If you run a retail business, you already know your numbers may not tell the truth, or at least not the whole truth. Your system says you have the product. The shelf says you don’t. And the customer? They already may be on their way to a competitor.
That gap between balance on hand, what your system claims is in stock, and on-shelf availability (OSA), what’s actually there for the shopper, is more than a numbers and operations headache. It’s a revenue drain and a competitive threat.
Across the industry, inventory accuracy is shockingly low. Gartner’s Hype Cycle for Revenue and Sales Technology says store-level accuracy can be as low as 60%. More than half of retailers, according to an IHL Group survey, admit their data isn’t even 80% correct. That means phantom inventory — stock the system thinks exists but doesn’t — is almost guaranteed in every store.
Shoppers notice, and take action. Nearly 30% will switch stores if they can’t find what they want, according to a study from Numerator. More than 70% will switch brands, according to Adobe. These lost sales fuel loyalty erosion, resulting in further missed opportunities. Add it up and the inventory distortion problem for retailers worldwide amounted to $1.7 trillion in 2024, according to IHL Group. The damage is slow but relentless. Every time a customer is disappointed, they’re a little less likely to come back.
Advertisement
The Problem you Can’t Count your Way Out of
The balance on hand and on-shelf availability disconnect has been around forever, but the forces making it worse are multiplying. Phantom inventory is the biggest culprit, caused by theft, shrinkage, inefficient promotion management, and misplaced items. Retail Insight analysis suggests it can drive as much as 80% of out-of-stocks.
But the converse is also a major problem. When balance on hand shows zero, even if stock exists, this triggers unnecessary reorders, driving excess stock, waste and holding costs — especially for perishables. Misplaced inventory adds to the issue: items may be in the backroom, overstock areas or displays, but not on the intended shelf. Associates waste time searching, shelves stay empty and customers leave frustrated. Relying on balance on hand alone distorts forecasting, fuels over-ordering and undermines on-shelf availability. A positive balance on hand doesn’t guarantee products are where shoppers expect them.
Modern supply chains make accuracy harder to control. Products move from factory to distribution center to backroom to shelf with multiple handoffs that can distort data. Omnichannel retail compounds risk, as the same inventory must serve stores, online orders, pickup and ship-from-store. Human error adds more gaps, such as manual counts, delayed updates and overstretched associates. Traditional fixes like quarterly counts or spot checks are too slow and partial. Inventory changes constantly; without shelf visibility, you’re running on lag. And in retail, lag is lethal.
The Way Forward
Improving high on-shelf availability levels is a constant challenge for retailers. With a 90% OSA rate across 30,000 SKUs, that translates to roughly 3,000 out-of-shelf events every day. Which of these can be replenished immediately? Which are out-of-stock? And which ones have high demand and high margin and should be addressed first? The product is available in the system, but cannot be found in the back of the house? The only way to close the gap between on-hand inventory and on-shelf availability is to see reality as it happens. That means embracing shelf visibility through AI and object recognition.
Computer vision and machine learning, which are fields of AI, can continuously scan shelves, detect gaps, misplaced items and compliance issues without needing manual checks. Together, these capabilities create a single, connected system that not only flags issues, it prevents them and ensures products are available where and when customers expect them.
Imagining the scenario with 3,000 out-of-stock events every day, the challenge goes beyond identifying such issues to receiving prioritized alerts, in order to address those that will really influence revenue and customer experience, all before the customer ever notices…
- When a shelf goes empty
- How much of it the retailer actually has
- Who’s available to fix it
That’s not a future vision. It’s live in stores today. An increasing number of retailers are embracing this shelf intelligence to stay ahead of today’s operational challenges. These solutions deliver actionable and prioritized insights on on-shelf availability — the real state of the shelf. And by consistently scanning shelves every day, store associates can catch errors as they happen and update the system, reducing the lag between when stock is missing and when it’s replenished.
This continuous, accurate feedback loop not only increases OSA, it minimizes inventory errors and phantom stock, ensuring the system reflects what’s actually on the shelf. Over time, this prevents the build-up of undetected stock errors, making inventory more and more accurate over time, which also will improve demand forecast, reordering and decision making. As an example: in just six months, one of Europe’s largest grocery chains, using AI-powered shelf intelligence, exceeded 95% on-shelf availability, up from 90%, with sales growing about 2%. That’s the power of removing blind spots.
And the impact goes beyond revenue. Having on-shelf availability and balance on hand in harmony builds trust. Customers believe you’ll have what they want, when they want it. Suppliers see you as a reliable partner. Competitors can’t lure customers away on stock availability alone. Store associates are happier and can work more efficiently. The reverse is also true: an imbalance breaks trust with everyone. Once shoppers find a retailer that rarely disappoints them, they won’t tolerate one that does.
That’s why this isn’t a side project for IT or operations. It should be a leadership priority. Retailers that understand the root causes and solve this set a standard. Those who delay must play catch-up in a game where customers don’t give second chances.
The path forward begins with high-ROI applications built into a single system: automated shelf monitoring, intelligent replenishment and prioritized alerts. The key is seamless integration within existing workflows, so stores run smarter without adding complexity. Associates should see this technology as an asset, not a burden. And move quickly, because every day wasted is one where a competitor can gain ground.
The balance on hand and on-shelf availability gap is retail’s most expensive blind spot. AI can help to close it. Not with guesswork. Not with delayed reports. But with hard, timely facts you can act on instantly.
Accurate data + visible shelves = revenue and trust.
As CTO and VP Product, Scandit Co-founder Christian Floerkemeier is responsible for Scandit’s product strategy and roadmap and is the technical lead behind Scandit’s patented Barcode Scanner technology. Before founding Scandit, he was the Associate Director of the Auto-ID Lab at MIT and a member of the MIT research team that developed the RFID technology which is today in use in major supply chains. Floerkemeier also co-founded Fosstrak, the leading open-source RFID software platform that implements the EPC Network specification. He was the technical program chair of the Internet of Things Conference in 2008 and IEEE RFID 2009 and general chair of IEEE RFID 2011. Christian received a PhD in Computer Science from ETH Zurich and a Bachelor and MEng degree in Electrical Engineering from the University of Cambridge.
