How do you calculate safety stock when supplier lead times vary?
Base the buffer on uncertainty in demand during the replenishment lead time, not just average daily usage. Use actual lead-time observations, define the service target and validate the proposed buffer against real demand and receipt history.
Measure the exposure period consistently
Define when lead time starts and ends: order release to usable receipt is often more useful than a supplier's quoted shipping time. Include relevant receiving or quality-release delays. Use the same time units for lead time and demand.
Check which observations belong together. Different suppliers, transport methods or production conditions may represent different replenishment processes. Combining them can hide the cause of variation.
Estimate demand during each lead-time window
Where sufficient representative history exists, calculate the demand accumulated over observed replenishment windows. Examine the distribution, including unusually high demand and late receipts. A model needs to reflect both demand uncertainty and supply uncertainty.
In a continuous-review system, a reorder point is commonly expressed as expected demand during lead time plus safety stock. A periodic-review policy must also consider exposure until the next review; the same calculation cannot be transferred blindly.
Define the service measure before choosing a buffer
A cycle service level describes the likelihood of completing a replenishment cycle without a stockout. A fill rate describes the share of demand supplied from available stock. These are not interchangeable targets.
One empirical approach is to select an appropriate percentile of representative lead-time demand for a cycle-service objective, then compare it with mean lead-time demand to estimate the buffer. Sparse history and changing demand make that estimate uncertain, so validation remains essential.
Know when a simple formula is not enough
Common normal-distribution approximations make assumptions about demand, lead-time variability and independence. They can be misleading when demand is intermittent, lead times are highly skewed or demand and delay are correlated. Recent operational changes can also make old observations unrepresentative.
Back-test the proposed policy and review order constraints, material criticality and the cost of shortages. Improving lead-time reliability or stock accuracy may be more useful than adding a larger buffer. Safety stock reduces exposure; it does not insure against every disruption.
Sources and further reading
General planning guidance, not a recommendation for a specific material or business. Validate policies against your own data and constraints.