Calculate a reorder point without confusing sales with demand

Build one row per SKU and stocking location. Record the demand window, units sold, days available, current sellable stock, allocated stock, open replenishment, and expected lead time. A case of twelve and a single unit cannot be summed until their units are reconciled. A warehouse's available stock may not be immediately available to another location.

Use sellable stock rather than physical count when some units are damaged or reserved. Keep returns distinguishable from fresh demand. Include an order only once, and separate cancelled lines. The inventory use case gives context for reviewing an export before turning it into a purchasing rule.

Suppose an item sells 120 units over 30 days when it was available throughout. Its observed average is four units per day. If it was unavailable for ten of those days, dividing by all 30 days understates the rate while it was in stock. Dividing by 20 yields six units per available day, but even that estimate assumes those days are representative.

A promotion or holiday can violate that assumption. Keep the raw sales, availability days, and the selected planning rate in separate columns. Label a forecast as a forecast. Observed sales during a stockout cannot reveal the orders that customers would have placed if the item had been available.

Using the four-unit rate, a ten-day replenishment lead time implies 40 units of expected lead-time demand. With an explicitly chosen 15-unit safety buffer, the reorder point is 55 units. Shopify's reorder-point guidance uses the same basic structure: lead-time demand plus safety stock.

The example's 15-unit buffer is a planning assumption, not a statistically calibrated service level. Do not describe it as providing a particular probability of avoiding stockout. Estimating that probability would need a model of lead-time and demand uncertainty, with assumptions checked against the available history.

Suppose there are 38 units on hand, eight allocated to existing orders, and 20 on a confirmed inbound order. Inventory position is 38 - 8 + 20 = 50 units. That is below the 55-unit reorder point, so it triggers a purchasing review under the stated rule. If inbound stock cannot arrive when needed, a single inventory-position number can hide a near-term shortage.

Keep expected receipt dates visible. A time-phased review may show that the item runs out before an already-ordered shipment arrives. Conversely, placing a duplicate order because an inbound shipment was omitted can create surplus. Check open purchase orders before translating an alert into a new commitment.

The reorder point is not an instruction to order 55 units. An order quantity must account for the desired coverage horizon, existing orders, pack sizes, minimum quantities, storage, and expiry constraints. State those rules separately so a trigger does not silently become an order size.

Reconcile inventory movement against receipts, sales, returns, and adjustments. The messy CSV guide covers import checks that can change the calculation. Upload an export to review free findings, then validate the planning assumptions before purchasing stock. A checked dataset improves the decision; it does not remove uncertainty from future demand.