Safety Stock and Reorder Point, Explained for Small Brands
Safety stock is the extra inventory you hold to absorb the difference between what you expect and what happens. Reorder point is the stock level that triggers a new purchase order, calculated as the units you will sell during the supplier lead time plus that safety stock. Both are arithmetic, both take about ten minutes to calculate, and the formula you choose changes the answer by thousands of dollars.
For a small brand the practical question is not whether to hold safety stock. It is which formula to use, because the common one is far more conservative than most people realize.
The two definitions
Safety stock is a buffer against two kinds of uncertainty: demand moving faster than forecast, and the supplier arriving later than promised. Hold too little and you stock out. Hold too much and you convert cash into cardboard.
Reorder point is the trigger level. When on hand quantity drops to it, you order. It answers a timing question, not a quantity question. How much to order is a separate calculation.
The relationship is straightforward:
Reorder point = (average daily sales × average lead time in days) + safety stock
A worked example
One SKU. Twelve months of history gives these inputs:
- Average daily sales: 42 units
- Standard deviation of daily sales: 14 units
- Highest daily sales observed: 78 units
- Average supplier lead time: 32 days
- Standard deviation of lead time: 6 days
- Longest lead time observed: 48 days
- Landed unit cost: $12.74
Demand during an average lead time is 42 times 32, or 1,344 units. That figure is the base of every version below.
Method 1: the maximum formula
This is the version most commonly published, and it is the most conservative.
Safety stock = (max daily sales × max lead time) – (average daily sales × average lead time)
Safety stock = (78 × 48) – (42 × 32) = 3,744 – 1,344 = 2,400 units
Reorder point = 1,344 + 2,400 = 3,744 units
Read what this actually assumes: your single worst sales day repeating every day, for the entire duration of your single worst lead time, at the same time. That compound event is extraordinarily unlikely, and you are paying to insure against it. At $12.74 landed cost, that safety stock represents $30,576 sitting in a warehouse.
Method 2: demand variability only
The statistical approach sizes the buffer to a chosen service level rather than to the worst case.
Safety stock = Z × standard deviation of daily demand × √(lead time)
Z is the service factor for your target service level: 1.28 for 90 percent, 1.65 for 95 percent, 2.05 for 98 percent, 2.33 for 99 percent.
At 95 percent: 1.65 × 14 × √32 = 1.65 × 14 × 5.66 = 131 units
Reorder point = 1,344 + 131 = 1,475 units
This one is usually too thin, because it ignores the supplier entirely. For a brand importing from overseas, lead time is typically the larger source of variability.
Method 3: both variabilities
This is the version worth using.
Safety stock = Z × √[(lead time × demand variance) + (average daily demand² × lead time variance)]
Working it through at a 95 percent service level:
- Lead time × demand variance = 32 × 14² = 32 × 196 = 6,272
- Average daily demand² × lead time variance = 42² × 6² = 1,764 × 36 = 63,504
- Sum = 69,776, and the square root is 264
- Safety stock = 1.65 × 264 = 436 units
Reorder point = 1,344 + 436 = 1,780 units
Notice which term dominates. Lead time variability contributes 63,504 of the 69,776 total, more than ten times what demand variability contributes. For this SKU, the supplier is the risk. Chasing a better demand forecast would be effort spent on the smaller half of the problem.
What the choice costs
| Method | Safety stock | Reorder point | Cash tied up at $12.74 |
|---|---|---|---|
| Maximum formula | 2,400 units | 3,744 | $30,576 |
| Demand variability only | 131 units | 1,475 | $1,669 |
| Both variabilities, 95% | 436 units | 1,780 | $5,555 |
The gap between the maximum formula and the full statistical version is 1,964 units, or roughly $25,000 of cash, on one SKU. Run a catalog of forty SKUs on the maximum formula and the difference is the reason the business feels poor while the profit and loss statement looks fine.
Service level is a cost decision
Moving from 95 percent to 99 percent on the same SKU changes Z from 1.65 to 2.33, raising safety stock from 436 to about 615 units. That is 179 extra units, roughly $2,280, to cut expected stockouts from five percent of cycles to one percent.
Whether that trade is worth it depends on the product. For an item with strong repeat purchase and easy substitution, a stockout costs a sale. For a hero SKU whose organic ranking depends on continuous availability, a stockout can cost months of recovery, and 99 percent is cheap. Set service level per SKU, not per catalog.
Four mistakes that make the output wrong
Using the quoted lead time instead of the actual one. Suppliers quote 30 days and deliver in 38. Use your own receiving records, and include customs and inland freight, not just the factory ship date.
Averaging across a seasonal year. A SKU selling 42 units a day annually might sell 20 in March and 140 in November. Calculate seasonally, or the buffer is simultaneously too big in spring and too small in Q4.
Forgetting that reorder point is measured in sellable stock. Units in transit, units at a preparation center, and units in an inbound shipment awaiting check in are not available to fulfill orders. Count what can actually ship today.
Setting it once. Both inputs drift. Recalculate quarterly, and immediately after any supplier change, since lead time variance resets entirely with a new factory.
Where the numbers come from
All six inputs live in systems you already have. Daily sales history and its variability come from marketplace reports. Lead time history comes from purchase order and receiving records, which is the piece most small brands do not keep in a form that can be averaged, and the single highest value record to start keeping today. Landed unit cost comes from your accounting system.
Pulling those together is what inventory functionality inside ecommerce accounting platforms is for. ConnectStock, the inventory layer inside ConnectBooks, tracks stock in real time against the same ledger that holds unit cost, and tools such as Finaloop, Webgility, and Entriwise take different approaches to the same problem. A spreadsheet works too, as long as someone updates the lead time log every time a container lands.
Start with one SKU. Calculate all three methods, look at the cash difference, and decide deliberately which risk you are buying down. That is a more useful exercise than adopting a formula because it appeared first in a search result.