A reorder point without spreadsheet theatre

Deciding when to reorder does not have to be a weekly guess or a complex model. For a replenishable product, one threshold is enough: how many units will sell before the next delivery arrives, plus safety stock to absorb swings in sales and supply. When stock falls to that level, you reorder. Your attention then goes to exceptions, not to a fresh forecast for every item.

Two parts of one threshold

MIT course materials on inventory management build the reorder point from two parts: expected demand over the lead time and safety stock. The professional body APICS (now ASCM) describes the same principle and explains safety stock as inventory that protects against stockouts caused by fluctuating demand, forecast error and variable lead times.

The first part is arithmetic. The second is a decision about risk, and this is where most mistakes happen: safety stock is set by gut feel or as 10–20% of regular stock. That is easy, but according to APICS it generally performs poorly, because two items with the same average sales can vary in entirely different ways.

Safety stock is not a one-off cost either. Average inventory consists of half the order quantity plus the whole safety stock, so that money sits in the warehouse permanently. That is why it is worth calculating rather than guessing.

What counts as lead time

Lead time is not the delivery promise in a supplier’s price list. For a reorder point it is the whole time from the moment stock drops below the threshold until the goods can be sold again: deciding and sending the order, supplier processing, transport and receiving.

The second common trap is how often you check. If you review stock once a week, an item can fall below the threshold the day after a review and wait a week before anyone notices. With periodic review, the protected period therefore extends by the review interval. If you order weekly, plan for lead time plus one week.

Third, do not compare the threshold only with units on the shelf. The decision should use stock on hand plus goods already ordered from the supplier, minus units that are sold but not yet shipped if your system has not already deducted them. Otherwise you will reorder goods that are already on their way.

An illustrative example, step by step

The following numbers are invented so that the calculation can be checked. A replenishable item sold an average of 10 units a week, with a standard deviation of 4 units, in the weeks when it was in stock. Out-of-stock weeks are excluded because they would understate demand. Two weeks pass from order to receiving, the purchase cost is CZK 400 excluding VAT, and a daily alert watches stock.

1. Sales during lead time: 10 × 2 = 20 units.
2. Variability over lead time: a weekly standard deviation is not multiplied across periods but scaled by the square root of the number of weeks: 4 × √2 ≈ 5.7 units.
3. Safety stock for 95%: 1.65 × 5.7 ≈ 9.3, rounded up to 10 units.
4. Reorder point: 20 + 10 = 30 units. Around CZK 4,000 sits permanently in stock as insurance.

Two changes of assumption show why the inputs matter:

• Weekly instead of daily review: the protected period is three weeks, sales 30 units, variability 4 × √3 ≈ 6.9, safety stock 12 and a reorder point of 42 units.
• An unreliable supplier: if lead time varies by half a week, a second source of uncertainty is added. For independent and roughly normally distributed variation, the two combine under a square root: √(2 × 4² + (10 × 0.5)²) ≈ 7.5; safety stock 13 and a reorder point of 33 units.

The example also shows where to look when safety stock is too expensive: more frequent review or a more reliable supplier can save more than lowering the target certainty.

What extra certainty costs

The factor 1.65 corresponds to a 95% cycle service level – the probability of not running out during one replenishment cycle. The relationship is non-linear, and every extra percentage point costs more than the last:

A 100% level is statistically unattainable. Instead of one level for the whole warehouse, choose higher certainty only for items where a stockout really hurts – because of margin, importance to customers or lack of a substitute – and lower certainty for easily replaceable goods.

Read the number carefully. 95% does not mean that only 5% of customer orders will find the item sold out. It means you expect a stockout in roughly one in twenty replenishment cycles, whatever its size. The share of demand actually met is a different metric and tends to be higher when sales are stable.

Where the calculation stops working

The formula assumes the item sells regularly and that its variation can be described by an average and a standard deviation. That assumption fails for three kinds of goods:

• Sparse sales. When weeks without sales dominate and then a larger order arrives, standard methods almost always produce inappropriate stock levels, and safety stock rules need adjusting. Specialised methods exist for such demand, but for a typical online store it is more honest to use a fixed minimum based on experience, or not to keep the item in stock.
• Unique and limited goods. Where the same item cannot be bought again, a reorder point has nothing to trigger.
• Unknown or new lead time. Without it, the result is a precise-looking number with no foundation. Measure two or three real deliveries first.

Treat seasons and promotions with the same care. Campaign weeks or seasonal peaks distort both the average and the deviation. Calculate the reorder point from regular weeks and adjust it deliberately before the season.

Set it once, then manage exceptions

The value of a reorder point lies not in precision but in no longer deciding about every item from scratch. Once it is set, check whether it behaves as expected: safety stock should be dipped into in roughly half of the cycles, and stockouts should not be more frequent than the chosen certainty implies. When results diverge, look for the cause first – a supplier change, a promotion, new competition – and only then change the numbers.

In Shoptet, the reorder point can be entered as a product’s minimum stock level. The stock overview then shows under- and overstock, can filter items below minimum stock and can create a supplier order export. Watch two things, though. According to the help page, understock is calculated against units in stock, so track goods already ordered but not yet received separately. And the export proposes a quantity equal to the understock, which only brings the item back to the threshold. The quantity you actually order should also cover sales until the next order.

What to do now

1. Pick five replenishable items that matter most. Leave out unique goods and items with sparse sales.
2. Calculate the weekly average and standard deviation of sales using only weeks when the item was in stock and not on special promotion.
3. Measure the real lead time from decision to receiving and add the interval at which you review stock.
4. Choose a certainty level based on the impact of a stockout and calculate the reorder point. Enter it as the minimum stock level.
5. Review exceptions once a month: items that ran out and items whose safety stock was never touched, and only then adjust the numbers.