A team of researchers went into eleven grocery stores and counted what was physically on hand for about 24,000 products, then compared it with what the stores' systems said should be there. The systems were right 35.3% of the time. For 39.4% of products, the record showed more stock than the shelf held. For the other 25.3%, it showed less.
That gap between the record and the shelf is the whole subject of inventory accuracy. It decides whether the store reorders on time, whether the online order gets filled or substituted, and whether the shopper finds what the app promised. In grocery, where products expire, get damaged, and move between the shelf, the backroom, and the promo end cap, the gap opens faster than almost anywhere else in retail.
What is inventory accuracy, and how do you calculate it?
Inventory accuracy measures how closely your stock records match what is physically there. The simplest version counts matches: of all the product-location records you checked, what share showed exactly the quantity you found?
Inventory accuracy = records that match the physical count ÷ total records counted × 100
That headline rate is useful, but it hides direction. A record that says 12 when the shelf holds 0 does more damage than one that says 0 when the shelf holds 12. The first leaves the shelf empty while the system thinks it's fine, so nothing gets reordered. The second ties up money and space. Track both, and track them by SKU and location, because an accurate store total can sit on top of badly wrong individual records.
How accurate are retail inventory records?
Less accurate than most systems assume. The grocery study above, by Rekik, Oliva, Glock, and Syntetos, found barely a third of records were right. It builds on earlier work by DeHoratius and Raman, who analyzed stock counts across 37 stores of a large US retailer and found that 65% of the inventory records were inaccurate, as the Rekik paper summarizes.
Those two numbers come from different retailers, years apart, using different methods. They agree on the part that matters: in a typical store, a record is more likely to be wrong than right.
Why does inventory accuracy break down in grocery?
Grocery adds causes that general merchandise mostly avoids.
Perishability. Products go out of date, get culled, or spoil in the backroom, and not every write-off makes it into the system. The Rekik study found inaccuracy rose with perishability.
Promotions. Promoted items sell fast, get restocked in a hurry, and often sit in more than one place at once. The same study linked promotions to higher inaccuracy.
Multiple locations. A product can live on the main shelf, an end cap, a seasonal display, and a backroom pallet. In a Grocery Gazette opinion piece, David Gottlieb of FORM put it bluntly: "Retailers' perpetual inventory data is often wrong, and many products are merchandised in more than one location simultaneously."
Online picking. When staff pick online orders from store shelves, every substitution and every short pick is a moment where the record and the shelf can drift further apart, unless the picking system updates stock as it goes.
What does poor inventory accuracy cost?
Sales, first. In the Rekik study, a store-wide stock count was associated with an increase in sales of about 11%, concentrated in the products where the system had overstated stock. Correct the record, and the store starts reordering the things it thought it already had.
Online, the cost shows up at the customer's door. Instacart, explaining why it bought a computer-vision company, said that "undetected out-of-stocks and catalog gaps are among the most persistent sources of customer dissatisfaction in online grocery," as Supermarket News reported. Every item the app shows as available but the picker can't find becomes a substitution, a refund, or a disappointed customer.
Shrink follows close behind. Inaccurate records hide where stock is being lost, which makes waste and theft harder to see and harder to fix.
Which tools improve inventory accuracy?
None of them fix accuracy alone. Most grocers combine several.
Targeted cycle counts. Counting everything is expensive. Counting the products most likely to be wrong is cheaper and catches more. The Rekik findings point to where to start: perishables, promoted items, and products with a history of overstated stock.
Shelf scanning with computer vision. Cameras on robots, carts, or fixed mounts check shelves continuously. Arpalus, the company Instacart acquired, says its system can identify individual items on shelves with more than 95% accuracy on average. That figure measures item identification, so treat it as the input to an accuracy program rather than the result.
Digital shelf labels. Electronic labels cut the manual work around pricing, which frees store time for counting and replenishment. CNBC reported that Walmart expects digital shelf labels in all its US stores by the end of 2026, and quoted an electronics team leader in Ohio who estimated they had cut her time on pricing duties by 75%.
Controlled storage for online orders. The most accurate inventory is stock that only moves when the system tells it to. In a dedicated fulfillment center or a grid-based micro-fulfillment site, every pick, putaway, and adjustment is a scanned transaction, so the record updates as the stock moves.
How does inventory accuracy affect online grocery?
Directly, and at scale. An online order is a promise built on the stock record: the storefront shows an item as available because the system says it is. When the record is wrong, the order fails on the floor, and the customer finds out at the door.
That's why online grocery puts more weight on accuracy than store retail does. A shopper in the aisle can see an empty shelf and pick something else. An online shopper can't. The retailers with the fewest substitutions are the ones whose stock records are closest to the truth when the order is placed.
Questions grocers ask about inventory accuracy
What is a good inventory accuracy rate?
It depends on how you measure it. Research on store inventory suggests most records are wrong, so the first goal is to measure accuracy honestly by SKU and location, then improve the categories that matter most to sales.
How often should grocers count inventory?
Use risk-based cycle counts. Count the products most likely to drift, such as perishables, promoted items, and lines with a history of overstated stock, more often than stable ones.
Does RFID work in grocery?
It works for some categories with consistent packaging. For fresh, loose, and variable-weight items, it is harder to apply, and most grocers rely on scanning and counts instead.
How does inventory accuracy relate to shrink?
They're linked but different. Shrink is stock that's lost. Inaccuracy is stock that's miscounted. Poor accuracy makes shrink harder to see, which makes it harder to reduce.
Accuracy you can pick from
For online grocery, the most reliable way to keep records accurate is to pick from inventory the system controls, where every movement is recorded as it happens. That's how Veloq's warehouse management system runs Rohlik Group's fulfillment centers. According to the Rohlik Group launch announcement, the platform has cut food waste to 0.5% of revenue, with more than 94% of orders delivered on time and error-free.
Grocers that pick from stores can bring the same discipline to a dedicated zone or a side-of-store micro-fulfillment center, with automated storage for the products that fit it.
The researchers in those eleven stores found the record right about a third of the time. The online customer, who never sees the shelf, deserves better odds. See how Veloq's WMS keeps the record and the shelf in step.
Sources
- Rekik, Oliva, Glock, and Syntetos, "Inventory record inaccuracy in grocery retailing: Impact of promotions and product perishability, and targeted effect of audits" (arXiv, revised June 8, 2026; Journal of Business Logistics, 2026)
- Grocery Gazette, "Grocery retail is revealing a bigger challenge for generative AI" (June 22, 2026)
- Supermarket News, "Instacart acquires computer vision company Arpalus" (July 16, 2026)
- CNBC, "Walmart digital price tags will be in every US store by end of 2026" (March 21, 2026)
- Rohlik Group, "Rohlik Group Introduces Veloq" (June 10, 2025)
