Safety stock is the cushion of inventory you hold on top of what you expect to sell before the next delivery arrives. It exists for one reason: reality is noisier than a forecast. Suppliers run late, a post goes viral, a carrier loses a carton. The buffer absorbs those surprises so you keep shipping.

It concerns anyone holding stock they reorder, from a creator restocking a hoodie every six weeks to a boutique with 200 SKUs. It is also one of the few inventory decisions that is pure trade-off: too little and you lose sales, too much and you lock cash in boxes that do not move.

What is safety stock?

Safety stock is the quantity of a product you intend never to touch in normal conditions. Your regular stock covers average demand during the supplier's lead time. Safety stock covers the gap between average and worst realistic case, on either side: demand higher than expected, or lead time longer than expected.

Picture a product that sells 10 units a day with a 14-day resupply. Regular cycle stock of 140 units covers the lead time on paper. If sales jump to 14 a day for a week, or the delivery arrives 5 days late, 140 is not enough. The 50 to 80 extra units that cover those cases are your safety stock.

What safety stock is not:

  • It is not the reorder point. The reorder point is the stock level that triggers a new purchase order. It equals expected demand during lead time plus safety stock. Safety stock is one component of it.
  • It is not dead stock. Dead stock is inventory that has stopped selling. Safety stock is inventory that sells fine but that you deliberately keep in reserve.
  • It is not the same for every product. A best seller with an unreliable overseas supplier needs a large buffer. A slow seller from a local maker who delivers in 3 days needs almost none.

Vocabulary you will meet: cycle stock (the regular stock that turns over between orders), reorder point or ROP, service level (the probability of not running out, for example 95%), stockout, demand variability and lead time variability.

Why it matters

Safety stock is where two costs collide, and both are easy to underestimate.

The cost of too little: a candle brand sells a $34 candle at 8 units a day, with a $9 unit cost. Their supplier's lead time is 21 days but has been as long as 30. With no safety stock, an ordinary 9-day delay means 72 lost sales, about $2,450 in revenue and $1,800 in gross margin. Plus the customers who tried to buy and went elsewhere. That happens two or three times a year for this brand, so roughly $5,000 of margin lost annually on one SKU.

The cost of too much: the same brand, scared by a stockout, decides to hold 60 days of the candle "just in case," 480 units. That is $4,320 of cash sitting on a shelf for months, plus storage, plus the risk that a scent goes out of fashion. If the brand has 12 scents and treats each the same way, that is $50,000 tied up, which for a small business is usually the entire cash reserve.

The right buffer for the candle, computed below, is about 90 units, $810 of cash. It covers the realistic delay and leaves the money for marketing, new products and a good night's sleep.

How to calculate it

The simplest reliable formula for a small store uses worst-case versus average:

Safety stock = (max daily sales × max lead time) − (average daily sales × average lead time)

Step by step, for the candle above:

  • Average daily sales: 8 units. Take the last 90 days and divide by 90, excluding one-off spikes you do not expect to repeat.
  • Max daily sales: 14 units, the busiest normal day in that period (not the day you were featured on TV, unless that happens every quarter).
  • Average lead time: 21 days, from order to stock ready to ship, based on your last three or four purchase orders.
  • Max lead time: 30 days, the longest of those orders.
  • Worst case demand during worst lead time: 14 × 30 = 420.
  • Average demand during average lead time: 8 × 21 = 168.
  • Safety stock: 420 − 168 = 252 units.

That is a conservative result because it assumes the sales peak and the delivery delay happen at the same time for the whole period, which is rare. Many small sellers scale it down by half, giving about 125 units, or use a lighter version:

Safety stock = average daily sales × (max lead time − average lead time), which here gives 8 × 9 = 72 units, covering delay only. Add a small demand cushion, around 20 units, and you land near 90.

Then set your reorder point:

Reorder point = (average daily sales × average lead time) + safety stock = 168 + 90 = 258 units. When stock drops to 258, you order.

Recompute every quarter or whenever a supplier or a sales trend changes.

Benchmarks and examples

Rules of thumb that hold up for small stores:

  • Local supplier, 1 to 7 days lead time, steady sales: 3 to 7 days of sales as safety stock.
  • Domestic supplier, 2 to 4 weeks: 1 to 2 weeks of sales.
  • Overseas by sea, 8 to 16 weeks: 3 to 6 weeks of sales, and a second supplier if the product is critical.
  • Print-on-demand or dropshipping: zero, because you hold no stock; your risk sits with the supplier's own buffer.

By seller type:

  • A creator with 3 merch products. The print shop delivers in 12 days, sometimes 20. Keep about 8 days of sales on each product. For a hoodie selling 5 a day, that is 40 units.
  • A skincare brand with 10 SKUs. Two best sellers get a full buffer, computed with the formula. The other 8, selling under 1 unit a day, get a flat 15 units each. Not every SKU deserves a spreadsheet.
  • A boutique with 200 SKUs. Rank by revenue. The top 20% of SKUs, which usually make 70 to 80% of revenue, get a computed buffer. The rest get a simple "reorder when below 2 weeks of sales" rule.

Service level matters too. Covering 95% of situations costs far less stock than covering 99%. For most small brands, accepting a stockout once or twice a year on a product is the right call; the last 4 points of service level often double the buffer.

Common mistakes

  • One buffer for everything. A flat "20 units on every SKU" is too much for slow items and far too little for best sellers. Size by product.
  • Using the supplier's promised lead time. Use the lead time you actually experienced. The difference between the two is exactly the variability you are trying to cover.
  • Never recomputing. Sales double, the buffer stays the same, and the stockout returns. Review quarterly.
  • Counting safety stock as available. If your store shows the buffer as sellable, it is not a buffer. Either subtract it in your head or, better, track a reorder point so you reorder before touching it.
  • Confusing safety stock with over-ordering. Buying 6 months of a product because the unit price was lower at that volume is a purchasing decision with its own logic; see economic order quantity. It is not safety stock.

How to improve it

  • Measure your real lead times. Write the order date and the received date on every purchase order. After four orders you have the two numbers the formula needs.
  • Shorten the lead time instead of growing the buffer. A supplier at 10 days instead of 25 cuts the needed buffer by more than half. Sometimes paying 5% more per unit is cheaper than holding 100 extra units.
  • Split by ABC. A items (top sellers) get a computed buffer and a monthly check. B items get a rule of thumb. C items get a minimum quantity or none.
  • Watch the reorder point, not the shelf. Set an alert when a SKU drops below its reorder point. That single habit prevents most stockouts.
  • Keep a second supplier tested for A items. A backup with a short lead time is a form of safety stock that costs nothing until you use it.
  • Use backorders as a controlled release valve. When the buffer is gone and a delivery is confirmed, dated backorders keep revenue flowing without more inventory.
  • Review dead stock every quarter. Money freed from products that stopped selling is money you can put into buffers for the ones that do.

In Roctify

Roctify tracks stock per product and per variant in one shared catalog, and every sale on your link in bio or your online store decrements it immediately, so the number you see is the real one across channels. That real number is the input for your reorder point: compare it with the threshold you computed and place your purchase order when it dips below. Sales reports and exports (Pro plan) give you the units-per-day figure the formula starts from, per SKU and per period, so you can recompute the buffer each quarter without a separate tool.

Roctify does not compute safety stock or send purchase orders to suppliers. It gives you accurate stock and sales data, sells at 0% transaction fees on every plan, and lets you adjust quantities in seconds when a delivery lands.

FAQ

How much safety stock should a small store keep?

Enough to cover the difference between your supplier's average lead time and their worst realistic one, at your normal daily sales rate. For a product selling 8 a day from a supplier that is usually 21 days but sometimes 30, that is around 70 to 90 units. Adjust upward for best sellers and downward for slow items, and recompute every quarter.

Is safety stock the same as a reorder point?

No. The reorder point is the stock level at which you place a new order. It equals the units you expect to sell during the lead time plus your safety stock. Safety stock is the part of the reorder point that protects against surprises; the rest covers ordinary sales while you wait for the delivery.

Should every product have safety stock?

No. Products with steady sales and a critical role deserve a computed buffer. Slow sellers, products you are phasing out and anything shipped by a supplier who delivers in a couple of days can run with a minimal buffer or none. A useful rule is to spend your buffer cash on the 20% of SKUs that make most of your revenue.