Safety Stock Calculator

Last updated: March 11, 2026
Reviewed by: LumoCalculator Team

Estimate safety stock with the service-level formula z x sqrt(L x sigma_d^2 + d^2 x sigma_L^2), then use the same inputs to check reorder point, buffer coverage, and carrying cost before you raise or lower a SKU policy.

Safety Stock Inputs

Use average demand, variability, and service level to size the statistical buffer and the reorder trigger for the same SKU.

Quick Scenarios

Cost assumptions
$
Safety stock = z x sqrt(L x sigma_d^2 + d^2 x sigma_L^2). If lead-time std. dev. is zero, the model reduces to z x sigma_d x sqrt(L).

Safety Stock Summary

Balanced policy

80 units

Service target 95% with z-score 1.64

Reorder point

830 units

Lead-time demand

750 units

Buffer coverage

1.07 days

Annual carrying cost

$563

This is a common operating target when teams want a practical trade-off between service level and carrying cost. Most of the buffer is coming from demand noise, so forecast cleanup and segmentation are the first levers to pull. The current carrying-cost load is relatively moderate, but it should still be reviewed against actual service performance.

Dominant driver: Demand variability. Current stockout risk assumption: 5%.

Detailed Breakdown

Safety-stock formula

SS = z x sqrt(L x sigma_d^2 + d^2 x sigma_L^2)

= 1.64 x sqrt(10 x 12^2 + 75^2 x 0.4^2)

Result: 80 units

Reorder-point formula

ROP = average lead-time demand + safety stock

= (75 x 10) + 80

Result: 830 units

MetricValue
Average daily demand75 units
Demand std. dev.12 units
Average lead time10 days
Lead-time std. dev.0.4 days
Cycle service level95%
Demand coefficient of variation16%
Demand variance share61.5%
Lead-time variance share38.5%
Buffer value$2,560
Daily carrying cost$2.17
Annual carrying cost$563
Carrying cost rate22%

Assumption notes

  • The model uses cycle service level plus standard deviations, not only max and average history.
  • Setting lead-time std. dev. to zero collapses the formula to the demand-only version.
  • Reorder point is the release trigger, not the recommended purchase quantity.

Current scenario highlights

  • Dominant driver: Demand variability
  • Buffer coverage: 1.07 days of average demand
  • Working-days assumption: 260 days per year

Editorial & Review Information

Reviewed on: 2026-03-11

Published on: 2025-09-11

Author: LumoCalculator Editorial Team

What we checked: Formula math, service-level assumptions, example arithmetic, boundary statements, and source accessibility.

Purpose and scope: This page supports replenishment planning for one SKU or item class. It does not replace ERP master data, MOQ rules, or supplier contracts.

How to use this review: Keep demand and lead-time units aligned, use recent history from the same operating pattern, and compare the carrying-cost impact before changing service targets across a full category.

Use Scenarios

Supplier-lane reset

Re-estimate the buffer after a port change, supplier switch, or customs delay pattern so the team can separate transit noise from demand noise.

ABC service policy

Apply one method across A, B, and C items, then pair the result with the Reorder Point Calculator when you need the on-hand trigger rather than only the statistical buffer.

Working-capital review

Use the carrying-cost output to challenge whether a higher service target is actually cheaper than expediting before applying the same service policy across a full category.

Formula Explanation

1) Safety-stock equation

Safety stock = z x sqrt(L x sigma_d^2 + d^2 x sigma_L^2)

The equation converts a service-level target into a z-score and applies it to the combined uncertainty from demand and lead time. If lead-time variability is zero, the formula collapses to the simpler demand-only version.

2) Demand uncertainty term

Demand term = L x sigma_d^2

This term grows when the replenishment window is long or when daily demand is noisy. It is the part of the buffer you attack with better forecasting, segmentation, or event cleanup.

3) Lead-time uncertainty term

Lead-time term = d^2 x sigma_L^2

A volatile supplier lane can dominate the full calculation because average demand is squared in this term. When the lead-time share is high, supplier reliability often matters more than squeezing the service target a little higher.

4) Reorder point and carrying cost

Reorder point = average lead-time demand + safety stock

Annual carrying cost = safety stock x unit cost x carrying-cost rate

The reorder point tells you when to release the order. The carrying-cost math translates the buffer into cash terms so the policy can be reviewed against service value instead of only units.

Example Cases

Case 1: Regional wholesaler

Inputs

  • Average demand: 80 units/day
  • Demand std. dev.: 14 units/day
  • Average lead time: 9 days
  • Lead-time std. dev.: 0.5 days
  • Service level: 95%

Computed Results

  • Safety stock: 96 units
  • Reorder point: 816 units
  • Annual carrying cost: $538
  • Buffer coverage: 1.2 days

Interpretation

This is a moderate buffer for a stable domestic lane. Demand and lead-time variability both matter, but neither one overwhelms the policy.

Decision Hint

Keep the 95% target for B items unless supplier performance worsens or stockout cost rises.

Case 2: Import lane with volatile arrivals

Inputs

  • Average demand: 110 units/day
  • Demand std. dev.: 18 units/day
  • Average lead time: 21 days
  • Lead-time std. dev.: 4 days
  • Service level: 98%

Computed Results

  • Safety stock: 920 units
  • Reorder point: 3,230 units
  • Annual carrying cost: $3,533
  • Lead-time variance share: 96.6%

Interpretation

The buffer is being driven almost entirely by supplier and transit variability, not by daily demand swings.

Decision Hint

Reduce lane variability first, because service-level tightening alone would make an already large buffer even more expensive.

Case 3: Critical spare parts

Inputs

  • Average demand: 12 units/day
  • Demand std. dev.: 5 units/day
  • Average lead time: 30 days
  • Lead-time std. dev.: 6 days
  • Service level: 99.5%

Computed Results

  • Safety stock: 199 units
  • Reorder point: 559 units
  • Annual carrying cost: $5,015
  • Buffer coverage: 16.6 days

Interpretation

Low-volume parts can still require a large buffer when the lane is long, variable, and tied to a near-zero stockout tolerance.

Decision Hint

Use this type of policy only when downtime or contract penalties cost materially more than the carrying cost.

Boundary Conditions

Demand history and lead time must use the same time unit. Daily demand with weekly lead time will distort the buffer.
This calculator assumes uncertainty can be summarized with averages and standard deviations. If you only have maximum and average history, use a max-average fallback separately.
Promotions, stockouts, and one-time disruptions can make standard deviation look worse than the true operating pattern, so cleanse history before using the result.
Cycle service level is a planning target, not a guarantee that every order will ship immediately or that every backorder risk is removed.
Zero measured variability produces zero statistical safety stock, but MOQ rules, pallet rounding, and contract minimums may still force a practical floor.
A larger buffer can hurt turns and tie up cash, so review the holding-cost trade-off with the Inventory Turnover Calculator before applying the same policy across a whole category.

Sources & References

Frequently Asked Questions

How does this safety stock calculator work?
This page uses the service-level method: safety stock = z x sqrt(L x sigma_d^2 + d^2 x sigma_L^2). The result combines demand variability, lead-time variability, and the cycle service level target. It then adds that buffer to average demand during lead time to produce the reorder point.
Why does the calculator ask for lead-time standard deviation?
Lead-time standard deviation measures how much arrival timing moves around the average. If supply timing is volatile, that variability can drive more safety stock than demand noise alone. Enter zero when your replenishment lane is stable and you want the demand-only version of the formula.
What if I only know maximum and average demand values?
You can still use a rough average-minus-max style method, but that is a different formula from the statistical service-level model on this page. Use this calculator when you have average demand, standard deviation, and a service target. Use the max-average fallback when your data is limited and treat the result as a rough planning estimate.
What service level should I choose?
Many teams use about 90% to 95% for lower-priority items, 97% to 99% for important items, and 99.5% only when stockouts are especially expensive or unacceptable. The right answer depends on stockout cost, margin, substitutability, contract commitments, and how much carrying cost the business can absorb.
How do I calculate demand standard deviation?
Use historical demand from the same time bucket as the lead-time assumption, such as daily demand with lead time in days. Clean out one-time promotion spikes, stockout periods, or unusual disruptions before calculating standard deviation. If the SKU is seasonal, calculate the metric by season or planning window instead of mixing the whole year together.
What is the difference between safety stock and reorder point?
Safety stock is the extra buffer. Reorder point is the inventory level where a new order should be released. Reorder point equals average demand during lead time plus safety stock, so the trigger includes both expected consumption and the statistical buffer.
Can safety stock be zero?
Yes. If both demand variability and lead-time variability are zero, the statistical buffer becomes zero. In practice, teams may still hold operational minimums because of order multiples, supplier minimums, or service promises that are not captured by a pure variability model.
How often should I review the inputs?
Review them whenever supplier performance changes, demand mix changes, or service targets move. Many teams refresh the assumptions monthly for fast-moving SKUs and quarterly for slower items, then immediately revisit the model after promotions, lane changes, or persistent service misses.