How to measure the out-of-stock rate and calculate the sales it costs
Who it is forGrocery chains, store operations, category managers, replenishment planners
Turkish retail uses four different expressions for the same problem: raf boşluğu (empty shelf), stoksuzluk (stockout), rafta bulunmama (not on shelf) and the English OOS. The confusion carries into measurement: most chains produce an availability report from system stock, the report looks fine, and meanwhile the shopper faces an empty facing. This guide separates the empty shelf from a warehouse stockout, shows how to catch it without a count and how to calculate the sales it costs. The basic SKU-level lost sales formula is in the fill rate guide; this guide covers the store and shopper behaviour side.
What is the difference between an empty shelf and a warehouse stockout?
All the shopper sees is the shelf, but an empty shelf has three separate causes and each is fixed by someone different. The first is a genuine stockout: the product is neither in the system nor in the store, so the problem lies in ordering or supply. The second is phantom stock: the system says the product exists but it is not in the store, so the problem lies in the record. The third is a replenishment gap: the product is in the backroom but has not reached the shelf, so the problem lies in in-store work.
An availability report based on system stock sees only the first case. In the second and third it shows the product as available; the product does not sell and nobody knows why.
Phantom stock is the most expensive. As long as system stock stays positive, automatic replenishment never fires; the shelf stays empty for days, until the next count or until someone notices. That is why on-shelf availability (OSA) is measured from the shelf itself, not from system stock.
| Case | System stock | In store | Root cause | Who fixes it |
|---|---|---|---|---|
| Genuine stockout | Zero | No | Short order, supplier delay, nothing in the warehouse | Replenishment planning, buying |
| Phantom stock | Positive | No | Unrecorded shrink, recording error, theft | Store, stock control |
| Replenishment gap | Positive | In the backroom | Fill timing, staffing, shelf capacity | Store operations |
On-shelf availability (OSA) is the complement of the out-of-stock rate: OSA = 100 − OOS rate. The two are faces of the same measure; state in the report heading which one is used.
How is the out-of-stock rate measured?
There are two methods and they complement each other. A shelf audit is direct measurement: at set times, for a set list of products, the shelf is checked item by item and every empty facing is recorded. It is precise but labour-intensive, so the list is limited to A-class products and the audit is run before and after the busiest hours.
The sales signal is indirect measurement and can be applied to every product. The logic is simple: if a product that normally sells steadily has not sold at all for longer than expected and system stock is positive, it is very probably not on the shelf. The signal catches both phantom stock and replenishment gaps.
When the signal counts as an alarm depends on the product's sales velocity. Assuming demand is random and steady, the probability of no sales over a given period follows from the velocity. For a product selling 6 units a day, a whole day with no sales has a probability of 0.25 per cent, a strong alarm. For a product selling 0.3 units a day, a day with no sales is normal, with a probability of 74 per cent, and says nothing.
- Out-of-stock (OOS) rate
Empty shelf observations ÷ Total shelf observations × 100- On-shelf availability (OSA)
100 − OOS rate- Probability of zero sales
e^(−Daily sales velocity × Number of days)- Alert window (days)
3 ÷ Daily sales velocityOver this window the probability of no sales falls below 5 per cent.
Run the sales signal excluding promotions, price changes and store closed hours. Otherwise the natural drop in a product whose price has gone up gets mistaken for an empty shelf.
How common are empty shelves, and what do shoppers do when they meet one?
The most widely cited study of out-of-stocks is the research Gruen, Corsten and Bharadwaj carried out for GMA in 2002, covering more than 71,000 consumers in 29 countries. According to the results cited in Pygmalios's 2026 report, the out-of-stock rate averages 8.3 per cent worldwide and 8.6 per cent in Europe: one in every twelve items on a shopper's list is not on the shelf.
According to the same research, 31 per cent of shoppers who meet an empty shelf buy the item at another store and 9 per cent do not buy it at all. In other words, for at least 40 per cent of shoppers the empty shelf moves that sale out of the store. According to Corsten and Gruen's 2004 analysis in Harvard Business Review, stockouts cost a fast-moving consumer goods retailer about 4 per cent of its sales.
No recent published out-of-stock measurement for Türkiye could be found. The figures below are global and dated; take your own rate from your own shelf audits and sales signal.
| Finding | Value | Basis |
|---|---|---|
| Out-of-stock rate, global average | 8.3% | GMA 2002, 29 countries, 71,000+ consumers |
| Out-of-stock rate, Europe | 8.6% | GMA 2002 |
| Shoppers who buy the item at another store | 31% | GMA 2002 |
| Shoppers who do not buy the item at all | 9% | GMA 2002 |
| Sales lost to stockouts | About 4% | Corsten and Gruen, Harvard Business Review, 2004 |
| Out-of-stocks on promoted products | About twice the normal rate | ECR Europe, 2003 |
| Global inventory distortion (2023) | USD 1.77 trillion: 1.2 trillion stockouts, 562 billion overstock | IHL Group |
How do you calculate the sales an empty shelf costs?
Lost sales never appear in the records; they are estimated. The first step is the stockout duration, measured in hours rather than days, because the shelf is usually empty for only part of the day. The share of daily sales that falls in the stocked-out hours is read from the hourly sales curve.
The second step is shopper behaviour. Some shoppers who meet an empty shelf take another brand or pack size, some put off the purchase; that part is not entirely lost to the store. The two shares in the GMA data, 31 per cent going to another store and 9 per cent not buying, are a definite loss to the store and serve as a lower bound in the calculation.
If you can measure your own substitution by category, use that rate instead: the rise in sales of similar products during the same hours shows the substituted share.
- Lost demand (units)
Normal daily sales × Share of daily sales in the stocked-out hours- Store loss (units)
Lost demand × (1 − Substitution and delay share)- Lost gross profit
Store loss × Unit gross profit
Lost sales must be added to the forecast series. If the raw sales of empty-shelf days feed the forecast, the order shrinks and the empty shelf becomes permanent; the fill rate guide covers this correction in detail.
Why do promoted products run out more often, and how do you reduce out-of-stocks?
According to ECR Europe's 2003 study, promoted products are out of stock about twice as often as regular products. The causes lie in three places: the demand spike turns out bigger than forecast, shelf capacity is set for normal sales, and promotional stock stays in the backroom instead of reaching the shelf in time.
The work that reduces out-of-stocks divides by root cause. Phantom stock is fixed through the record, replenishment gaps through in-store routines, genuine stockouts through ordering; treating all three as a single "availability project" fixes none of them.
- Produce the phantom stock list every dayProducts on the sales signal list are checked on the shelf. If the product is not in the store it is recounted, system stock is corrected and replenishment triggers by itself.
- Move shelf filling before the peakFilling is finished before the evening rush starts. A shelf that empties at peak costs many times more than one that empties in a quiet hour.
- Set shelf capacity to sales between deliveriesIf the shelf cannot hold the sales between two deliveries plus a safety allowance, it empties while the product sits in the backroom. Facings are set in the planogram from this calculation.
- Give promoted products separate capacityPromotional stock goes to a secondary display and the promotional forecast is kept apart from the normal forecast.
- Add lost sales to the demand seriesThe forecast is built on demand corrected for empty-shelf hours. Otherwise today's empty shelf shrinks tomorrow's order.
How are out-of-stocks tracked and closed in Mindpera?
In Mindpera, stock receipts and issues are analysed daily in Mind Ops and stock is monitored live. The shelf audit is completed on the shop floor as a checklist; an empty facing drops to its owner as an open item and escalates step by step if it is not closed. Agents flag a product whose sales have stopped for longer than expected despite positive system stock and open a shelf check task.
If the check reveals phantom stock, a recount task is opened in Mind Stock; the variance appears by product and location and the approved variance is posted to an accounting entry. On the genuine stockout side, Mind Send's unmet demand report shows which store ran short of what, and a replenishment proposal is produced for any store below its ideal stock level. Occupancy tracking in Mind Space separately lists shelves that are persistently empty.
- Mind Ops: daily stock movement analysis, shelf audit checklist, open items and escalation
- Mind Stock: recount tasks, variance analysis by product and location
- Mind Send: unmet demand report and replenishment proposals
- Mind Space: planned versus actual occupancy, list of persistently empty or overflowing shelves
If sales, stock and shelf audit data sit in separate systems, installation brings them into a single data model.
Frequently asked questions
- What does out of stock mean?
- Out of stock (OOS) means the product is not available at the point of sale; in a supermarket, the shelf is empty. Its complement is on-shelf availability (OSA). An empty shelf has three causes: the product is not there at all, it is in the system but not in the store (phantom stock), or it is in the backroom but has not reached the shelf.
- How is the out-of-stock rate measured?
- In two ways. In a shelf audit, a set list of products is checked on the shelf at set times and empty observations are divided by total observations. In the sales signal, products that have not sold for longer than expected despite positive system stock are flagged; the alert window is 3 divided by daily sales velocity. The audit is precise, the signal is comprehensive, and they are used together.
- How much do empty shelves cost in lost sales?
- According to Corsten and Gruen's 2004 analysis in Harvard Business Review, stockouts cost a fast-moving consumer goods retailer about 4 per cent of its sales. Your own loss is normal sales multiplied by the share of stocked-out hours and the share not substituted. According to GMA 2002, at least 40 per cent of shoppers do not buy the item in that store; use this as a lower bound.
- What is phantom stock?
- Stock that shows as positive in the system but is not physically in the store. It comes from unrecorded shrink, recording errors or theft. It is the most expensive kind of out-of-stock, because automatic replenishment never fires while system stock stays positive, and the shelf stays empty until the next count. It is caught by the sales signal and fixed by a recount.
- Why do promoted products run out more often?
- According to ECR Europe's 2003 study, promoted products are out of stock about twice as often as regular ones. The demand spike exceeds the forecast, shelf capacity is set for normal sales and promotional stock stays in the backroom. The fix is to give promoted products separate display capacity and keep the promotional forecast apart from the normal one.
- What should the out-of-stock rate be?
- There is no recent published benchmark for Türkiye; the most cited global average is the 8.3 per cent in GMA's 2002 study. Build the target from your own measurement: track the audited rate on A-class products, break it down by root cause and use the rate achieved by the best-quartile stores as the target for the rest. Do not use availability derived from system stock as the target.