during lead time plus safety stock.
Which service level should I choose?
Choose it by item importance and cost of a stockout. Many teams use 95% as a starting point and set higher values for critical items and lower ones for slow, cheap or substitutable items. Revisit the choice when customer expectations change.
How often should safety stock be recalculated?
Recalculate when demand patterns, lead times or service targets change, and on a regular schedule such as quarterly. Seasonal items may need different values by season. Stale parameters are a frequent cause of both stockouts and excess stock.