Every ecommerce brand has lived both versions of the same problem: the best seller that sold out for three weeks, and the pallet of a slow SKU that has been taking up space for a year. Both come from the same place. Someone decided when to reorder by feel. This guide shows how to replace the feel with two numbers per SKU, a reorder point and a safety stock, and how to keep them honest.
Reorder point = (average daily demand x supplier lead time in days) + safety stock. Safety stock = Z x standard deviation of daily demand x the square root of lead time, where Z is about 1.65 for a 95 percent service level. Use real lead times from supplier to your warehouse dock, count stock already on order, and set a different service level for each class of SKU.
The Two Numbers That Replace Guessing
Lead-time demand is how many units you expect to sell while a purchase order is in transit. Safety stock is the cushion for the times demand or delivery doesn't go as planned. Add them together and you have the point at which a new order must be placed.
Units sold per day over a representative recent window, excluding stockout days and one-off promotions.
Days from placing the PO to units being received and available to ship, not the supplier's quoted production time.
The standard deviation of daily (or weekly) sales. A spreadsheet computes it in one formula.
The share of reorder cycles you want to get through without a stockout. It is a business decision, not a math result.
A Worked Example
Take one SKU that sells an average of 40 units a day, with a daily standard deviation of 12 units. Your supplier takes 21 days to deliver to the warehouse. You want a 95 percent service level, which corresponds to Z = 1.65.
When available inventory, meaning on hand plus anything already on order, falls to 931 units, you place the next PO. This example is illustrative; plug in your own numbers.
Count inventory on order when you compare against the reorder point. Looking only at what is on the shelf triggers duplicate orders while the first one is still on the water.
What Service Level Really Costs
Each step up in service level needs a bigger Z, and the buffer grows faster than the protection. Here are standard values:
In the example above, 90 percent needs about 70 units of safety stock, 95 percent needs 91, and 99 percent needs about 128. Going from 95 to 99 percent adds roughly 40 percent more buffer for four more points of protection. Spend that on the SKUs where a stockout hurts most, such as top sellers, subscription items, and anything with a retailer penalty, not on the whole catalog.
When Lead Time Is the Variable
For most brands, especially those importing, supplier delivery time swings more than demand does. If lead time varies, the demand-only formula underprotects you. Use this version instead: safety stock = Z x the square root of (average lead time x demand variance + average demand squared x lead time variance).
Using the same SKU with a lead time standard deviation of 5 days, the buffer is 1.65 x the square root of (21 x 144 + 1,600 x 25) which comes to about 342 units, nearly four times the demand-only figure. That is why importers who "never run out until suddenly they do" often have a lead-time problem, not a forecast problem. Track actual days from PO to dock for each supplier and use that history.
Setting It Up in a Spreadsheet
-
Pull 8 to 12 weeks of sales per SKU
Use daily or weekly units shipped, and remove days where you were out of stock, since those understate real demand.
-
Class your SKUs
Sort by revenue or margin. Give the top group a 97 to 98 percent target, the middle group about 95, and the tail 90 or less.
-
Record real lead times
For each supplier, log the last several POs from order date to receiving date and calculate the average and spread.
-
Calculate and add a column for available inventory
Compute lead-time demand, safety stock, and reorder point, then compare against on-hand plus on-order to flag what to buy today.
-
Review on a schedule
Monthly for fast movers, quarterly for the rest, and any time a supplier or sales channel changes.
Why Good Math Still Fails
The formula is the easy part. These are the usual reasons a clean reorder point still produces stockouts:
- Wrong on-hand numbers. A reorder point compares against inventory the system believes exists. If counts drift, so does everything built on them. A regular cycle counting program fixes the input.
- Quoted lead times. Suppliers quote production time. You need dock-to-available time, including transit, customs, and receiving.
- Forgetting inbound stock. Units on order count toward availability.
- One setting for every SKU. Uniform buffers overstock the tail and still starve the leaders.
- Overbuffering as a reflex. After a stockout, the instinct is to double the safety stock. That is how brands end up with dead stock and a storage bill to match.
- Ignoring promotions and seasons. A trailing average misses the spike. Plan peaks as their own purchase events, as in our peak-season fulfillment guide.
Where a 3PL Helps and Where It Doesn't
A 3PL won't decide what you buy, and you wouldn't want it to. What it can give you is the data the formula depends on: accurate on-hand counts, advance notice of inbound receipts, receiving turnaround time, and sales velocity by SKU. Shorter and more predictable receiving also shrinks the lead-time piece of your buffer. Ask any warehouse you evaluate how fast they receive, how they report inventory accuracy, and whether they can alert you at a threshold you set.
Frequently Asked Questions
The reorder point is the on-hand-plus-on-order inventory level at which you place a new purchase order. It equals the demand you expect during supplier lead time plus safety stock. When available inventory drops to that number, you reorder so the new stock arrives just as the buffer is about to be used.
Safety stock is the extra inventory you hold to cover demand that runs higher than forecast or supplier deliveries that arrive late. Without it, any ordinary variation in sales or shipping time produces a stockout. With too much of it, you tie up cash and pay storage on units that sit.
A common formula is safety stock = Z x standard deviation of daily demand x the square root of lead time in days, where Z reflects your target service level (about 1.65 for 95 percent). If lead time also varies, use Z x the square root of (lead time x demand variance + average demand squared x lead time variance).
Most ecommerce sellers land between 90 and 98 percent per SKU, and the target should differ by SKU. Best sellers and items that drive repeat purchases deserve a higher level. Slow movers and seasonal tail items can run lower. Each step up costs more inventory than the last, so higher is not automatically better.
Review them monthly for fast-moving SKUs and quarterly for slower ones. Recalculate right away when a supplier changes lead time, when a promotion or channel launch changes demand, or after a stockout that the model should have prevented.
Use a demand forecast for the coming period, not the trailing average. A reorder point built on last quarter's sales will under-protect you going into peak and over-protect you coming out. Plan seasonal buys as separate events, as covered in our peak-season fulfillment guide.
Use the history you have, a comparable SKU, or a conservative estimate, and set a higher safety stock for the first few cycles. Tighten it as real demand data accumulates. For a brand-new product, a smaller first order with a second early reorder is usually safer than one large bet.
Most 3PLs do not set purchasing decisions, but a good one gives you accurate on-hand counts, inbound receiving dates, and sales velocity reports so you can. Some can also send low-stock alerts at a threshold you set. The reorder decision generally stays with the brand.
The usual causes are inaccurate on-hand counts, lead times measured from the supplier's quote instead of from actual receipt at the warehouse, ignoring stock already on order, and treating all SKUs the same. Fixing count accuracy and measuring real lead times solves most of it.
Holding cost includes storage fees, the cost of the cash tied up in inventory, and the risk of the stock becoming obsolete. Many operators estimate annual carrying cost as a meaningful percentage of inventory value, so a buffer that is twice as large as it needs to be quietly costs real money every month.
Running out of stock, or sitting on too much?
Tell Matt about your SKUs and suppliers. Simple Distribution will give you an honest read on whether tighter inventory planning or a better warehouse partner is the bigger fix.