Safety stock when lead times cross oceans

Safety stock rules are usually inherited from a world of domestic reorder cycles, where lead times are measured in days and a shipment that runs late can be chased down. Carry those rules into cross-border trade, where lead times run to months and vary from shipment to shipment, and they quietly stop working. The buffer is either too large, draining cash, or too small, failing exactly when a shipment runs long. Fixing it starts with being clear about what safety stock is for.

Safety stock is insurance against uncertainty

A buffer exists to absorb uncertainty, in demand and in supply, so that normal variation does not cause a stockout. For an importer, supply is usually the larger source of uncertainty, because transit times vary and the lead time is long enough that a lot can happen inside it. The buffer is an insurance premium against that uncertainty. Like any insurance, you can pay too much for it, and most flat rules do, because they size the premium to something other than the risk.

Length is not the trigger; variability is

The intuition that a long lead time demands a big buffer is half right and expensive. Duration alone does not require safety stock; uncertainty does. A lead time of ninety days that is reliably ninety days can be planned around precisely, and needs little buffer beyond demand variation. A lead time that averages forty days but swings from twenty-five to sixty needs a real buffer, because you cannot trust the average. Sizing to length overcharges the steady items and still leaves the erratic ones exposed. The thing to measure is the spread, not the duration.

The flat rule is wrong in both directions

A single weeks-of-cover rule across the catalog is simple and comforting, and it errs both ways at once. Steady, predictable items get more buffer than their uncertainty justifies, tying up cash and warehouse space. Erratic items get less than they need and stock out anyway, which prompts someone to raise the rule for everything, making the overstock worse. The rule fails because it ignores the one thing safety stock should respond to, which is how much each item actually varies. Items are not uniform in their uncertainty, so a uniform buffer cannot be right.

Size the buffer to the item

The alternative is to size each item's buffer to its measured uncertainty: the variability of its demand over the lead time, and the variability of the lead time itself. That produces larger buffers where they are earned and smaller ones where they are not, which both protects service and releases the working capital trapped in blanket cover. It turns safety stock from a policy set once into a number that reflects each item's real risk, and it connects directly to the days-of-supply view that drives reordering, because both are built from the same uncertainty.

The inputs have to be real and current

Right-sizing a buffer depends on two measured inputs: demand variability from sales or usage history, and lead-time variability from actual transit and receipt data, not quoted transit times. The quoted time is a promise; the buffer has to cover what actually happens. Holding demand history and real transit actuals together, current and joined across the ERP and carrier systems, is what makes the sizing trustworthy. Without that, safety stock stays a guess dressed up as a policy, which is where most of the overpayment hides.

If your safety stock is a flat rule carried over from shorter lead times, it is almost certainly costing you cash on the steady items and still letting the erratic ones run short. Tell us how your lead times behave, or see how the platform models inventory and buffer.