Inventory replenishment: how should your ERP decide when and how much to reorder?

What is inventory replenishment, and what should the ERP decide?

Inventory replenishment is the process of restoring stock before customers feel a shortage. For each stocked item and location, it answers three questions: when to reorder, how much to order and where the stock should come from. The source can be a supplier purchase order, a transfer from another warehouse or, for a manufacturer, a work order. An enterprise resource planning system (ERP) holds most of the records these answers need, so it is usually the right place to calculate them.

The ERP should propose the order; a buyer should own the exceptions. A reorder suggestion that nobody can explain gets overridden, and a rule that is overridden every week stops being a rule. The aim is a calculation simple enough for a buyer to check in a minute, using records the ERP already keeps accurate: demand, lead time, open orders, reservations and pack sizes.

Which replenishment method fits which items?

There is no single best method. A distributor usually runs two or three side by side, chosen by item behaviour. The table compares the common ones on the same criteria.

Material requirements planning (MRP) and distribution requirements planning (DRP) belong to the forecast-driven row. If your question is whether you need a separate planning tool at all, the MRP vs ERP guide covers where planning sits. The rest of this article uses the reorder point, because it is where most distributors start and where most of the avoidable errors appear.

How do you calculate the reorder point and safety stock? A worked example

Consider an illustrative distributor stocking hose clamp HC-210 in a main warehouse. The main warehouse sells an average of 40 units a day, including transfers to a branch, with a standard deviation of 12 units a day. The supplier quotes a 10-day lead time, but deliveries have varied by about 2 days either side. The business wants a 95% cycle service level: in 95 of every 100 replenishment cycles, stock should last until the order arrives.

Safety stock covers the variation. With a 95% service level, the factor from the standard normal distribution table is 1.65. Other common factors are 1.28 for 90% and 2.33 for 99%. When only demand varies, safety stock is the factor × daily demand variation × √(lead time). When lead time also varies, add both sources of variation inside the square root.

The third row is the one teams miss. Two days of supplier variation more than doubles the safety stock, from 63 to 147 units, because a late delivery leaves the item exposed to 40 units of demand for each extra day. If the ERP stores only the quoted lead time, the safety stock looks smaller and cheaper than the real risk requires. Measure actual lead time from purchase-order date to goods receipt, per supplier, and refresh it at least quarterly.

For imported items, weigh the order quantity against landed cost rather than supplier price. Freight, duty and customs fees can add a fifth to the cost of a container, and they change the balance between ordering more often and holding more stock. The landed cost guide works through one shipment.

A cycle service level is not the same as a fill rate. A 95% cycle service level allows a stockout in about one cycle in twenty, but in those cycles most demand is still filled. Agree which measure the business means before it sets the factor, and set it by item class: for example, 99% for items whose absence stops a customer's production line, 90% for slow items that customers will wait for.

Why does the reorder point trigger too late?

Back to the buyer in the introduction. HC-210 has 620 units on the shelf, which is above the 547-unit reorder point, so nothing triggers. But 180 of those units are already reserved for accepted B2B orders waiting to ship, and no purchase order or transfer is open. The inventory position is 620 − 180 + 0 = 440 units, which is 107 below the reorder point. The order should have gone to the supplier already. By the time on-hand stock falls below 547, the safety stock is being spent before the supplier has even received the order.

ERPs handle this in different ways, so check yours. ERPNext's automatic material request, for example, compares the item's projected quantity with the reorder level, not the stock on the shelf. Whatever the system, find out which figure the trigger reads, and whether reservations from every sales channel reach it.

Three inputs that go stale

• Orders from other channels. If ecommerce or portal orders reach the ERP in a nightly batch, today's reservations are invisible until tomorrow. The B2B customer portal guide covers how orders should reach the ERP.
• Stock held by a 3PL. First prove that warehouse and ERP balances describe the same unit, state and cutoff: the inventory reconciliation ledger and template shows how to bridge late movements without overwriting balances. A third-party logistics (3PL) warehouse that sends one stock snapshot a night leaves the ERP up to a day behind. For HC-210, treating that delay as one extra day of lead time raises the reorder point from 547 to 588 units. A faster stock feed is often cheaper than carrying the extra 41 units for ever.
• Units of measure. A purchase order entered in cases and a reservation counted in units must both convert to the stock unit before they are added. One unconverted case of 50 on a purchase order shows as 1 unit and makes the position 49 units too low.

How should replenishment work across several warehouses?

With more than one location, each one needs its own reorder point and its own source. In the example, the branch sells 8 HC-210 units a day, with a standard deviation of 4, and is restocked by transfer from the main warehouse in a reliable 2 days. Its safety stock is 1.65 × 4 × √2 = 10 units, and its reorder point is 16 + 10 = 26 units. The branch should raise a transfer request, not a purchase order: buying for the branch directly splits supplier volume and can miss the case minimum.

The main warehouse must then count branch transfers as demand. In the example, its 40 units a day are 32 units of its own customer orders plus 8 units of transfers. If its forecast counts only its own customers, and the variation stays the same, its reorder point falls from 547 to 443 units. The main warehouse runs short by about 104 units, exactly when the branch asks for stock.

When stock sits with a 3PL, decide who owns the replenishment decision before setting any numbers. The ERP should keep ownership and treat the 3PL as a location. The connection then has to carry stock, receipts and adjustments back promptly enough for the position to be current.

Which ERP records drive the replenishment decision?

Most replenishment errors trace back to a master record rather than the formula. Before tuning safety stock, check that each record below has a named owner and a review date.

Exclude exceptional orders from demand history. A one-off project order for 2,000 units is not part of normal daily demand; leave it in, and the average and variation both rise and the ERP holds excess stock for months. Mark such orders when they are entered, so the history can exclude them without manual cleaning.

How do you set up replenishment in an ERP without disrupting buyers?

Suggestion mode is the control point. Every difference between the ERP's suggestion and the buyer's order has one of three causes: a wrong record, a rule that does not fit the item, or knowledge the buyer has that the system lacks, such as a supplier shutdown. The first two get fixed; the third becomes a documented override.

How do you test replenishment before buyers trust it?

Run these scenarios on a copy of live data, or with test items in the live system, before switching on automatic requests. Each test has a clear pass condition, so the result does not depend on opinion.

When is a reorder point the wrong replenishment method?

A reorder point assumes that recent demand predicts the next lead time. That breaks for new items with no history, seasonal items, promoted lines and goods made or bought only against a customer order. Plan these from a forecast or from the order itself, and keep them out of the automatic rule until their demand settles.

Very slow, expensive items are another exception. An item that sells three times a year does not have a meaningful daily variation, and the formula will either hold too much or nothing at all. Keep a deliberate minimum, often one, with a named owner, or buy it only against a customer order.

Replenishment sits on top of the order and stock flows described in the wholesale order management guide. If those flows are reliable, the reorder rule becomes a quick check for buyers. If they are not, fix the reservations and stock feeds first. No safety-stock factor makes up for an inventory position that is a day out of date.