Safety stock formula: three methods compared

Safety stock is the extra inventory you hold to protect against two kinds of uncertainty: customers buying more than usual, and suppliers delivering later than usual. There is no single safety stock formula. This guide compares three methods, from the simplest to the most rigorous, with a hypothetical example for each, so you can choose the one that suits the data you actually have.

Method 1: days of cover

This is the simplest method and the one many small businesses start with. You decide how many days of extra stock you want, often based on experience with a supplier, and multiply by average daily demand.

Example (hypothetical numbers). An item sells 40 units a day on average. The buyer wants five days of buffer: 40 × 5 = 200 units.

The strength of this method is that it needs only average demand. Its weakness is that it ignores how variable demand and lead time really are. A stable item and an erratic item with the same average get the same buffer, so the first is overstocked and the second is still at risk.

Method 2: maximum minus average

This method uses the worst recent combination of demand and lead time. It needs a little more data: your highest daily demand and your longest lead time over a recent period.

Example (hypothetical numbers). Average demand is 40 units a day and average lead time is 12 days. Over the last six months, the highest daily demand was 55 units and the longest lead time was 16 days.

• Worst case: 55 × 16 = 880 units
• Average case: 40 × 12 = 480 units
• Safety stock: 880 − 480 = 400 units

The method is easy to explain, but it tends to produce large buffers because it assumes the highest demand and the longest delay happen at the same time. It is also sensitive to a single unusual day in the history.

Method 3: the statistical formula

The statistical method sets safety stock from the variability of demand and lead time and from the service level you want. The service level here means the probability of not running out during a replenishment cycle.

Here d is average daily demand, σd is the standard deviation of daily demand, L is average lead time in days, σL is the standard deviation of lead time and z is a factor taken from the standard normal distribution.

Service level	z
90%	1.28
95%	1.645
99%	2.33

Example (hypothetical numbers). Average demand is 40 units a day with a standard deviation of 10. Average lead time is 12 days with a standard deviation of 2 days. The target service level is 95%, so z ≈ 1.65.

• Demand-variability part: 12 × 10² = 1,200
• Lead-time-variability part: 40² × 2² = 6,400
• Square root of the total: √7,600 ≈ 87.2
• Safety stock: 1.65 × 87.2 ≈ 144 units

If the lead time never varied, only the first part would remain: 1.65 × 10 × √12 ≈ 57 units. Comparing the two answers shows that lead-time variability, not demand, drives most of the buffer in this example. That is a useful finding in itself: working with the supplier on delivery reliability could cut stock more than better forecasting would.

Comparing the three methods

Method	Data needed	Main strength	Main weakness
Days of cover	Average demand	Quick, easy to explain	Ignores variability
Maximum minus average	Average and peak demand and lead time	Simple worst-case logic	Often oversized; sensitive to outliers
Statistical	Averages and standard deviations	Links stock to a service level	Assumes roughly normal, stable demand

For the same hypothetical item, the three methods gave 200, 400 and 144 units. None of these is the right answer in general; each reflects different assumptions. The statistical method is usually the best starting point when you have at least several months of clean demand and delivery data.

Choosing a service level

A higher service level always costs more stock, and the cost grows faster as you approach 100%. Moving from 95% to 99% raises z from about 1.65 to 2.33, which increases safety stock by roughly 40% in the statistical method. Many businesses therefore set different service levels by item: higher for items whose shortage stops production or loses a key customer, lower for items with substitutes or long shelf life.

Common mistakes

• Mixing time units. If demand is measured per week, lead time and its standard deviation must also be in weeks.
• Using quoted rather than actual lead times. Measure lead time from your own order and receipt dates.
• Including one-off orders. A single large project order inflates both the average and the standard deviation. Remove it or treat it separately.
• Never recalculating. Revisit safety stock after a change in supplier, route, sales channel or season.
• Applying the formula to intermittent items. Items that sell rarely do not follow a normal distribution, so set their buffers by judgement or with specialist methods.

From safety stock to the reorder point

Safety stock is one part of the reorder point: the stock level at which you place the next order. Add expected demand during lead time to the safety stock, and you have the trigger. Our reorder point formula guide covers that step, and you can test your own numbers in the safety stock calculator or the reorder point calculator.

Additional sources

• Safety stock – Wikipedia
• Normal distribution table – NIST/SEMATECH e-Handbook of Statistical Methods
• Reorder point – Wikipedia

Related guides: ABC/XYZ classification, single-source supplier risk, inventory optimization consulting, and all free supply chain calculators.