What Is Safety Stock in a Supply Chain?

What Safety Stock Actually Is

Safety stock is the inventory you hold above what you expect to sell before your next shipment arrives. It exists to absorb the gap between what you planned for and what actually happens — a supplier that ships four days late, a demand spike from a promotion, a forecast that was off by 15%. Without it, any one of those turns into a stockout.

It's easy to confuse with cycle stock, which is the inventory you hold specifically to cover normal, expected demand between reorders. If you sell 40 units a day and reorder every two weeks, your cycle stock is roughly 40 × 14 = 560 units. Safety stock sits on top of that number — it's the buffer for when demand or lead time doesn't behave as planned.

The two get added together at the reorder point: the inventory level that triggers a new order. Get safety stock wrong and the reorder point is wrong, and you either stock out or tie up cash in units that sit on a shelf for months.

The Formula, Worked on Real Numbers

There's no single safety stock formula. The right one depends on whether your risk is mostly in demand variability, lead time variability, or both. The version used most often in practice is Greasley's formula (also called the standard deviation method), because it accounts for both:

• Safety stock = Z × σLT × D avg

Where Z is the service level you're targeting, expressed as a Z-score; σLT is the standard deviation of your lead time in days; and D avg is your average daily demand.

Worked example: a SKU sells 40 units a day on average. Lead time is usually 14 days but has run anywhere from 11 to 17 days over the last six months, giving a standard deviation of 3 days. You're targeting a 95% service level, which is a Z-score of 1.65.

Safety stock = 1.65 × 3 × 40 = 198 units, rounded to 200.

Your reorder point is then cycle stock plus safety stock: (40 × 14) + 200 = 760 units. That's the level at which you need to have already placed the next order — not the level at which you start thinking about it.

Why One Safety Stock Number for Every SKU Is a Mistake

The formula above assumes you pick one service level and apply it everywhere. Most businesses that do this over-stock the SKUs that matter least and under-stock the ones that matter most. ABC analysis fixes that by ranking SKUs on revenue or margin contribution and setting a different target service level for each tier.

The C-tier decision is the one people find hardest to accept: deliberately choosing a lower service level, and living with the occasional stockout, because insuring every slow-moving SKU against every possible disruption is a worse use of cash than the stockout itself.

The Mistakes That Actually Cause Stockouts

Across the more than $500M in managed revenue Full Circle has managed across 100+ brands, the single most common cause of a bad safety stock number isn't the formula — it's the inputs. Two show up constantly:

• Using average lead time instead of the standard deviation of lead time. A supplier who is usually on time but occasionally ships three weeks late will blow through safety stock calculated on the average, because the average hides the outlier that actually causes the stockout.
• Recalculating once and never again. Lead times drift — a factory adds a production line, a freight lane gets congested, a supplier changes minimum order quantities. A number that was right in January can be wrong by June with nothing about your sales having changed.

We've made the second mistake ourselves: sizing safety stock off a supplier's quoted lead time rather than their actual delivered lead time, and only catching the gap after a shipment ran two weeks late. The fix isn't a smarter formula. It's using received-date history, not quoted lead time, as the input.

The Amazon-specific version of this mistake is treating a stockout as a lost-sales problem only. On Amazon, a stockout also drops your organic rank, and rank doesn't fully recover the moment stock lands back on the shelf — a category like Epic Gardening's, with a hard seasonal peak, feels this especially, since a stockout in the peak week costs far more than the same stockout in a slow month. So the real cost of under-provisioning safety stock is higher than the unit margin on the units you didn't sell.

What to Do When the Number Is Wrong

Two failure modes, two different fixes.

You stocked out despite having safety stock. Pull the last six to twelve months of actual receiving dates for that SKU and recalculate σLT from real data, not the vendor's quoted lead time. If the standard deviation is bigger than what you used, that's the whole answer — the formula was fine, the input wasn't.

Safety stock keeps growing and cash is stuck in it. Check whether the SKU has drifted into a lower ABC tier — declining sales velocity should pull the target service level down, and most businesses never revisit the tier once it's set. If the tier is right and the number is still high, the lead time itself may have gotten longer or more erratic, which is a sourcing conversation, not an inventory-math one.

Either way, don't fix a bad safety stock number by nudging the multiplier up or down until the stockouts stop. That masks the real input error and usually creates the opposite problem a year later.

What Calculates This for You — and What Still Needs a Person

ERP systems and dedicated inventory management software will run these formulas automatically once lead time and demand data are flowing into them, and for a business running the whole operation — purchasing, warehousing, financials — through one system, that's the right place for safety stock to live. IoT tools like RFID tags and smart shelves tighten the demand-side input further by giving near-real-time stock counts instead of end-of-day snapshots.

None of that replaces judgment on the inputs. Software will faithfully calculate a wrong number from a wrong lead-time input forever. Someone still has to check that the lead time being fed into the formula is the one the supplier is actually delivering, not the one written on the purchase order.