Safety Stock Calculator — Multi-SKU, from CSV

Safety stock is the buffer that absorbs demand swings and late deliveries. Set the service level you want to protect, and we size that buffer item by item.

Calculated locally in your browser. No server uploads.

Safety Stock is calculated using demand history from the Forecast data, lead time and reliability from the Supplier Performance data, and the service level you set below. Stock on hand from the Operations data places each item in a coverage band.

Demand counted as real: Actual, Promotion (set on the Forecast page). Delivery rules reused from Supplier Performance: 3 buffer days, early deliveries accepted, over-delivery beyond 2% excluded.

Your inputs

Service level % per ABC class

The share of demand you want to serve from stock. Higher service level means a bigger buffer — protect your A items hardest.

Below this number of usable deliveries we fall back to the supplier average, then to your manual default.

Lead-time variability

With variability on we also buffer against suppliers arriving late by different amounts. Switch it off to treat lead time as a fixed number of days.

Used for items with no usable PO history at all — these are flagged in the results.

Review period

The buffer then covers the lead time plus that cycle.

days between order reviews
Rounding

Methodology & Guide

The safety stock formula, a worked example and answers to common questions.

Formula breakdown

When reliable lead-time history exists (combined demand and lead-time variability):

SS = Z × √( d² × σ_LT² + LT × σ_d² )

When lead-time variability is unreliable or switched off (fixed lead time):

SS = Z × σ_d × √LT
SS
Safety stock in units, rounded up to whole units (or to the pack size when set).
Z
Safety factor from the standard normal distribution for the target service level per ABC class (90% → 1.28, 95% → 1.65, 98% → 2.05).
d
Average daily demand = average monthly demand over 24 months ÷ 30.
σ_d
Daily demand standard deviation = monthly standard deviation ÷ 30.
LT
Average lead time in days (order date → delivery date). With periodic review, LT + review days.
σ_LT
Standard deviation of historical lead times, in days.

Worked example: a Class A spare part

A Class A item sells on average 600 units a month with a monthly standard deviation of 150. Its supplier's closed PO lines show an average lead time of 25 days with a standard deviation of 5 days. Class A targets a 98% service level.

  1. Daily demand d = 600 ÷ 30 = 20 units; daily volatility σ_d = 150 ÷ 30 = 5 units.
  2. 98% service level → Z ≈ 2.05.
  3. Lead-time term: d² × σ_LT² = 20² × 5² = 10,000. Demand term: LT × σ_d² = 25 × 5² = 625.
  4. SS = 2.05 × √(10,000 + 625) = 2.05 × 103.1 ≈ 211.3 → rounded up to 212 units.
  5. For comparison, ignoring lead-time variability: SS = 2.05 × 5 × √25 ≈ 51.3 → 52 units.

Safety stock = 212 units. Late and uneven deliveries, not demand swings, drive most of this buffer.

Quick FAQs

What service level should I use for each ABC class?

The defaults are A 98%, B 95% and C 90%: protect the items that carry most of your sales value hardest, and accept a little more risk on the long tail. Every step up in service level raises Z and the buffer, so 99%+ is usually reserved for truly critical items.

Why does an item show 'insufficient data' or use a fallback lead time?

Lead-time variability needs enough closed PO lines (default minimum 3). Below that, the tool uses the supplier's average lead time, then your manual default, and sets σ_LT to 0 — so the simpler fixed-lead-time formula applies. Items without demand history are listed with the reason.

Why is my safety stock higher than with the textbook Z × σ × √LT formula?

The textbook version assumes suppliers always deliver in exactly the same number of days. The combined formula also prices in delivery variability (d² × σ_LT²), which is often the larger risk for items with long or erratic lead times.

Full derivations, data safeguards and the privacy architecture are on the Methodology & Formulas page.