EN 
PT 
ES 
In a results meeting, the supply director presents the quarterly indicator: out-of-stocks came in at 6%. On the other side of the table, the regional manager hears that and does not recognize his own stores. In the aisles he walks every week, there is a lot more missing than that.
Both are looking at real numbers. Neither is lying. What is happening is that they are not measuring the same thing, and nobody in the room noticed.
That scene repeats itself in chains of every size, and it has a simple explanation. There is no consensus in the literature even on the definition of an out-of-stock, precisely because there are different ways of measuring one. That is on the record in the 2009 study by Vasconcellos and Sampaio, published in the Brazilian Administration Review, which underpins much of this article.
Before discussing targets, causes or investment, it is worth settling the question that comes first: what exactly is your indicator counting.
If what you are after is the state of the market and what the largest chains are doing, we have already written about out-of-stocks in grocery retail. If you want to turn the indicator into money, see what out-of-stocks really cost. Here the subject is different: how to measure.
This article is about measurement. For an overview of the whole topic, with the types and the difference between out-of-stocks and waste, start with the out-of-stock guide.
What an out-of-stock is and how the rate is calculated
An out-of-stock is the absence of a product at the point where the customer reaches out for it. The item is part of the store assortment, it has a defined place on the shelf, and it is not there when somebody comes to buy it.
The out-of-stock rate is the percentage of items that should be on sale and are not:
out-of-stock rate (%) = missing items ÷ items sold by the store × 100
The worked example belongs to the Brazilian study itself. In a store with 5,000 catalogued and commercialised items, a rate of 10% means that 500 of them are not available on the shelf for immediate purchase.
The calculation is simple, and that is exactly where the trap lies. It does not say at what moment you count what is missing, nor how you found out it was missing. Two stores using this same formula arrive at different numbers depending on those two choices, and the rest of this article is about them.
Keep an eye on the denominator as well. It is the items catalogued and sold by the store. Not the number of sales, not the number of customers and, above all, not revenue. The out-of-stock rate counts products, not money.
For a market reference, the most widely followed Brazilian index came in at 10.8% in July 2026, and the figures updated by category are in our article on the sector. What matters here is something else: understanding why that number and the number your own store measures can be quite different without either of them being wrong.
The three definitions of out-of-stock in use today
Three respected sources publish figures on out-of-stocks, and each uses a definition of its own.
The first is the academic, shelf-based definition. An out-of-stock is when an item that is regularly sold, occupying a defined place on the shelf, is not available to the consumer at the moment of purchase. The authors add a phrase worth holding on to: an out-of-stock is characterised by an inefficient shelf replenishment process.
The second is the catalog definition, used by the Neogrid out-of-stock index. It is the percentage of items missing against the store's total catalog. The index is calculated over each store's assortment and takes no account of sales history or demand. And there is a detail almost nobody reads: the word stock there covers the entire physical space of the store, including the shelf and the storage area holding products that have not yet reached it.
The third is the outcome definition, used by the IHL Group. It counts an out-of-stock every time a customer arrives ready to buy and leaves without the item for any reason other than price. That covers the empty shelf, but also the locked case with nobody to open it, and the stock the system confirms and nobody can find.
Now apply all three to the same case. A case of parboiled rice arrived at the store, it is sitting in the stockroom, and nobody took it out to the shelf.
Under the shelf definition it is an out-of-stock, and a very typical one. Under the catalog definition it is not, because the item is inside the store. Under the outcome definition it is, provided some customer came looking and left without it.
The customer does not walk away with the rice under any of the three. The only thing that changes is the report.
The two types of out-of-stock, and why the indicator usually sees only one
This difference in definition is not a debate about vocabulary. It separates two problems that have different causes, different owners and different solutions.
There is the shelf out-of-stock, where the product is already in the house and does not reach the shelf. And there is the systemic out-of-stock, where the product simply does not exist in the chain, because ordering or forecasting failed.
The simplest way to tell them apart is to ask where the product is at the moment the customer cannot find it. There are four possible answers, and only the first is good news:
- it is on the shelf, and there is no out-of-stock
- it is in the store, in the stockroom or the back area, and nobody replenished it
- it is at the distribution center, and was not shipped to the store
- it is nowhere in the chain.
The two in the middle are shelf out-of-stocks in the broad sense, because the product is yours and has already been paid for. The last one is a systemic out-of-stock.
The largest survey ever carried out on the causes of out-of-stocks measured the weight of each case. The 2002 study by Gruen, Corsten and Bharadwaj, conducted for the Grocery Manufacturers of America and the Food Marketing Institute, triangulated 52 studies and surveyed 71,000 consumers across 29 countries. The worldwide average of causes came out as follows:
- store ordering, placed late or in insufficient quantity: 34%
- shelf replenishment, with the product in the store and off the shelf: 25%
- head office or manufacturer decisions: 14%
- store demand forecasting: 13%
- distribution center: 10%
- other causes: 4%.
Notice that ordering and forecasting appear separately, and that the study explains the relationship between them: almost half the causes trace back to ordering problems, meaning the retailer buying late or buying too little, often because the forecast it relied on was not reliable.
And pay particular attention to the two lines that make up replenishment. The product sitting in the store is worth 25%, and the product sitting at the distribution center adds another 10%. The authors treat the two together and say, in as many words, that around one third of the causes are replenishment problems, predominantly the product that is in the store and does not reach the shelf, combined with the flow of goods that never leaves the warehouse for the store.
In other words: in a chain with its own distribution center, the question is the product already mine? has two stages, not one. It can be sitting in the back of the store or sitting at the DC, and in both cases the money has been spent and the sale does not happen.
The authors' general conclusion is that two thirds to three quarters of out-of-stocks originate inside the store, and one quarter to one third come from levels above it, at the distribution center or at head office.
By region the difference is large. In Europe, 38% of out-of-stocks happen with the product inside the store, against 22% in the United States and 15% in Asia. The analysts record that the European result was counter-intuitive even to them, since they expected the opposite given smaller back rooms and efficient transport. Out-of-stocks caused by product available at the distribution center but not shipped to the store, on the other hand, appear fairly evenly distributed around the world.
What happens when you cross the two types with the three definitions
Here the conversation turns practical. A shelf-based indicator sees both types, because it looks at the shelf and does not ask where the product is. An outcome-based indicator does too, because it starts from the customer who left without the item. A catalog-based indicator, which treats the stockroom as availability, sees the systemic out-of-stock and does not see the shelf one.
Anyone following only that last indicator watches the number improve every time buying gets better, and never sees the replenishment problem, which is still there, worth a quarter of the total inside the store and another tenth on the road between the distribution center and the store.
It is worth noting that there is a third case, one every operation knows well, which is neither of the two types: the item is in surplus at one store and missing at another in the same chain. The product is in the house and the problem is distribution. We did not find a primary study that quantifies this share, so we are not going to invent a figure for it.
The Brazilian case
The survey of 95 supermarkets in the state of São Paulo, members of the APAS association, put this specific cause to the managers. The question covered the shelf not replenished by the merchandiser with the product sitting in the stockroom, on a scale of 1 to 7.
At conventional supermarkets, which account for 60 of the 95 stores surveyed, the average came to 3.84 with a p-value of 0.000, meaning statistically significant. At hypermarkets the figure was 3.36 and at compact supermarkets 3.11, neither of them significant.
In plain terms: managers at the most common format in the sample confirmed that the product sitting in the stockroom and never reaching the shelf is a real and frequent cause. It was measured in Brazil, with a method, and it is nobody's opinion.
The four measurement methods, and what each one gets wrong
On this point the 2009 study is still the best reference available, because it is where the methods appear side by side with the flaw of each. The four recognized by a committee of retail professionals:
Visual shelf auditing checks availability directly on the shelf. It is the most effective measurement from the customer's point of view. The problem is that it is expensive and eats into the organization's scarce resources, which is why it ends up being periodic and sampled.
Store inventory level calculates availability from records of goods in, goods out and balance. It is cheap, systematic and the only one that covers 100% of the range. The problem is that the share of inconsistencies is high, and that the product may be in the stockroom rather than on the shelf. Note that the flaw identified in 2009 is exactly the shelf out-of-stock that the global study sizes at 25%.
Asking the consumer at the checkout is cheap, but it is not systematic. It depends on the operator asking every time, and the customer may simply have failed to find the product and reported it as missing.
Information from the supplier is reliable, but only a small share of a store's items has a merchandiser paid for by the manufacturer.
ECR Retail Loss, looking at European retail, goes as far as listing ten variations in use, from distribution center service level to items below 70% of shelf capacity and sales exceptions against historical patterns. And it concludes that every one of them has limitations, which produces long internal arguments about the real size of the problem and, next in line, about who answers for it.
That this argument is not theoretical, ECR measured. At an event bringing retailers and manufacturers together in Brussels in 2012, 48% of those present said their organization had no out-of-stock indicator that was respected internally. And 30% of the retailers answered that no function was accountable for on-shelf availability.
Why the system balance does not match the shelf
The cheapest method is the one whose weakness is best documented, and the documentation is recent.
In a three-year project published in 2020, Rekik, Syntetos and Glock measured for ECR Retail Loss the accuracy of inventory records at seven of the largest retailers in Europe, spread across four countries, covering around one million items in approximately 100 stores. The authors make a point of saying that the volume of data is of a different order of magnitude from anything attempted before.
The central result: 59.54% of the items audited had a physical quantity different from the one recorded in the system at the time of the count. In grocery and general merchandise the figure rises to 63.39%. It is consistent with what DeHoratius and Raman had already found in 2008 in Management Science, across almost 370,000 records from 37 stores, where 65% were inaccurate.
Two findings change how an inventory report should be read.
The first is that the discrepancy does not only run one way. In 27.46% of items the physical count was higher than the system. In grocery and general merchandise, 38.52% had a negative discrepancy against 24.87% with a positive one, and the authors attribute the difference to the losses typical of the sector, such as spoilage and damage.
The second is more uncomfortable. The authors record that, as far as they know, there is no software package that factors inventory record inaccuracy into replenishment calculations. The system places the order as though the balance were correct.
The item frozen at zero
This is the study's most useful finding for anyone designing a measurement, and it is the kind of thing no monthly report shows.
At one of the retailers analyzed, 15.07% of items had zero physical stock at the time of the count, which took their sales to zero the following week. Worse: 2,402 items, or 8.8% of the total, stayed in that state for the entire experiment, producing no point-of-sale signal at all.
Stop on that point. An item frozen at zero does not sell, and because it does not sell it produces no sales exception, never falls below its historical average and triggers no alert whatsoever. Every method that looks for out-of-stocks in sales behavior is blind to it by construction. Only a physical count or a shelf audit will find that item.
It is the extreme version of phantom inventory, and it can last for months without anyone noticing.
Why today's number does not compare with 2009
That survey of 95 supermarkets in São Paulo found an average rate of 8.3%, with a standard deviation of 6.8. By format, it was 9.1% at compact supermarkets, 8.5% at conventional ones and 7.1% at hypermarkets. Analysis of variance found no significant difference between the formats, meaning store size did not change the result.
The temptation is to take today's index, subtract 8.3 and announce that out-of-stocks have got worse. You cannot.
The definitions are different, shelf against catalog with the stockroom included. The methods are different, manager-reported perception against continuous measurement of store data. And the samples are different.
What can be stated is more modest and more useful: the numbers sit in the same order of magnitude as a problem the literature was already describing as chronic. The study itself compiles estimates above 8% across forty years, from 1963 to 2003, and the authors are explicit about the limit of the thing when they say that out-of-stock rates will never be zero.
That answers, sideways, the question that comes up most. No serious source establishes an acceptable rate. What exists is a historically stable level and a warning from the Brazilian authors: managers tend to treat the measured figure as a normal level of occurrence, and a subject that becomes routine rarely improves.
One honest caveat about the age of the sources, because it deserves to be said out loud.
The Brazilian survey is from 2009. The worldwide survey of causes, the source of the 25% for replenishment and the 34% for ordering, is from 2002. We looked for more recent replacements with the same rigour and did not find them, and we are not alone: the 2002 survey is still the reference the later literature cites whenever it needs to discuss root causes, and the figures circulating today as a global average are usually its own, reused without fresh measurement.
Where newer data exists, we used the newer data. Inventory accuracy comes from a 2020 measurement across one million items. Where the older studies still stand alone, as in the comparison of methods and the attribution of cause, we kept them with the date in plain view for you to judge.
What the gap in perception reveals
That same survey contains the result the authors call unexpected, and it explains a great many stalled meetings.
Managers across all formats pointed to failures in the supplier's logistics process, late delivery and incorrect delivery, as the main cause of out-of-stocks. At compact supermarkets, the highest average went to late delivery, with 4.40 and a p-value of 0.006. At conventional ones, the supplier not having the product available scored 4.18 with a p-value of 0.000.
What makes the result unexpected is that the global study attributes 70% to 75% of out-of-stocks directly to the store's own practices.
The nuance sits in the Brazilian data itself, and it is revealing. At conventional supermarkets, alongside the supplier, two internal causes also came out significant: replenishment not carried out with the product in the stockroom, and delay in generating the order. Compact supermarkets pointed to insufficient shelf space, which is internal planning. And hypermarkets did not name a shortage of shelf stackers among the main causes, even though much of the international literature elects exactly that as retail's principal problem.
There is a second contrast, about priority. Asked which attributes weigh on a consumer's choice of supermarket, managers placed out-of-stocks fifth out of eight, behind friendly service, low prices, promotions and variety. An American study from 2006 identified out-of-stocks as the single most important attribute in explaining customer satisfaction at a chain in the United States.
The Brazilian authors comment on this without hedging. A manager's perception can differ from the customer's, and the fact that out-of-stocks are not a manager's priority may, on its own, help explain the high rates.
Adding up what was measured afterwards, the balance tips. With 25% of out-of-stocks originating in replenishment, around 60% of inventory records diverging from the physical count and 8.8% of items capable of freezing at zero without producing a signal, most of the problem lives inside the store. That does not prove the Brazilian managers are wrong about the supply chain here, which is different from the European and American ones. But it is a concrete reason to be sceptical of any attribution of cause that does not come from measurement.
How to measure out-of-stocks in practice
Pulling together what the sources show, five decisions solve most of the problem.
State the definition before the number, and say what it does with the stockroom. That is the first question to ask about any index, yours or a third party's. Does a product that is in the store and not on the shelf count or not? The answer moves a quarter of the problem.
Measure the two types separately. Shelf out-of-stocks and systemic out-of-stocks have different causes, different owners and different solutions. Added into a single indicator, one hides the other, and an improvement in buying masks a deterioration in replenishment.
Choose a primary method, name what it cannot see, and cover that blind spot with a second one. Shelf auditing cannot see what happens between two counts. The system balance is wrong on around 60% of items and cannot see the product in the stockroom. Any method based on sales is blind to the frozen item. That is what the stores surveyed did in practice, combining methods, even without writing it down in any manual.
Break the data down by category and by sales velocity. The aggregate index hides what matters, both in out-of-stocks and in record discrepancy, which concentrates by category, and in counting effort, which pays off most on fast movers with high discrepancy.
Treat counting as an investment, not an obligation. Correcting inventory records lifted sales at all seven European retailers studied by ECR, and it is the strongest argument available for defending the budget of a measurement program. The figures behind that calculation, and the three errors that sink a loss estimate in a meeting, are in what out-of-stocks really cost. To estimate the value on your own numbers, our out-of-stock calculator builds the figure from the number of stores, the revenue and the rate in use.
The right indicator is the one that points to a decision
The difference between a chain that reduces out-of-stocks and one that lives with them rarely comes down to effort. It comes down to knowing which question the number answers.
An index that treats the stockroom as availability answers well whether buying is on target. It does not answer whether the shelf is full. Anyone using the first to decide about the second will invest in the wrong place, and will do so with the clear conscience of somebody holding an indicator.
It was with that kind of blind spot in mind that we built INTEGRA, our autonomous replenishment platform, which handles each type of out-of-stock differently. It cross-checks expected sales against actual sales and inventory balance to find the item that is sitting in the store and not selling, which is the signature of a shelf out-of-stock, and it projects demand against supplier lead time to warn the buyer before the gap appears. If you want to understand what out-of-stocks cost in your operation, talk to one of our specialists and we will run that calculation on your own numbers.
Frequently asked questions
What is an out-of-stock?
It is the absence of the product at the point where the customer reaches out for it. The item is part of the store assortment and has a defined place on the shelf, but it is not there at the moment of purchase. The out-of-stock rate is the number of missing items divided by the total number of items the store sells, multiplied by one hundred. In a store with 5,000 catalogued and commercialised items, a rate of 10% means 500 items unavailable on the shelf. The formula, however, does not define when you count or how you observed the gap, and it is from those two choices that the differences between indicators come.
Does a product sitting in the stockroom but not replenished on the shelf count as an out-of-stock?
It depends on the definition, and this is the most important difference between indicators. Under the academic definition it does, because an out-of-stock is characterised by an inefficient shelf replenishment process. Under the Neogrid out-of-stock index it does not, because the term stock there includes the storage area. Under the IHL Group criterion it does, if the customer came looking and left without it. The 2002 global study by Gruen, Corsten and Bharadwaj sizes this case at around 25% of all out-of-stocks on the worldwide average, reaching 38% in Europe, 22% in the United States and 15% in Asia.
What are the types of out-of-stock?
There are two, and the difference lies in where the product is. In a shelf out-of-stock the product is already yours and is sitting somewhere, either in the back of the store, which accounts for 25% of causes on the worldwide average, or in the distribution center without having been shipped, which accounts for a further 10%. The global study treats the two together and says that about one third of out-of-stocks are replenishment problems. In a systemic out-of-stock the product does not exist in the chain at all, and here we find store ordering placed late or in insufficient quantity, with 34%, store demand forecasting, with 13%, and head office or manufacturer decisions, with 14%. There is also an operational case that is neither of the two, which is an item being in surplus at one store and missing at another in the same chain.
What are the methods for measuring out-of-stocks?
The Brazilian study presents four recognized by a committee of retail professionals: visual shelf auditing, store inventory level, asking the consumer at the checkout and information from the supplier. ECR Retail Loss lists ten variations in use in European retail. None of the sources points to a method without limitations, and the stores surveyed usually combine more than one.
Why does the system balance not match the shelf?
Because the record is wrong on a large scale and because the product may be in the stockroom. In the largest field study ever conducted on the subject, covering seven of the largest European retailers and around one million items, 59.54% of them had a physical quantity different from the one in the system, reaching 63.39% in grocery. And the discrepancy does not only go one way: in 27.46% of items there was more physical stock than the system recorded.
What is phantom inventory?
It is when the system shows an available balance and the shelf is empty. The extreme version appears in the ECR Retail Loss study: items with zero physical stock that stay frozen in that state and stop generating any sales signal. At one of the retailers analyzed, 8.8% of items remained like this throughout the entire experiment. Because the item does not sell, it does not trigger any sales-drop alert, and only a physical count or a shelf audit will find it.
What out-of-stock rate is acceptable?
No serious source sets a target. What exists is the record that estimates have stayed consistently above 8% for forty years and the authors' observation that out-of-stock rates will never be zero. Comparing your own historical series, always measured the same way and with the same definition, is more useful than chasing a benchmark from another market.
Who is responsible for out-of-stocks?
The global study attributes 70% to 75% of out-of-stocks directly to the store's own practices, across ordering, forecasting and replenishment. The Brazilian managers surveyed in 2009, on the other hand, pointed mainly at the supplier, a result the authors themselves classify as unexpected. The inventory accuracy data measured later reinforces the first side.
This content was produced by the Integrity-UX team, an IT consultancy specializing in infrastructure, cloud, security and business continuity for medium and large companies.
