How to Calculate Reorder Point When Lead Time Varies

The reorder point is the inventory control parameter that triggers a replenishment order, holding enough stock to cover expected demand across the lead time plus a buffer against what varies. The standard formula assumes the supplier delivers in a fixed number of days. On one supplier record — 100 units a day, a demand standard deviation of 20 units, a 20-day (4-week) average lead time and a lead time standard deviation of 5 days — that assumption returns 2,000 units where the correct trigger is 2,838 units, a shortfall of 838 units or 29.5%. Five inputs set the answer: average daily demand, the demand standard deviation, average lead time, the lead time standard deviation, and the Z-score for your chosen cycle service level. This page gives four formula variants, a seven-step procedure, a worked example and a validation pass against receipt history. The same figures drive continuous review triggers, periodic review target levels, min/max parameters and ABC-segmented service levels.

On this page

1. What Is a Reorder Point?
2. Why the Basic Formula Fails
3. The Formula and Its Four Variants
4. The 7-Step Procedure
5. Worked Example
6. Comparing the Four Variants
7. The Inventory Position Problem
8. Under Periodic Review
9. The Max-Method Shortcut
10. Building It in Excel
11. Multiple SKUs and Suppliers
12. When to Recalculate
13. Mistakes
14. FAQs
15. The Bottom Line

What Is a Reorder Point?

A reorder point is the stock control threshold that releases a replenishment order, and it holds two quantities at once: the demand expected across the lead time, and a buffer sized against the variability in both demand and the lead time. The reorder point answers when to order. Economic order quantity (EOQ) answers how much.

The reorder point (ROP) exists inside a continuous review system, where the inventory position is checked after every transaction. On the inputs used throughout this page, the reorder point is 2,838 units, which is 28.4 days of cover at 100 units a day.

Key Takeaways

Reorder Point vs Safety Stock vs Order Quantity

Safety stock sits inside the reorder point, not beside it. Safety stock is the 838-unit buffer that absorbs variability. The reorder point is the full 2,838-unit trigger, which also carries the 2,000 units of demand expected during the 20-day lead time. An order is placed before the buffer is touched, which is the whole point of separating the two terms.

Order quantity is a third quantity entirely. Ford W. Harris set out the economic order quantity in 1913, and R.H. Wilson later carried the same square-root result into practice as the Wilson formula. EOQ sizes the 1,500-unit purchase order; the reorder point decides the day it is released.

Minimum stock is not a synonym for the reorder point either. The reorder point is always the higher of the two, because it covers consumption during the lead time as well as the buffer. In a min/max system the min field is loaded with the reorder point, which is where the confusion starts.

The phrase "reorder point" carries four separate senses, and a planning review stalls when two people hold two of them. Select each one below.

Why the Basic Reorder Point Formula Fails With Variable Lead Time

The basic reorder point formula fails because it prices the lead time as a constant. It returns 2,000 units where the correct trigger is 2,838 units, understating the replenishment trigger by 838 units, or 29.5%, on a supplier whose deliveries land anywhere between 13 and 30 days.

That form assumes a deterministic lead time, zero safety stock and unlimited supply. Each assumption fails against a real supplier record. A constant lead time is a contractual rarity: the 21 receipts behind this article carry a 17-day spread between the fastest and slowest delivery. Zero safety stock means a 50% cycle service level, because demand lands above its average in half of all replenishment cycles.

Two further failures are structural rather than statistical. A fixed reorder point set by hand and never revisited tracks nothing, and reordering by eye against a static min level tracks less. Neither approach carries a service level, so neither can be audited when a stockout happens.

The gap between 2,000 and 2,838 units is not a rounding difference. It is 8.4 days of cover, and it is the entire lead time volatility buffer.

The Reorder Point Formula With Variable Lead Time

The reorder point formula with variable lead time adds a stochastic lead time safety term under a single square root. Define the six symbols before using them: D is average daily demand in units, LT is average lead time in days, σD is the standard deviation of daily demand in units, σLT is the standard deviation of lead time in days, Z is the service factor read from the normal distribution for the chosen cycle service level, and R is the review period in days where periodic review applies.

Hadley and Whitin set out the continuous review model under a stochastic lead time, and Paul Zipkin gives the modern treatment of the same result. Edward A. Silver and Rein Peterson carry it into planning practice, and the Association for Supply Chain Management (ASCM), whose APICS Dictionary defines the reorder point as the set stock level at which an order is placed, uses the same continuous review framing. The derivation assumes demand and lead time are independent and roughly Gaussian, which the Central Limit Theorem supports over a 20-day horizon and Step 7 tests directly.

Round once, at the end: keep intermediate results to two decimal places, round safety stock to the nearest whole unit, and round the reorder point up to the next whole unit.

What Each Part Contributes

The Four Variants and When Each Applies

There are 4 reorder point formula variants, including constant demand with variable lead time. Each one answers a different data situation, and each returns a different number on the identical input set. State the applicability condition first, then the formula, then the result.

Variant 1 — Constant Demand, Constant Lead Time

Variant 1 applies where both demand and the lead time are contractually fixed and measured, which describes a kanban loop on a just-in-time (JIT) line and almost nothing else.

Variant 2 — Variable Demand, Constant Lead Time

Variant 2 applies where the supplier delivers on a guaranteed day and only demand moves. This is the variant most ERP systems implement by default.

Variant 3 — Constant Demand, Variable Lead Time

Variant 3 applies where demand is steady and the delivery date moves, which describes contract manufacturing and most overseas replenishment.

Variant 4 — Variable Demand, Variable Lead Time

Variant 4 applies where both inputs move, which describes the large majority of purchased items. It is the only variant that prices both sources of uncertainty, and it is the default this page recommends.

Choosing Your Z-Score

To choose the Z-score, set the cycle service level first, then read the service factor from the normal distribution. Cycle service level is the probability of avoiding a stockout during one replenishment cycle. Fill rate measures something different — the fraction of demand served from stock on hand — and the two return different numbers on the same item.

Moving one SKU from 95% to 99% costs 345 units, or 3.4 days of cover. Moving it from 95% to 99.9% costs 732 units. That reorder point sensitivity is why service levels belong at the ABC class level rather than set once for the catalog.

How to Calculate Reorder Point Step by Step

To calculate reorder point step by step, run 7 steps in order: pull the receipt history, compute the two averages, compute the two deviations, match the time units, size the buffer, add lead time demand, and validate the result against what actually happened.

Step 1 — Pull the Order-Level Receipt History

To pull the order-level receipt history, export every purchase order line for one supplier-SKU pair with its order date and its goods-receipt date, then compute elapsed days per line. Order-level records are the requirement. A monthly average of lead times destroys the dispersion you are trying to measure, and an uncertain supply interval cannot be reconstructed from summary data.

The 21 receipts used here run from 13 days to 30 days:

13, 14, 14, 15, 16, 17, 17, 17, 18, 19, 19, 19, 20, 21, 22, 23, 24, 25, 27, 30, 30

Step 2 — Calculate Average Daily Demand and Its Deviation

To calculate average daily demand and its deviation, run =AVERAGE(range) and =STDEV.S(range) over the daily issue history for the same period as the receipt history. Those functions return D = 100 units per day and σD = 20 units per day on this item.

Use the issue or shipment history rather than the sales forecast. A forecast already carries a bias, and feeding it into a probabilistic reorder calculation buries that bias inside the buffer where nobody audits it.

Step 3 — Calculate Average Lead Time and Its Deviation

To calculate average lead time and its deviation, run the same two functions over the elapsed-days column from Step 1. On the 21 receipts that returns LT = 20 days (4 weeks) and σLT = 5 days.

Use the sample standard deviation, not the population form. Divide by n − 1, not n, and call STDEV.S rather than STDEV.P. Receipt history is a sample of the supplier's future behavior, not the complete data set, so the sample denominator is the correct one. Several circulating sources give the σLT formula with n in the denominator, which understates the deviation and therefore understates the replenishment trigger. On these 21 records the population form returns 4.88 days against the correct 5.00 days, and the reorder point falls by 19 units on that error alone.

Measure the lead time you actually get, not the one you were promised. The three senses below produce three different triggers.

Step 4 — Match the Time Units

To match the time units, put σD and D on the same time base as LT and σLT before either value enters the formula. Mismatched time bases are the most common source of an answer that is wrong by a whole multiple.

Demand recorded weekly converts by the length of the bucket. Weekly demand of 500 units over a 5-day working week gives D = 100 units per day. The deviation converts by the square root of the bucket length, not the bucket length itself:

Dropping a weekly deviation of 44.72 units straight into a daily formula inflates the demand term from 8,000 to 40,000, a factor of 5. Convert first, then calculate.

Step 5 — Compute Safety Stock

To compute safety stock, multiply the Z-score by the square root of both variability terms combined. Keep the radical over the whole expression.

The lead time term contributes 250,000 of that 258,000 total, or 96.9%. The demand term contributes 8,000, or 3.1%. A lead time variance adjustment of a single day moves the buffer far more than a day off the average does.

Step 6 — Add Lead Time Demand

To add lead time demand, multiply average daily demand by average lead time and add the buffer from Step 5.

Lead time demand is 70.5% of the trigger and safety stock is 29.5%. Both figures move when either input changes, which is why a dynamic reorder threshold is recalculated on a schedule rather than set once.

Step 7 — Validate Against History

To validate against history, count how many of your historical replenishment cycles would have stocked out at the calculated trigger, then compare that count against the service level you asked for. No competing page on this topic includes this step, and it is the one that catches a wrong answer.

Run the check on the 21 receipts. A cycle stocks out when demand during the actual lead time exceeds 2,838 units. The two 30-day deliveries consume 3,000 units on average demand alone and breach the trigger outright. The 27-day delivery consumes 2,700 units and breaches it whenever demand runs 138 units above average, which the demand distribution puts at roughly 1 cycle in 11. Expected shortfalls across the sample: 2.1 out of 21 cycles.

Realized cycle service lands near 90%, not 95%. The cause is distributional: the formula assumes a normal distribution, real supplier lead times are right-skewed, and the gamma and lognormal distributions fit that shape better. Here the average of 20 days sits above the median of 19 days, which is the signature of that skew.

Percentile-based sizing is the practical fallback rather than a different distribution. Replace average lead time with the observed 95th percentile of 30 days:

That lead time percentile reorder trigger costs 343 units more than the statistical answer and covers 20 of the 21 observed cycles. Use the 90th percentile instead, if the tail is driven by events that will not repeat.

Worked Example: Reorder Point for a Variable-Lead-Time Supplier

A distributor stocks one stock keeping unit (SKU) at $12 per unit, sells 100 units a day and replenishes from a single overseas supplier across a fluctuating delivery cycle. Every figure below comes from the 21 receipts in Step 1 and the matching demand history.

The arithmetic runs in full, with no line skipped:

The distributor releases a 1,500-unit purchase order when inventory position reaches 2,838 units. Safety stock of 838 units is 8.4 days of cover and $10,056 of working capital in this one item. The full trigger of 2,838 units is 28.4 days of cover.

Comparing the Four Variants on the Same Data

Variant 3 lands within 0.5% of the full formula while Variant 2 misses by 24.3%, and that single comparison identifies which input your system is failing to capture. Most ERP systems implement Variant 1 or Variant 2.

Read the table as a diagnostic. A system returning 2,148 units is running Variant 2 and pricing demand variability only, which misses 690 units of lead time volatility buffer. A system returning 2,000 units is running Variant 1 and pricing nothing.

Variant 3 is the practical shortcut where demand is steady. It needs no demand standard deviation at all, and it costs 13 units of accuracy on this data set. Use Variant 4 where you have both deviations, and Variant 3 where you have only the lead time records.

The Inventory Position Problem

Inventory position, not stock on hand, is what a reorder point is measured against. A correct formula paired with the wrong balance releases duplicate purchase orders, and this is where most working replenishment systems actually fail.

Why On-Hand Stock Is the Wrong Trigger

Stock on hand counts only what is in the building. It excludes the quantity already on order and still in transit, so an on-hand trigger fires again on every transaction until the first delivery lands. Simfoni's glossary makes the same operational observation: a correct reorder point fails when the balances feeding it are wrong.

Calculating Inventory Position

Quantity on order covers every open purchase order line, including goods in transit and stock held for inspection. Backorders are committed demand not yet shipped. Allocated stock is a fourth component some systems deduct separately; deduct it once, in one place, and document which.

When Lead Time Exceeds the Order Cycle

The lead time is 20 days and the order cycle is 15 days, because 1,500 units at 100 units a day is 15 days of supply. Divide 20 by 15 and round up: 2 purchase orders are open at once, permanently. The moment the reorder point exceeds the order quantity — 2,838 units against 1,500 — this condition is already true.

Work the case in full. Stock on hand is 1,200 units, two purchase orders of 1,500 units each are open for 3,000 units, and backorders are zero.

A system checking stock on hand alone sees 1,200 against 2,838 and releases a third purchase order for 1,500 units that nobody needs. That order arrives into 4,200 units of coverage and becomes excess stock, and the same test fires again the next day. Material requirements planning (MRP) and distribution requirements planning (DRP) both net against scheduled receipts for this reason; a hand-built reorder report usually does not.

Reorder Point Under Periodic Review

No, the reorder point does not exist under periodic review, and treating the two review modes as interchangeable is the error that produces chronic shortages on P-system items. A continuous review system, the Q-system, checks inventory position after every transaction and triggers on the reorder point. A periodic review system, the P-system, checks on a fixed calendar and triggers on the review date, comparing inventory position against a target level instead.

The exposure window changes with the review mode. Under continuous review the risk period is the lead time. Under periodic review it is the lead time plus the review period, because a shortage arising the day after a review is not seen until the next one.

On a 7-day review period the target level is 3,543 units against a reorder point of 2,838 units, a difference of 705 units or 7.0 days of cover. Order quantity under the P-system is the target level minus inventory position on the review date, so it varies every cycle.

The Max-Method Shortcut and Where It Overbuffers

The max method multiplies the highest observed daily demand by the longest observed lead time and stops there. On this data set it returns 4,500 units against the statistical answer of 2,838 units, over-buffering by 1,662 units, 16.6 days of cover, or $19,944 per SKU at $12 a unit.

The method costs that much because it stacks two worst cases that have never coincided. The 150-unit day and the 30-day delivery are each rare; the joint probability of both landing in the same replenishment cycle is far below the 5% risk a 95% cycle service level accepts. The max method also has no service level attached, so it cannot be tuned by ABC class and it cannot be audited after a stockout.

Two properties make it worth keeping as a sanity bound rather than a method. It needs 2 numbers instead of 5, and it never understates. Use it as an upper bound on a new item with fewer than 20 receipts, then replace it once the history supports a probabilistic reorder calculation.

How to Build the Reorder Point Calculation in Excel

To build the reorder point calculation in Microsoft Excel or Google Sheets, put the receipt history in one block of cells and the five parameters in another, then reference across. Three functions do the work: AVERAGE, STDEV.S and NORM.S.INV.

Lay out the receipt history first. Column A holds the purchase order date, column B holds the goods-receipt date, and column C computes elapsed days with =B2-A2 filled down 21 rows. Column E holds the daily issue history for the same period.

Two guards belong in the same sheet. H12 flags any supplier-SKU pair below 20 receipts, and a comparison of H11 against H2 flags a lead time distribution skewed far enough that Step 7 needs running. Copy the block down one row per supplier-SKU pair and the reorder point simulation scales to the whole catalog.

NORM.S.INV(0.95) returns 1.6449, rounded to 1.65 throughout this page. PERCENTILE.INC returns 27 days on this history, against an average of 20 days.

Setting Reorder Points Across Multiple SKUs and Suppliers

Calculate one reorder point per supplier-SKU pair, and per lane where one SKU arrives by more than one route. A single company-wide lead time standard deviation is the most common error in practice, and it fails in both directions at once: it overbuffers the reliable supplier and underbuffers the erratic one from the same middling figure.

Segmentation runs on three axes. The supplier-SKU pair is the base unit, because the same SKU from two vendors carries two distributions. The lane matters where a sea freight route and an air route feed the same item, since a 30-day ocean move and a 5-day air move cannot share one σLT. The ABC class sets the service level, and therefore the Z-score, as in the table above.

Sample size decides whether the pair-level figure is usable. Use 20 to 30 receipts per supplier-SKU pair as a working minimum, and 40 or more where lead time exceeds 30 days. Below 20 receipts, borrow the deviation from a comparable supplier on the same lane, size the trigger on that borrowed figure, and flag the item for review at the next receipt.

When to Recalculate Your Reorder Point

Recalculate quarterly for stable categories and monthly where lead times are moving, then add event triggers on top of the calendar. A calendar alone misses the changes that matter most, because a supplier's performance shifts between review dates rather than on them.

Six events call for an immediate recalculation:

• A supplier change or a second source added. The distribution is new, so the old σLT describes a vendor you no longer buy from.
• A lane or transport mode change. Moving an item from sea freight to air changes both the average and the spread, and transit modelling is the fastest way to see by how much.
• A demand step change above 20%. D enters the lead time term squared, so a volume move drives the buffer harder than it drives lead time demand.
• Three consecutive receipts outside 2 standard deviations. That pattern says the distribution has moved, not that three deliveries were unlucky.
• A service level or ABC class reassignment. Promoting an item from B to A moves Z from 1.65 to 2.33 and the trigger from 2,838 to 3,183 units.
• A change in the order quantity. A smaller order quantity shortens the order cycle and changes how many purchase orders stay open at once.

Test for drift before pooling 12 months of receipts. A supplier whose lead time moved from 20 days to 30 days across a year has a trend, not variance, and no reorder point formula corrects for a trend. Split the history in half, compare the two averages, and reset the parameter on the recent half where they differ by more than 2 days.

Mistakes That Break Reorder Point Calculations

Six errors account for most incorrect reorder points, and each one has a visible signature in the output.

• Triggering on stock on hand instead of inventory position. The signature is duplicate purchase orders on long-lead items, as in the 1,200-against-4,200 case above.
• Using the quoted lead time instead of the measured one. A quoted duration carries no spread, so σLT collapses toward zero and the trigger falls by hundreds of units.
• Padding the planned lead time parameter. Padding to 25 days moves the average rather than narrowing the spread, releases every purchase order early, and double-counts a buffer that safety stock already holds.
• Calling STDEV.P instead of STDEV.S. The population denominator understates the deviation on every sample, and the understatement grows as the receipt count falls.
• Mixing time bases. A weekly demand deviation of 44.72 units dropped into a daily formula inflates the demand term by a factor of 5.
• Setting one reorder point for the catalog. A fixed reorder point, a static min level, manual reordering and reordering by eye all share the same defect: no service level is attached, so nothing can be audited after a stockout.

Three assumptions behind the formula also deserve naming, because each one marks a boundary rather than a mistake. The formula assumes a deterministic relationship between demand and lead time that is actually independence, so it misstates the buffer where a supplier's delays correlate with demand peaks. It assumes unlimited supply at the vendor, which allocation during a shortage breaks. And it assumes repeat replenishment, so it does not apply to a single-period newsvendor decision or to a make-to-order item, neither of which carries a reorder point at all.

FAQs About Reorder Point With Variable Lead Time

How do you calculate reorder point with variable lead time?

Use ROP = (D × LT) + Z × √(LT × σD² + D² × σLT²). On 100 units a day, a demand standard deviation of 20 units, a 20-day average lead time, a lead time standard deviation of 5 days and Z = 1.65 for a 95% cycle service level, that gives 2,000 units of lead time demand plus 838 units of safety stock, so the reorder point is 2,838 units or 28.4 days of cover. The square root encloses both terms, and both standard deviations come from STDEV.S on your own receipt history.

What is the difference between reorder point and safety stock?

Safety stock sits inside the reorder point rather than beside it. Safety stock is the 838-unit buffer that absorbs demand and lead time variability, and the reorder point is the full 2,838-unit trigger that also covers the 2,000 units of expected demand during the 20-day lead time. You place the purchase order at 2,838 units so that stock is still 838 units above zero when the delivery lands on an average day.

Is the reorder point the same as minimum stock?

No. The reorder point is always higher than minimum stock because it covers consumption during the lead time as well as the buffer. On these inputs minimum stock is the 838-unit safety stock level and the reorder point is 2,838 units. In a min/max system the min field is loaded with the reorder point, which is why the two terms get confused in practice.

Should you use average lead time or maximum lead time?

Use average lead time in the formula and keep maximum lead time as a diagnostic only. The average of 20 days with a 5-day standard deviation returns 2,838 units, while the max method of 150 units a day times 30 days returns 4,500 units, an over-buffer of 1,662 units or 59%. Where the lead time distribution is heavily right-skewed, size on the 90th or 95th percentile instead of the maximum.

How many purchase orders do you need before the calculation is reliable?

Use 20 to 30 receipts per supplier-SKU pair as a working minimum, and 40 or more where the lead time exceeds 30 days. A standard deviation computed from 5 receipts is noise and moves by several days when the sixth receipt lands. Below 20 receipts, borrow the deviation from a comparable supplier on the same lane, size the trigger on that figure, and flag the item for review.

Does the reorder point change under periodic review?

No, because the reorder point does not exist under periodic review. A continuous review system, the Q-system, triggers on a reorder point. A periodic review system, the P-system, triggers on the calendar and compares inventory position against a target level, which must cover the review period as well as the lead time. On a 7-day review period the target level is 3,543 units against a reorder point of 2,838 units.

Why do stockouts still happen when the reorder point is set correctly?

Three causes account for most of it. The system triggers on stock on hand instead of inventory position, so orders fire late or duplicate. The lead time is drifting rather than varying, which no reorder point formula corrects. Or the normal distribution assumption understates a right-skewed supplier, which the validation step in Step 7 exposes: on the 21-receipt history used here a 2,838-unit trigger delivers roughly 90% realized service against a 95% target.

How often should reorder points be recalculated?

Recalculate quarterly for stable categories and monthly where lead times are moving, then add event triggers on top of the calendar. Recalculate immediately after a supplier change, a lane or mode change, a demand step change of more than 20%, or 3 consecutive receipts outside 2 standard deviations. A reorder point set once and carried forward is the single largest source of both stockouts and excess stock.

The Bottom Line

Four formula variants answer the same question on one input set, and the one you need is Variant 4: ROP = (D × LT) + Z × √(LT × σD² + D² × σLT²), which returns 2,838 units, or 28.4 days of cover, against 2,000 units under the basic formula. Run the seven steps in order — receipt history, the two averages, the two deviations with STDEV.S, matched time units, safety stock of 838 units, lead time demand of 2,000 units, and validation. Measure the trigger against inventory position, 1,200 on hand plus 3,000 on order minus 0 backorders, never against stock on hand alone. Then count the historical replenishment cycles that would have breached it: realized service of 90% against a 95% target means the lead time distribution is right-skewed and the percentile fallback applies. The reorder point is only as good as the receipt history behind it and the inventory position it is measured against.

Related Articles

• How Lead Time Variance Inflates Safety Stock — why σLT holds 96.9% of the radical, on the same input set.
• What Is Lead Time? Definition, Types, How to Measure — the definitional groundwork behind LT and σLT.
• How to Reduce Lead Time: 23 Strategies — tactics for the average, with days recoverable for each.
• Lead Time Calculator — model a replenishment stage by stage.
• Procurement Lead Time Calculator — requisition to purchase order issue.
• Supply Chain Lead Time Calculator — end-to-end multi-stage timelines.
• Sea Freight Lead Time Calculator — port-to-port transit and clearance.
• Lead Time Calculator in Excel — build the receipt data set STDEV.S reads from.
• All lead time calculators — the full set of 21 tools.