The size you ran out of never shows up in your sales

The shopper the report never meets

Picture a shirt that sold well last month. The report says thirty went out of the door. What the report cannot say is how many more people walked up to the rail, looked for their size, did not find it and left. They bought nothing, so they left no bill, and a sales report is made of bills.

That gap is the subject of this note. Running out of stock is an old problem, and the lost sale is the obvious cost. The second cost, in our view, is easier to miss: the empty size quietly teaches the next order to buy too little of it.

Chart 1Hidden in plain sight

The sale nobody records

Three findings from studies of stock in stores, reported in 2002 and 2023

$1.2trillionthe world's cost of out-of-stocks, as IHL Group reported in 2023

Over 65%of stock records wrong at one large retailer, in a 2002 study

16%of items could not be found by shoppers who asked for help at another, though not out of stock

How big the hole is

The analyst firm IHL Group, reporting in 2023, put the world's combined cost of out-of-stocks and overstocks at $1.77 trillion. Out-of-stocks made up 68.2% of that, about $1.2 trillion; overstocks were the other 31.8% ¹. Among the reasons shoppers gave for leaving without what they came for, the study lists sizes not being available, no staff to help, products locked up or in the stockroom, and empty shelves ¹.

Chart 2

Running out costs more than having too much

Share of the $1.77 trillion cost of out-of-stocks and overstocks worldwide, IHL Group, reported 2023

Inventory distortion

68.2%31.8%

Out-of-stocksOverstocks

Only one of those is an empty shelf. A missing size is a gap that a full-looking rail can hide, and a product locked up or in the stockroom is stock that exists but is out of reach. An older study makes the same point from inside the store. In 2002, researchers from Chicago and Harvard drew on audits at two large retailers known for their operations ². At one, 16 percent of items could not be found by shoppers who asked a sales associate for help, and the items were not out of stock; they were misplaced in storage or on the wrong aisle ².

Sales are not demand

Here is the part that is easy to miss. When a size runs out, the demand for it does not show up anywhere. Statisticians call this censored data ³. As one 2024 paper puts it, unmet demand is lost "and, more importantly, most of the time is also unknown", so sales figures "underestimate the underlying demand" and push forecasts down ³.

Chart 3

What the report sees, and what happened

Our illustration of censored demand: when a size runs out, its demand leaves no trace

What the sales report shows

What happened on the floor

A size sold out

More shoppers wanted it than there were pieces

No sale after that

Shoppers asked, found nothing and left

Sales by size look settled

The best-wanted size is under-counted

Reorder in the same ratio

The same size runs short again

In a shop, the effect is concrete. Say a buyer looks at last season's sales by size and orders in the same proportions. The size that ran out first was capped by what was on the rail. Its true demand was higher than its sales. Order to the sales, and you order too little of exactly the size that was most wanted.

Sales represent a constrained demand, not the demand itself.

In our view, this is how a store can run short of the same size season after season while every report looks reasonable. Nobody made a mistake that shows up in the numbers. The numbers themselves are short.

Chart 4

How a short size stays short

Our reading of how censored demand feeds back into buying

1A popular size sells out first

2The sales report counts only what sold

3The next order follows the report

4The same size runs out sooner

→↓

↓

←↓

↑↺

the short-size loop

What the shopper does next

Whether an empty shelf costs you the sale depends on what the shopper does. In an April 2026 survey of 1,000 US shoppers by Profitero, the most common response to an out-of-stock was to buy a similar product from a different brand in the same store (36%), just ahead of staying with the same brand (33%) ⁴.

Chart 5

A rival brand is the most common answer

What 1,000 US shoppers said they do when an item is out of stock, Profitero, April 2026

36%buy a similar product from a different brand in the same store

33%stay with the same brand

Two cautions. That survey covers shoppers in general, not clothing; and the figures say what people report, not what they did. Still, the split is useful. A brand loses when the shopper switches brand. A store loses when the shopper switches store. In clothing, our view is that the store has the weaker hand: a missing size has no close substitute on the next rail, the way one brand of biscuits stands in for another.

The stock you think you have

There is a quieter version of the problem. Sometimes the size is in the building, but the system does not know it, or does not know where.

In the 2002 study, more than 65 percent of the inventory records at one retailer were inaccurate: the recorded quantity did not match what was found in the store, across 37 stores ². The researchers say inaccurate records can mean carrying too much stock without realising it, or too little and not reordering it, a problem that reduced that retailer's profits by more than 10 percent ². These were companies "well regarded in the industry for their excellent operations" ².

If the system says three of a size are in stock and the rail is empty, nothing triggers a reorder. The size is missing twice: from the rail and from the report.

Chart 6Up to 95% accurate

Tags fix the records

Inventory accuracy with barcodes and with item-level RFID, Project Zipper (GS1 US and Auburn University), 2018

63%

95%

BarcodeRFID

Item-level tagging is one answer. In Project Zipper, a 2018 study by GS1 US and Auburn University's RFID Lab with five retailers and eight brands, RFID raised inventory accuracy from 63 percent to 95 percent compared with barcodes, and reduced out-of-stocks by up to 50 percent ⁵. Put another way, more than one record in three was wrong with barcodes alone, against one in twenty with tags.

How Zara closes the loop

Zara is the usual example, and the detail is worth knowing. In 2016, trade press described how Inditex, Zara's owner, coded and read each garment at its logistics centre. When shipments reached the stores, two times each week, the system identified which sizes and models needed replenishing ⁶. The same system let staff check whether a size is available in the store, in nearby stores or online ⁶. Inditex's then chief executive, Pablo Isla, called it "one of the most significant changes in how the group's stores operate" ⁶.

Chart 7

Zara's restock rhythm

How trade press described Inditex's system in 2016

At the warehouseevery garment is coded and read

Two times a weekshipments reach the stores

Each deliverythe sizes and models to replenish are identified

On the floorstaff can check a size in store, nearby or online

In our reading, the lesson is not the technology. It is the rhythm: frequent, small deliveries, each one sized by what actually left the rail since the last one. Each delivery corrects the last order before a broken size has weeks to cost sales. In our view, a store without tags can copy the rhythm, if not the precision.

Where this does not work

Buying more of everything is not the answer. Overstock was close to a third of IHL's 2023 figure ¹, and in our view, in clothing the extra units tend to end up marked down. The aim is better proportions and faster correction, not deeper stock.

In our view, tags cost money, and for a small store with a few hundred styles a careful weekly count may do much of the same job. Frequent deliveries need a supplier or warehouse that can send small quantities often; a store that gets stock once a season cannot react inside the season. And none of this fixes a style nobody wants. Censored demand is about a good product running short, not about a bad one.

What to do on Monday

1. Log the walk-aways. Keep a simple sheet at the counter: style, size, colour, for every shopper who asked for a size you did not have. It is the only record of demand your bills will never show.
2. Check the system against the rail. Each week, pick ten fast styles, count every size on the floor and in the back, and compare with what the system says. Fix the records before you reorder.
3. Watch the core sizes, not the style. A style is only fully on sale while its most-bought sizes are in stock. When the first of those breaks, act then, not when the style looks low overall.
4. Order to demand, not to sales. When you rebuy, add back the walk-aways and the days a size was missing, so the next order is not shaped by last season's gaps.
5. Ask for smaller, more frequent deliveries. Two small top-ups beat one large one, because the second can follow what actually sold.