Lead time variability: the quiet killer of service levels

When service levels slip, the first suspect is usually demand. Customers ordered more than expected, so we ran out. Often, though, the real culprit sits on the other side of the equation: suppliers who promise four weeks and deliver in three, then five, then seven.

Lead time variability is easy to overlook because the average looks fine. A supplier whose deliveries average 28 days appears reliable on a dashboard. But safety stock does not protect against averages; it protects against the spread around them. If replenishment can arrive two weeks late, the buffer must cover two extra weeks of demand, and that cost is paid on every unit, every day it sits on the shelf.

The arithmetic makes the point. In the common safety stock formula that combines both sources of uncertainty, the variance of lead time is multiplied by the square of average demand. For a fast-moving item, a few days of delivery uncertainty can require more buffer than all of its demand fluctuation combined. Halving the spread of lead times can free more working capital than months of forecasting work.

That changes where planners should spend their effort. Measure actual receipt dates against promised dates, by supplier and by lane, not just the average. Share the data with suppliers; many have never seen their own variability. Reward consistency, not only speed: a dependable six weeks is often worth more than an unpredictable four. Where the spread comes from your own side, such as slow order release or receiving backlogs, fix that first, because it is entirely within your control.

Shorter lead times are welcome, but predictable lead times are what let you hold less inventory with confidence. The next time a stockout lands on your desk, look at the calendar before you look at the forecast.