The reorder point answers “when do we reorder?”. Safety stock answers “what buffer do we keep against the unexpected?”. Confusing the two either ties up too much product, or starts replenishment too late.
Start from consumption during the lead time
Under continuous review, a starting formula is: reorder point = average demand during the replenishment lead time + safety stock. If demand is expressed in units per day, the lead time must use the same convention for days. Mixing working days and calendar days distorts the result immediately.
Suppose a reference is consumed at 8 units a day, with 12 days between triggering the purchase and the goods being available to sell. Expected demand during that lead time is 96 units. With a safety buffer set at 40 units, the reorder point becomes 136 units.
Reorder point136 units
- Expected demand during the lead time96 units
- Chosen safety buffer40 units
The buffer is a management assumption. The threshold is compared with the stock position and does not give the quantity to buy.
Compare the threshold with the stock position
Physical stock does not tell the whole story. Part of it may already be committed, blocked or still being received. A simplified stock position can be calculated like this: usable stock + confirmed incoming supply still to be received − firm demand not yet served. Reservations must be deducted only once.
| Element | Quantity | Treatment |
|---|---|---|
| Physical stock | 150 units | Starting point |
| Blocked stock | 10 units | Exclude from usable stock |
| Confirmed incoming supply | 60 units | Add if still expected and relevant to the horizon |
| Firm orders not yet served | 70 units | Subtract, without double-counting a reservation already deducted |
| Simplified position | 130 units | 150 − 10 + 60 − 70 |
The position is below the threshold of 136: a new order needs to be considered. But the dates matter too. If the 60 units arrive after the orders that have to be served, that balance can hide an interim stockout. A dated projection of inflows and outflows therefore complements the aggregate threshold.
Measure the complete lead time, not just transport
The lead time that matters starts when you decide to reorder and ends when the product can actually be sold or picked. It can include internal approval, waiting on the supplier, manufacturing, transport, inspection and putaway. Goods received but held in inspection are not available.
Quantify a plausible delay
In the 8-units-a-day example, a lead time of 17 days instead of 12 consumes 40 extra units. The 40-unit buffer absorbs exactly those five days if demand stays constant. If demand rises at the same time, it can become insufficient. That scenario is easier for the team to read than a safety percentage chosen without explanation.
For irregular sales, study several scenarios and the forecast errors. A statistical formula can help, provided its assumptions are stated. A target of “95%” in particular must say whether it measures cycles without a stockout or the share of units demanded that were served: those indicators are not interchangeable.
Adapt the rule to your review frequency
The starting calculation assumes the alert is monitored continuously. If you only place your purchase orders each Monday, consumption can run for several days before the next decision. A periodic review policy has to protect both the replenishment lead time and the interval until the next review.
As an illustration, with a 12-day lead time and a review every 7 days, a target level based on 19 days of demand comes to 152 units before safety stock, assuming constant demand of 8 units a day. This is an order-up-to-level logic, not the same threshold of 136 applied unchanged.
Separate the trigger from the quantity purchased
Reaching the reorder point does not mean buying a quantity equal to that point. The quantity depends on your target level, the case pack, the supplier minimum, storage capacity and product shelf life. Rounding up to the next case must stay visible.
A product with a short shelf life may need a smaller buffer and more frequent deliveries. An essential but slow-moving product may justify a buffer of its own. The right decision compares holding cost, obsolescence risk and stockout cost, without imposing the same formula on the whole product catalogue.
Check the alerts against real events
- →After a stockout: was it caused by demand, by the lead time, by wrong stock data or by an ignored alert?
- →After overstock: was a promotion or an exceptional order mistaken for regular demand?
- →At a launch: distinguish the initial assumptions from the history actually observed.
- →At every import: check units, dates and receipt statuses so that a purchase already received is not counted twice.
Finally, recorded sales sometimes understate demand when an item is unavailable. Keep the unserved demand where you can measure it. The dashboard should make stockouts and delays visible: a fall in sales caused by unavailability should not automatically reduce the next forecast.
By Anatole Briant
By Anatole Briant
By Anatole Briant
COO of Salesia
Pierre Lerebours is COO of Salesia. He is your contact for implementation, training and ongoing support.