A shopper taps a product, sees “sold out,” and buys something similar somewhere else. Nothing was wrong with the product or the price. The restock was due on Thursday, and the shelf ran empty on Monday.
Safety stock is the inventory set aside to cover days like those. The tricky part is sizing it, because every extra unit protects a sale but also ties up cash on a shelf until it is needed. Here is how to find the right amount, in plain numbers.
TL;DR
- Safety stock is inventory held beyond what you expect to sell before the next delivery.
- Two things use it up: customers buying faster than planned, and restocks arriving later than promised.
- Late deliveries are often the bigger risk, so a safety stock calculation that only looks at sales can leave you badly short.
- The most complete safety stock equation covers both risks using average sales, average lead time, and how much each one swings.
- Protection should vary by SKU. Best sellers deserve more cover than slow or easily swapped products, and every part of a kit needs its own buffer.
What Is Safety Stock?
A quick safety stock definition: the extra inventory you keep on top of the stock you expect to sell while waiting for your next delivery. If everything goes to plan, you never touch it. It only gets used when demand runs ahead of the forecast or a shipment shows up late.
That sets it apart from cycle stock, the inventory you buy to sell as normal. The two meet at the reorder point, the stock level that tells you it is time to place the next order:
Reorder point = (average daily sales × lead time in days) + safety stock
So what is safety stock inventory in money terms? It is stock you pay for, store, and hope to use rarely. On a 3PL storage and fulfillment invoice, it shows up every month as storage, which is why the goal is the smallest buffer that still protects the sale.
What Safety Stock Inventory Protects You From
Customers Buy Faster Than Planned
A good forecast gets the average right, but none gets every day right. A product can sell 40 units on Tuesday and 65 on Wednesday for no clear reason. Safety stock absorbs that everyday noise.
The Restock Arrives Late
The other half of the risk sits with suppliers and freight. A production run slips or a container waits at port, and every extra day is a day of sales with nothing new arriving. Delays are common right now: in the New York Fed’s September 2026 survey of manufacturers in New York State, 25.7% of firms said deliveries were taking longer than a month earlier, while only 6.9% saw them speed up.
How Do You Calculate Safety Stock? Build The Equation In Three Steps
Take a scented candle that sells an average of 40 units a day. On a typical day, sales move up or down by about 12 units (that 12 is the standard deviation of daily demand). The supplier quotes 30 days, but real deliveries have varied by about 6 days either way. The brand wants 95% protection, meaning it accepts running short in about one restock cycle out of 20.
That 95% becomes a Z score, the multiplier that sets how much protection you buy: 1.65. For other levels, 90% is 1.28, 98% is 2.05, and 99% is 2.33.
Step 1: Cover The Sales Swings
Safety stock = Z × standard deviation of daily demand × √lead time
1.65 × 12 × √30 = about 108 candles.
This works when your supplier delivers on schedule every time. The candle’s supplier does not, and 108 candles cover less than three days of delay.
Step 2: Add The Late Deliveries
If only deliveries moved, the formula would be:
Safety stock = Z × average daily demand × standard deviation of lead time
1.65 × 40 × 6 = 396 candles.
Most products have both problems. When sales and deliveries vary for unrelated reasons, the Association for Supply Chain Management combines them in one safety stock equation, with every time input in the same unit (days, here):
Safety stock = Z × √(lead time × demand standard deviation² + average daily demand² × lead time standard deviation²)
1.65 × √(30 × 144 + 1,600 × 36) = 1.65 × √61,920 = about 411 candles.
Look inside that square root. The sales part is 4,320 and the delivery part is 57,600, more than 13 times larger, so nearly all of this candle’s risk comes from the supplier. A single week of delay means 280 candles of demand with nothing on the way, which is why Step 1 falls so short. With 411 as the buffer, the reorder point is (40 × 30) + 411 = 1,611 units.
Step 3: Check It Against The Quick Shortcut
Many brands skip the statistics and use the max and average method:
Safety stock = (maximum daily sales × maximum lead time) minus (average daily sales × average lead time)
If the candle’s busiest day was 70 units and its slowest delivery took 42 days, that is 2,940 minus 1,200 = 1,740 candles. The shortcut assumes your best sales day and your worst delivery always land together, so it ties up more than four times the stock the full formula calls for. Fine for a rough first number, but not for running safety stock calculations across a catalog.
Choosing How Much Protection Each SKU Gets
The Z score is a business decision dressed up as a number. Each step up costs more stock, so the right level depends on what running out costs for that SKU, and that often goes beyond the missed order. Penn State researchers studying more than 290,000 grocery products on Amazon’s marketplaces in the US, Canada, and three European countries found that when products were unavailable over a 90-day period, average sales rankings were 14% to 67% higher, a sign of weaker sales. On a marketplace, a stockout keeps costing you after the shelf is full again.
A simple way to sort your catalog:
- Hero SKUs that carry your ad spend, your marketplace ranking, or a retail account: 98% or higher.
- Steady core products with repeat buyers: around 95%.
- Slow or easily swapped variants, such as a scent shoppers will happily trade for another: 90% may be enough.
For the candle, moving from 95% to 98% raises the buffer from about 411 to about 510 units: worth it on the best-selling scent, wasted on a seasonal one.
Where The Standard Safety Stock Calculation Needs Help
Kits Run Out At Their Scarcest Part
A gift set, a subscription box, or a concert VIP package can only ship as many units as its least stocked component allows. Five hundred tees and 40 lanyards make 40 kits. So a kit’s safety stock is set item by item, each with its own lead time: a custom enamel pin from a slow overseas supplier needs far more cover than a blank tote bought locally. In VIP fan packages, where a tee, a laminate, and a lanyard ship as one unit, the small pieces deserve the same attention as the tee.
A Viral Spike Is Not A Bad Week
Every formula above assumes demand moves within a normal range. A product that takes off on social media can sell several times its usual volume in a day, far outside what a Z score covers, and stocking for that just in case means months of inventory for a moment that may never come. For SKUs that depend on TikTok Shop fulfillment or creator campaigns, keep a normal buffer and plan the spike separately: production capacity agreed with your supplier in advance, and a stock level that triggers it.
Common Mistakes When Setting Safety Stock Levels
- One rule for the whole catalog. “Two weeks of cover on everything” overprotects steady sellers with reliable suppliers and underprotects products that travel a long way.
- Calculating by product instead of by SKU. A hoodie sold in five sizes needs five buffers, not one, and the smaller-selling sizes swing more relative to their volume.
- Treating the buffer as stock to sell. Fill a wholesale order or a promotion from safety stock without reordering the difference, and the protection is gone before the surprise arrives. Planned demand belongs in the order quantity.
- Setting it once. A new supplier, a new channel, or a year of growth changes every input, so the number has to change too.
Focused Catalogs, Watched Closely
Merchdrop is not the right warehouse for every brand. A catalog of thousands of SKUs that each sell a handful a week needs a different kind of operation. We fit the opposite profile: a focused range of beauty, personal care, wellness, home goods, or merch products moving in volume, where one hero SKU running dry costs revenue every day it stays empty.
Our own team and floor are set up for exactly that catalog, with per-SKU stock counts you can check at any hour, so the buffer you set is one you can see. If that sounds like your business, get in touch with our team.
Ready To Find Out If Merchdrop Is A Good Fit?
The fastest way to find out whether we are a good match for your needs and growth plans is a short conversation, not a long form. Send us your numbers and we will come back with a quote within 24 hours.
FAQ
What is safety stock and how is it calculated in practice?
It is extra inventory held to cover sales that run higher than expected or a delivery that arrives late. The most complete method is the combined formula: Z score × √(lead time × demand standard deviation² + average daily demand² × lead time standard deviation²), with inputs taken from recent daily sales and the dates of past deliveries.
Is safety stock the same as buffer stock?
The two terms are often used interchangeably. Some planners use buffer stock more broadly, for any extra inventory built up ahead of a known risk such as a factory shutdown, and keep safety stock for the everyday cover against ordinary swings in sales and lead time.
How do I find safety stock if I have no lead time records?
Start collecting them now. For every purchase order, log the date it was placed and the date the stock was ready to sell. Until you have a handful of deliveries, use a cautious estimate of how much lead time varies, such as a week either way, and replace it with real figures as they come in.
How often should safety stock be recalculated?
Every quarter works for most steady products. Recalculate sooner after a supplier change, a new sales channel, a price change, or any month where deliveries took noticeably longer or shorter than usual.
Who sets safety stock levels, the brand or the 3PL?
Usually the brand, because it owns the purchasing decisions and the supplier relationships. The 3PL holds the stock, keeps the counts accurate, and should supply the inputs: units sold per day by SKU, the dates deliveries were received, and alerts when stock falls toward the reorder point.
