How to Find Out-of-Stocks in a Retail Chain, Store by Store

An empty shelf rarely shows up in a chain report. The report adds up the stock of every store, and the total looks fine while one store has had nothing to sell since Saturday. In Italy, where the grocery trade measures this, shoppers met an empty shelf on 9.9% of their shopping trips and could not find 4% of the products they came for².

You don't need shelf cameras to find those shelves. The till and the stock record already hold most of the answer, read store by store. Four checks find them, from the cheapest to the most work.

What an empty shelf costs

Not every out-of-stock is a lost sale. When ECR Italia had several hundred shoppers keep a diary of every product they could not find, a third of the time they took something else, and almost as often they went without²:

The two outlined parts are sales the chain lost, and some of them are customers who just learned where else to shop. Substitution is not free either: the shopper who switched brands may have paid less, or picked a product with a thinner margin.

Check 1: zero stock on a product that sells

The first check is the one your data already shows. List every product with zero stock in a store where it sold in the last few weeks. Do it per store, never on the chain total:

Then put a price on each line, because a list of two hundred empty shelves gets read from the top. Use what the product takes in that store on a day it sells, averaged over the last four weeks. An average over all days undercounts, since the days it was already out of stock pull it down. Multiply by the days the shelf has been empty and sort each store's list by that amount. The empty-shelf playbook covers what a day of absence is worth in more detail.

A negative stock figure belongs on this list too. It means the record has gone wrong somewhere, and the inventory shortage essay walks through how to find where.

Check 2: stock on paper, nothing at the till

The harder out-of-stocks are the ones the stock record does not show. The system says there are units on hand; the shelf is empty, or the units are in the back room, damaged or stolen. Because the record never reaches zero, the store does not reorder either.

This is common. When researchers counted about 500 products in each of 10 stores of dm, the European drugstore chain, 31% of them did not match the record, and records showing more than was on the shelf were the more frequent error in every store. Empty shelves behind a positive record ranged from 0.4% to 2.0% of products, depending on the store³. At a UK grocer, 39.4% of stock records showed more than was actually there⁴.

The till gives these shelves away. A product that sells about two a day goes a day without a sale now and then; four days in a row, with stock on the record, happens by chance about once in three thousand runs. A run that long means the product is not reaching the shopper:

Italy's grocery industry measures availability this way for the whole market: GS1 Italy and Circana read sales per product, per store and per day from 2,524 stores, and count a product as out of stock on a day it sold nothing but was expected to sell¹. In September 2025 that gave an out-of-stock rate of 3.4% and lost sales of 4.2%. Promoted products ran at 0.7%: when a store expects a spike, it fills the shelf.

Send these lines to the store as a list to count and check. It pays off: in the dm study, counting the 4% of products a zero-sales model ranked highest found 42% of the hidden empty shelves³, and at the UK grocer a full count lifted store sales by about 11% over the following two months, all of it on products whose record had shown more than was there⁴.

Check 3: the store against the others

A chain has an advantage over a single shop: the same product sells in other stores at the same time. If it sells everywhere this week except in one store, demand did not change; that store's shelf did. Researchers who built a detection method on this idea read the same product in other stores and other products in the same store, rather than one product's history alone⁵.

The simple version fits in a spreadsheet. For every product in every store, compare this week's sales with that store's own average for the last ten weeks, and flag the lines that fell below half, the rule those researchers tested on a retailer's data⁵. Then drop the products that fell in every store: that is demand, a season or a price change. What is left is a list of shelves to look at.

Check 4: what runs out before the next delivery

The first three checks find shelves that are already empty. The fourth finds the ones about to be. Divide each product's stock in each store by its average daily sales over the last month: that is its days of supply. Any line with fewer days of supply than the days until the store's next delivery will run out before stock arrives. The dead-stock guide uses the same number from the other end, to find stock that will not sell.

One habit makes this worse. After an out-of-stock, the sales history shows a quiet week, because nothing was there to sell. A study of a supermarket chain's automated ordering found that managers were more likely to cut the proposed order after a stock-out, which made the next stock-out more likely⁶. The quiet week was the empty shelf, not the demand.

One list per store, every morning

The four checks belong in one list for each store, in this order, so the store manager sees the most money first:

All four checks depend on the stores' data being comparable: the same product code in every store, sales dated by business day, stock read after the last sale of the day. The guide to consolidating POS data across stores covers that groundwork.

Test yourself: five calls on empty shelves

Can you find the empty shelf?

Make five calls on stock and sales in a store chain and leave with a list you can run every morning.

• 42 unitsChain stock
• 20 selling daysLast 28 days
• 14 on handRecord
• below halfOne store
• 3 empty daysLast week

Start the quizStore by store, Till and stock, 5 calls

How marql finds out-of-stocks

marql reads sales and stock from a chain's POS and ERP. After every stock sync it lists, store by store, the products at zero that sold in that store in the last 30 days, and raises one card per store with the eight that cost the most and the sales at risk per week across all of them. Each product is priced by what it takes on a day it sells, over the last four weeks. Warehouses are left out.

Pick a store in Analyze, under Products and Stock, and every product shows its days of supply from the last 30 days' sales: critical under three days, low under seven, with the revenue at risk from both.