It is no secret that many consumers now opt for the self-checkout (SCO). In fact, 66 percent of US consumers prefer the method over staffed checkouts, and the global self-checkout system market size is projected to grow from $5.64 billion in 2024 to $18.01 billion by 2032. It isn’t hard to imagine why, either; SCO offers a new level of convenience and caters to customers who are in a hurry or seek a solo experience, not wanting to interact in that moment with cashiers or clerks.
But every rose has its thorn. With SCO being a high-growth point-of-sale (POS) option, it is also increasingly a popular part of the store where shoppers are committing fraud—whether on purpose or accidentally. Over 20 million Americans have committed theft at the self-checkout, with nearly 9 million indicating that they would do it again, according to Capital One Shopping Research.
Fraud can also be unintentional; 21 percent of respondents in a LendingTree survey reported having accidentally taken an item at the SCO, with 61 percent deciding to keep it anyway rather than returning it.
Without scaling back on a method that consumers are enjoying, how can retailers limit fraud, even the unintentional kind? One potential solution is computer vision.
You can think of computer vision as simply pattern recognition. By combining camera-enabled computer vision technology and artificial intelligence, a strategic SCO can identify visual patterns in an item to verify its identity—rather than identifying it by a barcode—with accuracy above 99 percent. SCOs can also now make use of visual deep learning, an advanced AI model that offers enhanced cost efficiency, energy savings, and stronger data protection.
Computer vision can reduce theft at SCOs in a myriad of ways, including its ability to detect hand movement, validate items, or verify age requirements. Let’s walk through the advantages.
Detecting Hand Movements
With camera recognition, the SCO can identify patterns in something as small as hand movement, determining whether a shopper is deliberately trying to bypass the scanner.
What’s more, once an alert is triggered, the consumer will be prompted to self-correct the issue. And only if necessary will a staff member be made aware. If they are, the staff member receives a short video snippet showing exactly why the alarm was raised—both the consumer and staff member view the video to identify the item and action in question, allowing the issue to be fixed immediately and transparently.
By keeping video snippets on a need-to-show basis, the SCO process becomes less staff-intensive and gives the customer the chance to self-correct before staff intervention is required. This also helps maintain the relationship between the store and the customer, as they can avoid a potentially embarrassing situation that might otherwise deter them from returning.
Validating Items to Ensure Accuracy
A common form of fraud at the SCO occurs when customers falsely identify an item—usually a premium product—as a less expensive alternative. This is especially common with produce and bakery items, which are easy to “accidentally” mix up.
Take apples, for example. Let’s say a customer wants to purchase a Fuji apple but only wants to pay the price of a less expensive Gala apple. At the SCO, they may select “Gala apple” on their screen—whether intentionally or not—and leave the store with the Fuji apple.
If a computer vision system is in place, it can visually recognize the subtle differences between the two apples, reducing or eliminating the need for manual selection altogether. In fraud prevention mode, the system can flag a discrepancy, prompt the customer to self-correct, or require employee intervention before completing the transaction. This allows the shopper to correct mistakes while also protecting the store from intentional fraud and deterring future misuse.
Reducing the Effort to Ensure Age Requirements Are Met
Retailers are required to verify that customers meet age requirements for restricted items, a task that is time-consuming and often delays checkout.
Here’s where computer vision can help. Instead of requiring staff to intervene for every alcohol purchase, AI-driven age verification technology can pre-screen customers, flagging only those who may need an ID check. This reduces staff workload, speeds up transactions, and maintains a secure yet convenient experience.
Protecting Both Your Stores and Your Customer Experience
At the heart of every retail strategy is the customer—and the good news is that fraud prevention doesn’t have to come at the expense of customer satisfaction.
By leveraging the right technologies, retailers can streamline the self-checkout process while minimizing theft, preserving speed and convenience, and protecting profitability. It’s a win-win.
Michael Jaszczyk is the CEO of GK Americas, where he works to maintain and enhance the company’s global reputation as the supplier of one of the most innovative and complete retail software platforms and suite of services. Jaszczyk has been a part of GK for more than 12 years, previously serving as CTO. He draws on an extensive wealth of experience, both in software development for the retail sector and as a manager at international IT companies, including MCRL AG, Pironet AG, and SA2 Retail AG. GK Software provides a future-proof foundation to support retailers’ customer engagement strategies.
