A best-selling product earns money only on the days it can be bought. When it keeps running out, the store loses more than the units it failed to sell. Some customers leave with the rest of their basket too, and some do not come back. The decision to hold more stock therefore rests on a simple comparison: how much margin stock-outs take away, and what the stock that would prevent them would cost. Both can be estimated from data a typical online store already has.
A stock-out costs more than one unsold unit
The most thorough measurement of stock-out costs comes from a field experiment at a US catalogue retailer with around 22,000 customers. Items that were in stock when ordered ended up sold 86% of the time; out-of-stock items only 62%. The stock-out also reduced sales of other items in the same order, from 83.1% to 71.2%. This indirect loss made up about a third of the total short-run cost of a stock-out. And the effect persisted: customers for whom everything they ordered was out of stock had roughly 22% lower demand over the next thirteen months.
The authors themselves caution that the specific figures do not transfer to other firms and markets. The direction does: a stock-out takes away sales of the missing item, part of the basket and future purchases. That makes sense online. A customer who cannot find the main product often abandons the rest of the order rather than pay for shipping twice.
Stock-outs are not spread at random either. A review of more than fifty studies in grocery and household retail found higher stock-out rates for fast-moving and promoted items than for the range as a whole, and in one study the fastest 10% of items accounted for 45% of stock-outs. These are physical stores and older data, but the logic holds generally: what sells fastest runs out first.
Days out of stock are not days without demand
Sales data on stock-out days show zero, but that zero does not mean nobody wanted the product. Research on vending-machine demand showed that ignoring availability biases demand estimates, and that the true effect of stock-outs on profitability comes out considerably larger. For an online store this leads to a simple three-step approach.
1. Count the days out of stock
In Shoptet, the product’s stock history sits on its stock card and can be exported to CSV. It tells you how many of the last 90 days the best seller was at zero. If you use Shoptet’s back-in-stock alerts, their overview also shows how many customers asked to be notified when the product returns. That is a direct, if incomplete, signal of demand during the stock-out.
2. Measure the sales rate on in-stock days only
If a product sold 120 units in 90 days but was in stock for only 60 of them, its rate is not 1.3 units a day but 2. Check that a one-off promotion or season did not inflate the rate. Such a peak may not repeat.
3. Express the loss as a range
Not every customer leaves when an item is out of stock. In the review of physical-store studies, 31% of customers bought elsewhere and 9% did not buy at all. The rest switched to another brand or variant, or postponed the purchase. The retailer lost almost half of the intended purchases. Online, the share may differ. A competitor is one click away, but where a good substitute exists, some customers stay. Work with two bounds: an upper bound where every missed unit is lost, and a more cautious one where roughly half is lost.
Illustrative example with invented numbers:
| Value | |
|---|---|
| Sales rate on in-stock days | 2 units a day |
| Days out of stock in 90 days | 30 |
| Missed demand | 60 units |
| Gross margin per unit excl. VAT | CZK 500 |
| Lost margin – cautious bound (half) | CZK 15,000 |
| Lost margin – upper bound (all) | CZK 30,000 |
Even the upper bound leaves out lost basket items and future purchases. Your everyday data cannot measure those reliably; you only know they exist.
How much extra stock pays off
Against the lost margin stands the cost of extra stock. The classic inventory model calls these the underage and overage costs. Underage is what you lose when a unit is missing: simply put, the margin. Overage is what you lose on a unit left over: the purchase cost minus what you eventually clear it for. The optimal stock level gives a probability of covering demand equal to the share of underage in the sum of both costs.
Illustratively: a product sells for CZK 1,200 excluding VAT, costs CZK 700 to buy, and leftover units can be cleared at CZK 550. Underage is CZK 500, overage CZK 150. The critical ratio is 500 / (500 + 150) ≈ 0.77. It pays to hold enough stock to cover demand in roughly three out of four ordering cycles. The higher the margin and the cheaper it is to clear surplus, the more stock pays off.
Count lost baskets and future purchases in the underage cost and the ratio rises further. In the experiment cited above, including the short- and long-run cost of a stock-out cut the optimal stock-out rate from 20% to 12.2%. The firm’s actual rate was 21.9% of ordered items, because its planning gave no weight to stock-out costs at all.
The model assumes a single selling period. For goods you reorder continuously, surplus usually hurts less: a leftover unit sells in the next cycle and mainly ties up cash a little longer. Treat the formula as a guide to direction, not an exact order quantity, which also depends on lead time and minimum order size.
Out-of-stock products your ads still point to
See which out-of-stock products are still getting paid clicks and start your lost-sales estimate there.
Start with the biggest gaps, not all of them
Trying never to run out is expensive. A study of an online grocer found that stock-outs have a dramatic but nonlinear effect on profitability. A firm can capture much of the benefit with a modest reduction in stock-outs, provided it protects inventory first where it matters most.
Rank your best sellers by the lost margin from the calculation above, not by revenue. At the top you will find items with high margins, fast sales and repeated stock-outs. That is where more safety stock, more frequent orders or a conversation with the supplier about shorter lead times pays off. Items with a cheap, good substitute can wait.
When stock cannot fix a stock-out
Some stock-outs cannot be prevented, or preventing them does not pay. The supplier has no goods, the product is unique or seasonal, or the extra stock would tie up cash the business needs elsewhere. Past sales rates may also stop applying. Demand for a best seller may have been lifted by a one-off promotion, a trend or a competitor’s own stock-out. Before a larger order, check that the rate holds in recent in-stock weeks.
When a stock-out happens, how you handle it matters. Accurate availability information and back-in-stock alerts help keep interest alive. Be careful with discounts for waiting: in the experiment cited above they were the least profitable response, and among customers who were offered one, the stock-out did the most damage to their later purchases.
What to do now
- Pick the ten best-selling replenishable items from the last 90 days. Leave out unique pieces and goods you are clearing.
- For each, count the days out of stock from the stock card and calculate the sales rate on in-stock days only.
- Estimate lost margin as a range: a cautious bound (half the missed units) and an upper bound (all missed units).
- For the three items with the highest lost margin, compare underage and overage costs. Use the critical ratio to decide whether to raise stock, speed up replenishment or accept the stock-out.
- Write down what you do not know: the share of customers who waited or bought a substitute, and whether the rate will change after a promotion or season.
Korzaro stores the store’s stock level from Shoptet daily alongside orders and flags out-of-stock products that were still receiving paid clicks from Zboží.cz or Google Ads. These are often the first candidates for the calculation above. Korzaro does not calculate missed demand or order quantities; those remain your estimate.