
The inventory a shipment requires is afunction of two things: how much you sell per day, and how uncertain thearrival date is. The first is a forecast. The second is measurable, and it isthe one most importers never measure.
Inventory exists to cover the gap betweenwhen you buy and when you can sell. The size of that gap is a freight question,and most importers answer it by feel rather than by arithmetic.
The arithmetic is not complicated, anddoing it changes what you are willing to pay for a faster sailing.
There is the lead time you plan for and thelead time you experience. The first is a quotation. The second is adistribution, and the distribution has a long tail.
Planning against the average means halfyour shipments arrive late. Planning against the worst case means carryinginventory you do not need. The correct basis is a high percentile of the actualdistribution — usually the 90th or 95th — calculated from your own arrivalrecords, not from the carrier's transit time.
The distinction has a practicalconsequence: you can improve the second number without changing your supplier.A different carrier, a different gateway or a pre-booked drayage appointmentall reduce variance, and reducing variance reduces the inventory requirementimmediately.
The measurement itself needs care. Arrivalat the port is not arrival at the warehouse, and the gap between them is partof the lead time the inventory has to cover. Using the vessel arrival dateunderstates the requirement by several days on every shipment, consistently inthe same direction, which is the worst kind of error because it never averagesout.
1. Collect the last twelve monthsof vessel arrival dates and warehouse receipt dates. The gap between them isyour real destination lead time.
2. Calculate the average and thestandard deviation of the total door-to-door days.
3. Set the planning lead time atthe average plus 1.65 standard deviations for a 95 percent service level, orplus 1.28 for 90 percent.
4. Multiply the difference betweenthe planning lead time and the average by your daily demand. That product isthe safety stock the variance requires.
The service level factor is a choice, not alaw. A factor of 1.65 produces a 95 percent probability of not stocking outduring a lead time; a factor of 2.33 produces 99 percent. The second requiresroughly 40 percent more safety stock than the first, which is a large amount ofworking capital for four percentage points of protection.
If a routing saves five days and your dailydemand is a thousand units, that routing reduces the inventory you must hold byfive thousand units. Multiply by your unit cost and by the cost of capital, andyou have the annual value of the speed.
Compare that to the rate premium for thefaster service. On high-value, fast-moving goods the faster service often paysfor itself; on low-value, slow-moving goods it usually does not.
The same arithmetic works in reverse for aslower routing. A service that saves 15 percent on freight but adds six days oftransit requires six more days of inventory, and on high-value goods thecarrying cost of that inventory usually exceeds the freight saving within asingle quarter.
Situation | Average transit | Days of safety stock needed at 95 percent |
Stable lane, reliable carrier | 26 days | Roughly 4 to 6 days |
Mixed carriers, frequent blanks | 26 days | Roughly 9 to 14 days |
Peak season on a congested gateway | 26 days | Roughly 15 to 20 days |
Two lanes with the same average can requirevery different inventory. That is why the cheapest carrier is sometimes themost expensive supplier of working capital.
That is the quiet argument for paying apremium for reliability. Two carriers with the same average transit butdifferent variance produce different inventory requirements, and the differencein working capital is a permanent annual cost that never appears on a freightinvoice.
Most importers set a safety stock levelonce and revisit it after a stockout. The better practice is to recalculatequarterly from the actual arrival data, because both the average and thevariance move with the market.
When a gateway becomes erratic, theinventory plan should reflect it before the stockouts do. When a lanestabilises, the inventory should be released back into cash.
Transit time costs money twice: once asfreight, and again as the inventory that covers its uncertainty.
The loop needs one more input: the salesforecast. Safety stock protects against lead time and demand variance together,and a promotional period raises demand variance for its duration. Planning apromotion without adjusting the safety stock is planning to stock out.
The loop also needs a review of the servicelevel itself. A 98 percent target set for a product launch is expensive tomaintain two years later, and a 90 percent target set for a commodity may betoo thin once a competitor starts stocking the same item. The target is abusiness decision that should be revisited, not inherited.