How to calculate safety stock and the reorder point
Who it is forGrocery chain buying, inventory planning, DC-to-store replenishment teams
Safety stock is the balance between two errors: hold too little and you lose sales, hold too much and you tie up capital. In most businesses that balance is struck by instinct rather than calculation — a rule like "keep two weeks of cover" applied identically to every product. But a product with steady demand and one with volatile demand do not need the same buffer. This guide covers the formula, the service level choice and how the reorder point is set.
What safety stock covers
Safety stock is not held to cover average demand; it is held to cover deviation above the average. Average demand is already covered by the normal order quantity.
There are two sources of uncertainty. The first is demand uncertainty: a product averaging 50 units a week sells 30 in some weeks and 80 in others. The second is supply uncertainty: a promised seven-day delivery sometimes takes twelve.
A stock-out happens when both go wrong at once — a delayed delivery during a high-demand week. The size of the safety stock is determined by how much of that combined risk you accept, and that acceptance level is called the service level.
The safety stock formulas
There are three common formulas requiring different amounts of data. Use the most advanced one your data supports; they are all versions of the same logic at different levels of detail.
The simple formula uses only maximum and average values. It is practical when data is thin but systematically produces excess stock, because it assumes the worst case always happens.
The statistical formula uses standard deviation and lets you choose the service level explicitly. If lead time also varies, the extended formula is required; in imports and long supply chains that variability matters more than demand variability.
- Simple formula
(Max daily demand × Max lead time) − (Average demand × Average lead time)- Statistical formula
Z × σ(daily demand) × √(Lead time)- With variable lead time
Z × √( LT × σ²(demand) + Demand² × σ²(LT) )- Reorder point
(Average daily demand × Lead time) + Safety stock
Exclude days with no stock when calculating standard deviation. A day that sold zero because the shelf was empty does not mean demand was zero, and it artificially lowers the deviation.
Choosing the service level: 95 or 99 per cent
Service level is the probability of not running out during the lead time. A 95 per cent service level means accepting a stock-out in one of every twenty replenishment cycles.
The choice is a cost comparison and it is not symmetric: as service level rises, the safety stock required rises at an accelerating rate. Going from 90 to 95 per cent raises stock by 29 per cent; going from 95 to 99 adds another 41 per cent.
Which is why a single service level is not applied to every product. High service levels go to A-class items, low ones to C-class, and which product sits in which class comes from the ABC analysis.
| Service level | Z factor | Example safety stock | Increase over 90% | Suitable class |
|---|---|---|---|---|
| 90% | 1.28 | 54 units | — | C class, low value |
| 95% | 1.65 | 69 units | +29% | B class, standard |
| 97% | 1.88 | 79 units | +47% | A class |
| 99% | 2.33 | 98 units | +82% | Critical, contracted supply |
| 99.9% | 3.09 | 130 units | +141% | Rarely justifiable |
The safety stock required rises at an accelerating rate as service level goes up. Moving from 90 to 95 per cent adds 15 units; moving from 99 to 99.9 adds 32 — the last half point costs more than the first five.
Reorder point and order quantity
Safety stock alone does not tell you when to order; the reorder point does. It is the goods that will be consumed during the lead time plus the safety stock.
When stock falls to that level, an order is placed. The order quantity is a separate decision, set by economic order quantity or a fixed-period method; in practice the supplier's minimum order quantity and case size usually dominate the calculation.
In a fixed-period ordering system — where orders go out weekly — safety stock has to cover the order interval as well as the lead time, otherwise the gap between two orders is left unprotected.
- Reorder point
(Daily demand × Lead time) + Safety stock- In a fixed-period system
Z × σ(demand) × √(Lead time + Order interval)- Target stock level
(Daily demand × (Lead time + Order interval)) + Safety stock
The reorder point is checked against available stock, which includes goods already on order. Systems that look only at stock on hand raise a second order at the same reorder point.
Implementation errors that break safety stock
Even with the formula set up correctly, three implementation errors distort the result and all three are common.
- Calculating safety stock from annual rather than monthly data — for seasonal lines such as ice cream, bottled water and holiday products the deviation differs greatly within the period
- Recording zero demand on out-of-stock days — deviation reads low and safety stock is set too thin
- Mixing promotional periods into base demand — one campaign week permanently inflates the deviation
- Calculating safety stock once and never updating it — when the demand pattern changes the buffer is either short or excessive
- Clean the demand dataFlag out-of-stock days and campaign weeks; calculate the mean and deviation with those flags excluded.
- Measure lead time from actualsUse the achieved duration of the last twenty deliveries, not the contracted figure. Calculate its mean and its deviation together.
- Assign service level by ABC classA single service level is not applied across all products; for example, 97-99 per cent can be targeted for A class and 90 per cent for C.
- Recalculate quarterlyDemand patterns and supplier performance change. Safety stock calculated once a year is wrong for most of the year.
Frequently asked questions
- How is safety stock calculated?
- In the statistical formula, the service level factor is multiplied by the standard deviation of daily demand and the square root of lead time. For a product with a daily deviation of 14 units and a 9-day lead time, safety stock at 95 per cent service is 1.65 × 14 × 3 = 69 units.
- What is the safety stock formula?
- The base formula is Safety Stock = Z × σ(demand) × √(Lead Time). Where lead time also varies, the extended form is used: Z × √(LT × σ²demand + demand² × σ²LT).
- How is the reorder point calculated?
- Safety stock is added to the demand expected during the lead time. For a product selling 50 units a day with a 9-day lead time and 179 units of safety stock, the reorder point is (50 × 9) + 179 = 629 units.
- What service level should I use?
- It depends on the product class. 97-99 per cent for A-class items, 95 per cent for B, 90 per cent is enough for C. Going from 95 to 99 raises safety stock by roughly 41 per cent, and that cost cannot be justified for every product.
- What if I do not have enough data?
- Use the simple formula: maximum daily demand times maximum lead time, less average demand times average lead time. It systematically produces excess stock, so switch to the statistical formula once three months of data has accumulated.
- Should out-of-stock days be included in demand?
- No, they should be excluded. A day that sold zero because there was no stock does not mean demand was zero. Including those days lowers both the mean and the standard deviation, and safety stock is set too thin.