Demand-risk band around lead time

The service factor expands beyond the one-standard-deviation band to create 349.24 safety units. The band is descriptive under the normal approximation and is not a promise that demand stays inside it.

Reorder-point formulas

The calculation assumes independent daily demand and lead-time variability. If supplier delay occurs precisely when demand spikes, independence understates risk. Historical simulations or scenario models can preserve that correlation.

Worked component-stock example

Average demand is 100 units per day and average supplier lead time is 12 days, so expected lead-time demand is 1,200 units. Daily demand standard deviation is 20 units and lead-time standard deviation is two days. The combined lead-time uncertainty is 211.66 units.

At a 1.65 service factor, safety stock is 349.24 units. The unrounded reorder point is 1,549.24 units, and the action threshold is rounded up to 1,550 whole units. This represents 15.49 days of average demand, including 3.49 safety days above mean lead time.

On hand is 1,400, on order is 200, and backorders are 100, producing a 1,500-unit inventory position. It is 49.24 units below the unrounded threshold, so the result says order now. A planned 3,000-unit order would raise position to 4,500 and represents 30 average demand days.

Reorder point, safety stock, and order quantity do different jobs

Do not add order quantity to the reorder point as another safety buffer. The trigger and replenishment amount belong to one policy but answer separate questions.

Monitor inventory position, not shelf count alone

Inventory position commonly equals usable on hand plus reliable on-order supply minus backorders or commitments. Counting a purchase order that the supplier has not confirmed can overstate protection. Ignoring backorders can delay replenishment. Damaged, quarantined, expired, consigned, or reserved units may not be usable on hand.

Define status codes and ownership. In-transit goods can be on order operationally and owned inventory financially, depending on terms. Avoid counting transfer inventory at both origin and destination.

Update the signal after receipts, demand, cancellations, adjustments, allocations, and supplier changes. A nightly batch can be too slow for high-velocity or scarce components.

Measure end-to-end replenishment lead time

Lead time can include approval, purchase-order transmission, supplier queue, manufacturing, consolidation, origin handling, transport, customs, receiving, inspection, and put-away. The clock should end when stock is available to use or promise, not merely when a truck reaches the gate.

Average and standard deviation should come from comparable supplier, item, lane, mode, and season. Exclude data errors but do not discard real disruptions solely because they are inconvenient. Consider a separate extreme-risk buffer or dual sourcing for tail events.

Supplier quotes may state production time only. Capture timestamps from request through usable receipt and review bias by promised versus actual date.

Forecast error can be more useful than raw demand variation

If replenishment is driven by a forecast, safety stock should often protect forecast error over lead time rather than variability in the demand series itself. Promotions, price changes, product launches, customer wins, substitution, and seasonality should first enter the forecast.

Intermittent or lumpy demand violates a simple normal approximation. A part demanded once per quarter can have many zero days and rare large orders. Use intermittent-demand methods, empirical lead-time distributions, bootstrapping, or simulation.

Separate recurring consumption from projects and known orders. Customer commitments can enter deterministic lead-time demand while uncertainty covers the residual.

Cycle service and fill rate are not the same target

Cycle service is the probability of avoiding a stockout during a replenishment cycle under the model. Fill rate is the share of demand filled immediately. A given z factor does not guarantee the same fill rate across order quantities and demand distributions.

Select service by item criticality, margin, substitution, customer promise, shortage cost, downtime, and recovery speed. A lifesaving component and a slow accessory should not inherit the same target automatically.

Measure actual stockout events, units short, backorder days, fill rate, lost contribution, expediting, and customer churn. Adjust the buffer based on outcomes and data, not only a table.

Seasonality makes the signal move

A static 100-unit daily average can understate peak demand and overstate off-season needs. Recalculate lead-time forecast and variability at each review. Holiday closures can extend supplier, carrier, customs, and receiving calendars simultaneously.

Build forward demand over the exact expected lead-time window. If demand rises during the next 12 days, use the daily forecast sum rather than 12 times an annual average. Apply safety stock to residual uncertainty.

Product end-of-life requires a declining trigger and a final-buy decision. Replenishing a full order quantity near phaseout can create obsolescence that costs more than a stockout.

Supplier and logistics constraints can override the ideal trigger

Minimum order quantity, case pack, pallet, container, truck capacity, production campaign, order calendar, quantity discount, freight minimum, supplier capacity, cash, storage, and shelf life constrain the order. Round quantity only after calculating the position gap.

When several items share a supplier or container, coordinate review and transport without allowing one slow SKU to inflate stock. Evaluate the total landed and holding cost of consolidation.

If lead time exceeds the order cycle, multiple orders can be open. Maintain line-level expected dates and do not treat all on-order quantity as equally reliable.

Operational reorder data and financial inventory must reconcile

Inventory position contains operational commitments that may not yet be recognized as owned inventory, while financial records may include in-transit goods. Build a reconciliation rather than forcing one field to serve purchasing, accounting, and tax.

IRS Publication 334 discusses inventory and COGS for small businesses.

Preserve demand history, forecasts, purchase orders, confirmations, receipts, lead timestamps, backorders, inventory adjustments, service policy, and parameter versions.

Govern the signal as a living policy

Assign parameter owners and review frequency. Re-estimate demand, forecast error, lead-time mean and variation, service factor, minimums, order quantity, and usable inventory status. Flag inputs with too little history or recent structural change.

Backtest: apply the old parameter set to historical periods and count stockouts, excess stock, and emergency orders. Compare the model with a simpler lead-time-demand-plus-fixed-days rule to ensure added complexity improves decisions.

Log overrides and reasons. Buyers can have information not in the model, but repeated override patterns indicate missing data, poor parameters, or incentives.

Reorder-signal checklist

1. Use a consistent day calendar.
2. Forecast the lead-time window.
3. Measure end-to-end lead time.
4. Estimate variability by SKU and supplier.
5. Select a service objective.
6. Remove unusable on-hand stock.
7. Confirm open purchase orders.
8. Subtract backorders and allocations.
9. Check minimums and pack sizes.
10. Coordinate shared freight.
11. Backtest stockouts and excess.
12. Version parameter changes.

Frequently asked questions

References

• National Institute of Standards and Technology, supply-chain inventory dynamics and reorder-point discussion
• U.S. Government Accountability Office, ordering and carrying-cost analysis
• IRS Publication 334, business inventory and COGS
• IRS Publication 583, Starting a Business and Keeping Records