When and how much to restock: reorder point and safety stock explained with numbers

Every out-of-stock product is a lost sale and, often, a customer who goes to the competition. Every surplus product is money sitting on a shelf. Between the two there is a simple way to decide when and how much to restock, and you do not need to be a logistics expert: four figures and a spreadsheet are enough.

The four figures you need

• Average daily demand (D): units you sell per day, on average.
• Lead time (L): days from placing the order until you can sell the product. It includes the supplier's preparation time, transport and receipt.
• Safety stock (SS): a cushion for when you sell more than usual or the supplier takes longer.
• Available stock: what you have plus what is already ordered and has not arrived yet.

The reorder point

It is the stock level at which you have to place the order. The formula is:

The idea: what you have left has to last the whole lead time, plus a margin in case things go worse than expected.

Two ways to calculate safety stock

Simple method (worst case)

SS = (maximum daily sales × maximum lead time) − (average daily sales × average lead time).

Example: a product sells 6 units a day on average and 9 on its best days. The supplier takes 10 days on average and, occasionally, 14.

• SS = 9 × 14 − 6 × 10 = 126 − 60 = 66 units.
• Reorder point = 6 × 10 + 66 = 126 units.

It is easy to understand and almost never leaves you out of stock, but it is usually excessive: it assumes that the worst sales day coincides with the worst lead time.

Statistical method (tighter)

SS = Z × σ × √L, where σ is the standard deviation of daily sales and Z depends on the service level you want: 1.28 for 90%, 1.65 for 95%, 2.33 for 99%.

With the same product, if daily sales vary with a standard deviation of 2 units and you want a 95% service level:

• SS = 1.65 × 2 × √10 = 1.65 × 2 × 3.16 ≈ 11 units.
• Reorder point = 60 + 11 = 71 units.

The difference from the simple method is 55 idle units. Multiplied across the whole catalogue, that is a lot of money. In a spreadsheet, σ is obtained with the STDEV function on daily sales.

How much to order

A practical way is to set a target cover: how many days you want each order to last (for example, 30).

Continuing with the example (statistical method), if you have 70 units left and nothing on order: 6 × 30 + 11 − 70 = 121 units. Then adjust to the supplier's minimum order or to full cases.

The mistake that skews almost every forecast

Daily demand seems easy to calculate: the month's sales divided by 30. But if the product was out of stock for part of the month, that calculation falls short, and leads you to order less than you need, so it runs out again. It is a vicious circle.

The same happens with new products: if an item has been active for ten days, dividing its sales by 30 underestimates its demand by two thirds. Demand must be calculated over the days the product was active and in stock.

Not all products deserve the same attention

Classify the catalogue with an ABC analysis:

• A: the 20% of products that usually make around 80% of turnover. Review them every week and with a high service level (95-99%).
• B: the next ones. Fortnightly review, 90-95% service level.
• C: the long tail. Monthly review, little stock or even made-to-order sales using the supplier's stock and a longer delivery time stated on the product page.

Cases the formula does not solve on its own

• Seasonality: use the demand for the same period last year, not last month's.
• Promotions: if you are going to run a campaign, add it to the forecast by hand.
• Products with an expiry date: the maximum stock is set by the shelf life, not just the demand. There is no point buying cheap if part of it is going to expire.
• Minimum orders and shipping: sometimes it pays to bring an order forward to reach free shipping; calculate whether the saving outweighs the cost of holding more stock.

To start tomorrow

1. Export the daily sales of the last 90 days for your 50 best-selling products.
2. Exclude the days without stock when calculating the average.
3. Record each supplier's real lead time (order date and receipt date of the latest shipments).
4. Calculate the reorder point and quantity with the formulas above.
5. Check every week which products have crossed their reorder point.

When the catalogue grows, the spreadsheet falls short: the sensible thing is for the store itself to calculate the need for each product from its sales and its days in stock, and for the supplier order to come from there.