Blog · Procurement and supply chain · Supply chain
Safety stock covers uncertainty beyond expected replenishment demand. Work through demand and lead-time inputs, compare service targets, and reconcile the buffer to a reorder point.
The inventory safety stock formula converts demand and replenishment uncertainty into an extra quantity to hold. It needs a stated service target, consistent time units, and a clear replenishment policy. Expected demand belongs in the replenishment plan; safety stock covers the uncertainty around it.
For continuous review with fixed replenishment lead time and independent daily demand:
Safety stock = z × daily demand standard deviation × SQRT(lead time in days)
When lead time also varies, a common approximation is:
Safety stock = z × SQRT(L × σd² + d² × σL²)
Reorder point = d × L + safety stock
Here, d is average daily demand, σd is its standard deviation, L is average lead time in days, σL is the standard deviation of lead time, and z is the service factor. Stanford's supply-chain lecture notes set out the mean and variance of lead-time demand and distinguish cycle service from the share of demand filled.
The combined expression assumes stable, independent demand increments and independence between demand and lead time. Using a normal service factor also assumes the lead-time-demand distribution is reasonably approximated by a normal distribution. Intermittent demand, shared disruptions, or orders arriving out of sequence require closer examination of actual replenishment experience.
A cycle service target means the intended probability of avoiding a shortage during a replenishment cycle. It is not a promise to fill the same percentage of units immediately. Peter King's safety-stock article, hosted by MIT, explains this distinction and why predictable seasonality should be separated from random variability.
An Ohio maintenance-parts distributor is reviewing one SKU at one warehouse. The following deliberately short lists make the arithmetic reproducible; they are insufficient to establish a production stocking policy. Demand days and completed replenishments are separate observations, not matched pairs.
| Demand day | Requested units | Deviation from mean | Squared deviation |
|---|---|---|---|
| D1 | 80 | -20 | 400 |
| D2 | 80 | -20 | 400 |
| D3 | 100 | 0 | 0 |
| D4 | 120 | 20 | 400 |
| D5 | 120 | 20 | 400 |
| Total | 500 | 0 | 1,600 |
Mean daily demand is 500 / 5 = 100 units. Sample variance is 1,600 / (5 - 1) = 400, so daily demand standard deviation is SQRT(400) = 20 units.
| Completed replenishment | Lead time in working days | Deviation from mean | Squared deviation |
|---|---|---|---|
| R1 | 4 | -1 | 1 |
| R2 | 5 | 0 | 0 |
| R3 | 5 | 0 | 0 |
| R4 | 5 | 0 | 0 |
| R5 | 6 | 1 | 1 |
| Total | 25 | 0 | 2 |
Mean lead time is 25 / 5 = 5 working days. Sample variance is 2 / (5 - 1) = 0.5 days squared; standard deviation is approximately 0.707107 days. Demand is also measured per working day, so the calendars agree.
In the underlying exports, retain SKU, site, requested quantity, demand date, purchase-order line, release date, and usable receipt date. Match replenishment start and finish consistently, including receiving or inspection delays when stock remains unusable. The supplier lead-time variability guide covers that receipt analysis separately.
Include confirmed zero-demand working days. Distinguish closures from missing records and unavailable stock from absent demand; an inferred out-of-stock is evidence to investigate before accepting the sales series.
Management chooses an illustrative 95% cycle service target. The normal service factor is approximately 1.644854. Use full precision in the calculation, display intermediate quantities to six decimals where needed, and round the final buffer upward to whole units.
| Step | Calculation | Result |
|---|---|---|
| Demand contribution to variance | 5 × 20² | 2,000 |
| Lead-time contribution to variance | 100² × 0.5 | 5,000 |
| Total lead-time-demand variance | 2,000 + 5,000 | 7,000 |
| Lead-time-demand standard deviation | SQRT(7,000) | 83.666003 units |
| Unrounded buffer | 1.644854 × 83.666003, using full precision | 137.618328 units |
| Safety stock | Round buffer upward | 138 units |
| Expected lead-time demand | 100 × 5 | 500 units |
| Reorder point | 500 + 138 | 638 units |
The identity is 638 - 500 = 138: subtract expected replenishment demand from the reorder point and recover the buffer. Squaring the unrounded standard deviation must recover 7,000, the sum of the variance contributions. Both checks catch a missing square or an accidental addition of standard deviations.
Under continuous review, use inventory position to assess the reorder trigger: on-hand stock plus eligible open replenishment minus outstanding demand. Define outstanding demand once so allocated orders and backorders are not deducted twice. The base-stock inventory method addresses target replenishment policies; safety stock is only the uncertainty component.
Place the demand observations in B2:B6 and the lead-time observations in E2:E6. Set H2 to the service target, entered as 95%. The whole-unit buffer is:
=ROUNDUP(NORM.S.INV(H2)*SQRT(AVERAGE(E2:E6)*STDEV.S(B2:B6)^2+AVERAGE(B2:B6)^2*STDEV.S(E2:E6)^2),0)
If that buffer is in H3, calculate the reorder point with:
=AVERAGE(B2:B6)*AVERAGE(E2:E6)+H3
Microsoft documents STDEV.S as a sample standard deviation using the n - 1 method. Its NORM.S.INV documentation defines the inverse standard normal cumulative distribution, which supplies z. Keep numeric zeros in the ranges and resolve missing observations explicitly.
Covirage is not affiliated with Microsoft.
Use the same demand rows and an assumed unit acquisition cost of $12.50 throughout. Only the lead-time-variability assumption or the service target changes below.
| Scenario | Service factor, displayed | Lead-time-demand variance | Rounded buffer, units | Reorder point, units | Buffer value (USD) |
|---|---|---|---|---|---|
| Fixed lead time, 95% target | 1.644854 | 2,000 | 74 | 574 | 925.00 |
| Observed lead-time variability, 95% target | 1.644854 | 7,000 | 138 | 638 | 1,725.00 |
| Observed lead-time variability, 99% target | 2.326348 | 7,000 | 195 | 695 | 2,437.50 |
The fixed-time calculation is ROUNDUP(1.644854 × SQRT(2,000), 0) = 74 using full precision. Moving to observed lead-time variability adds 138 - 74 = 64 units, or 64 × $12.50 = $800. Raising the target next adds 195 - 138 = 57 units, or $712.50.
Buffer value is quantity multiplied by unit cost, not annual carrying expense. Procurement owns the receipt assumptions, planning owns the demand series, and commercial and finance leaders should agree on service priorities and cash exposure. Historical days inventory outstanding measures a different question: how much inventory the business has carried relative to cost of sales.
Bring demand history, purchase-order and receipt records, and an inventory export; Covirage maps the columns, and its tools compute and check the supplied SKU and site scenarios. The external AI model explains the results with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.
See Covirage for supply chain teams for the supporting business context.
Safety stock is the uncertainty buffer. A reorder point adds expected demand during replenishment to that buffer. Under continuous review, compare the reorder point with inventory position, including eligible open supply and outstanding demand, rather than treating physical shelf stock as the entire position.
For unchanged demand and lead-time uncertainty, a higher normal-distribution service factor produces a larger buffer. The target should reflect shortage consequences and carrying capacity. A target based on cycles without shortages is different from a target based on units fulfilled immediately.
Sales history is useful when it reflects demand. During a stockout, shipped sales can fall because inventory was unavailable. Keep backorders, lost-demand evidence, promotions, and closures visible so the calculation does not interpret constrained sales as a permanent drop in demand.
Review inputs when demand patterns, supplier performance, service commitments, or replenishment rules change. Use a regular planning review as well. Preserve the prior assumptions and compare subsequent shortage experience with the chosen target before increasing or reducing the buffer.