Safety stock is the extra stock that covers the days when sales run above average. Enter how much your daily demand swings and how long delivery takes: the table shows the safety stock for every service level, so you see what each extra percent of cover costs.
| Service level | z | Safety stock | Reorder point | Extra vs 90 % |
|---|
The calculator runs in your browser. Nothing you type is sent anywhere.
Safety stock = z × σ × √(lead time + review period)
Read the last column of the table: a percent near the top costs far more stock than a percent lower down. In the example below, the half point from 99 % to 99.5 % costs two thirds as much as the five points from 90 % to 95 %. That is why most items do not need the top rows.
Golden Inventory's demand planning starts from this formula, with the same z table, then tests the minimum against your past sales before it proposes it.
A café sells 30 cartons of milk a day, but some days 20 and some days 45: σ = 9. The dairy delivers 3 days after the order, and the café orders every 2 days.
| Service level | Safety stock |
|---|---|
| 90 % | 25.76 |
| 95 % | 33.21 |
| 99 % | 46.89 |
| Max–min rule: 48 × 4 − 30 × 3 | 102 |
At 95 % the café keeps about 33 cartons of safety stock, and its reorder point is 30 × 3 + 33.21 = 123.21, rounded up to 124. The max–min rule asks for 102 cartons of safety stock — three times as much, for milk that expires. It assumes the busiest day and the slowest delivery come together every time.
Take the units sold per day for the last 30 to 90 days you were open. In Excel, =STDEV.S(range) is σ and =AVERAGE(range) is the daily demand. Leave out days with no stock: a zero there is a missed sale, not a slow day.
When you forecast, what matters is how wrong the forecast is, not how much sales move. Golden Inventory's demand planning takes the mean absolute error of its own forecast in a backtest and converts it: σ ≈ 1.25 × the mean absolute error of one forecast period, scaled down to one day (÷ √ days in the period). A steady item that is easy to forecast gets a small buffer even if its sales are large.
Put the units sold on each day in column A, one row per day — for example A2:A91 for 90 days. Then, in the cells beside them:
=AVERAGE(A2:A91)=STDEV.S(A2:A91)=D6*D3*SQRT(D4+D5)=MAX(A2:A91)*D9-D2*D4With the café's milk — 30 a day, σ 9, lead time 3, review 2, 95 % — D7 gives 33.21. The max–min rule gives 102 when the busiest day was 48 and the slowest delivery 4 days.
The extra stock you hold above the expected demand during the lead time. It absorbs the days when sales run higher than average, so a normal swing does not empty the shelf.
Start at 95 % for most items. Go higher for items whose stock-out loses the customer, and lower for slow items with a high holding cost. Each step up costs more stock than the one before.
Not wrong, but blunt. It assumes the worst day and the slowest delivery happen together, every time. It often gives more stock than a 99 % service level.
This calculator treats the lead time as fixed. If deliveries swing a lot, use the longest lead time you see often (not the single worst one), or enter it in the max–min comparison to see the upper bound.
Add the demand during the lead time: reorder point = daily demand × lead time + safety stock. The reorder point calculator does both steps and the order-up-to level.
Demand planning (Pro plan) measures σ from its forecast error and proposes a minimum for every item. Start on the free plan, no card.