A reorder point without a lead time component is not a reorder point. It is a stockout schedule.
The reorder trigger has one job: create enough replenishment time for incoming stock to arrive before available inventory reaches zero. That requires two quantities, not one: expected demand during the supplier’s lead time, and a buffer for uncertainty.
The basic reorder point formula
The starting formula is:
ROP = demand rate × lead time + safety stock
Each variable has a specific meaning:
- ROP is the reorder point, measured in units.
- Demand rate is expected unit demand per time period, such as units per day.
- Lead time is the elapsed time between placing an order and receiving the stock, using the same time unit as demand rate.
- Safety stock is the additional inventory held to absorb demand or delivery uncertainty.
Suppose a retailer sells 12 units of a product per day, and the supplier usually takes 18 days to deliver. Expected demand during lead time is:
12 units per day × 18 days = 216 units
If safety stock is 60 units, the reorder point is:
216 + 60 = 276 units
When inventory position falls to 276 units, the retailer should place the order. Inventory position is more useful than the shelf count. It normally means on hand inventory plus stock already ordered, minus committed demand such as open customer orders. Using only physical stock can cause duplicate orders when a replenishment shipment is already in transit.
The formula assumes that demand rate and lead time are expressed consistently. If demand is recorded weekly, convert lead time to weeks. If a supplier takes 18 calendar days but the demand rate is based on selling days, the conversion must reflect the retailer’s operating calendar. A mismatch here can create a systematic error rather than random noise.
What the lead time term actually changes
Lead time is not an administrative field attached to a supplier. It changes the amount of demand that must be covered before replenishment arrives.
With a two day lead time, an item selling 12 units per day needs to cover about 24 units of expected demand. With an 18 day lead time, the same item needs to cover 216 units. The demand pattern has not changed. The exposure window has.
This is why a reorder rule based only on average daily sales often fails. A trigger such as “reorder at 50 units” may work for a local supplier that delivers in three days. It is structurally too low for an overseas supplier with a three week delivery cycle, even when the product has identical sales.
Lead time also affects order frequency and working capital. A higher reorder point does not automatically mean larger purchase orders. The reorder point determines when to order. The order quantity determines how much to order. Those are separate decisions and should not be combined into one unexplained stock setting.
A synthetic example from Valencia
Consider a three location women’s clothing retailer in Valencia. A supplier in Portugal provides a seasonal blouse. Across the three locations, the retailer sells an average of 10 units per day. The supplier’s average lead time is 18 days.
Expected demand during the average lead time is:
10 × 18 = 180 units
If the retailer uses 70 units of safety stock, its initial reorder point is:
ROP = 180 + 70 = 250 units
This number applies to the combined inventory position if replenishment is managed centrally. If each location orders independently, the demand rate, lead time exposure, and safety stock need to be calculated at location level. Pooling the demand can reduce some demand variability, but it also introduces transfer time and allocation decisions. A central calculation is not automatically correct for decentralized ordering.
The scenario also shows why the word “average” needs care. Eighteen days is useful as an expected value, but it does not describe whether deliveries normally arrive between 17 and 19 days or between 8 and 35 days. Those two suppliers have the same average lead time and very different replenishment risk.
Lead time variability needs its own buffer
If lead time changes from order to order, safety stock should account for that variation. A common approximation, when demand and lead time are independent, is:
Safety stock = z × square root of ((average lead time × demand standard deviation squared) + (average demand squared × lead time standard deviation squared))
Here, z is the service factor, demand standard deviation describes daily demand variation, and lead time standard deviation describes delivery variation. The first term captures uncertain demand during the delivery window. The second captures the fact that the delivery window itself can be longer or shorter than expected.
For the Valencia example, use these synthetic values:
- Average demand: 10 units per day
- Demand standard deviation: 3 units per day
- Average lead time: 18 days
- Lead time standard deviation: 5 days
- Service factor: 1.65
The uncertainty calculation is:
Square root of ((18 × 3²) + (10² × 5²)) = square root of 2,662 = approximately 51.6 units
Safety stock is therefore:
1.65 × 51.6 = approximately 85 units
The resulting reorder point is:
10 × 18 + 85 = 265 units
Compared with the earlier 250 unit setting, the difference is not large because the original safety stock was already 70 units. The important point is the reason for the increase. Five days of lead time variation contributes 50 units of standard deviation before the service factor is applied. Ignoring it would treat a late shipment as if it were a demand problem, then leave the retailer without enough inventory to absorb the delay.
The calculation is an approximation. Demand and lead time may be correlated. For example, a supplier may take longer during the same promotional periods when demand rises. In that case, the independent variable formula understates joint risk. Historical order level data should be examined for that pattern rather than assumed away.
Measure the lead time that inventory actually experiences
The useful lead time is not necessarily the number in a supplier contract. It is the time between the event that triggers the purchase order and the event that makes the units available for sale.
Depending on the operation, that interval may include:
- Approval time before the order reaches the supplier.
- Supplier processing time.
- Transit time.
- Receiving and quality checks.
- Time required to put stock into the available inventory record.
If an order is approved two days after a planner creates it, those two days belong in the replenishment process. If a shipment arrives at a warehouse but cannot be sold until receiving is completed, the operational lead time ends at receiving, not at the delivery dock.
Calculate lead time from timestamped purchase order and receipt events where possible. Keep the observations by supplier, product family, location, and order method. A single supplier average can hide meaningful differences between standard orders, seasonal orders, and urgent requests.
Use a distribution, not only an average. The median shows a typical result. The upper percentiles show how long replenishment takes during slower cases. Neither replaces the other. A high average caused by a few extreme delays can require a different response from a consistently long but predictable lead time.
Where the formula stops being enough
Some replenishment flows do not have a stable lead time. Spot buys are negotiated and accepted case by case. Emergency restocks may use a different carrier or a local wholesaler. Partial shipments may deliver some units quickly and the balance weeks later.
For a partial shipment, treating the entire purchase order as received on the first delivery can overstate available inventory. Treating the entire order as unavailable until the final delivery can make the system order unnecessarily. The inventory model should represent receipt lines or shipment lots separately, with each quantity becoming available at its actual date.
Emergency restocks also need a separate policy. Their lead time may be shorter, but their purchase price, minimum quantity, and availability can differ from the normal supplier. Mixing emergency orders into the standard lead time average can make normal replenishment look faster than it is.
There is also a counterpoint to adding more lead time detail: bad lead time data can make the model worse. If a system records a purchase order as delivered when a carrier label is created, it will learn an artificially short lead time. If cancelled orders remain in the history, it may learn an artificially long one. If a supplier changes its dispatch process but the data is not segmented by period, the average can describe no current operating condition.
A precise formula fed with incorrect timestamps produces a precise wrong answer. Before increasing safety stock, inspect the source events, remove cancelled orders, separate partial receipts, and check whether the current supplier process matches the historical sample.
The edge case that most often trips up a reorder rule is a shipment that arrives on time but is not sellable until receiving, inspection, or allocation is complete. Set the lead time endpoint at the moment stock can actually satisfy demand, not the moment a truck reaches the building.
