OPERATIONAL EFFICIENCY

Retailers face a myriad of challenges, from competing with online sellers, attracting and retaining workers, to creating seamless customer experiences. But one issue continues to escalate: retail loss prevention, particularly shoplifting.

“I’ve heard everything from shoplifting doubling over the last 12 months to quadrupling over the last month,” says Matt Redwood, Vice President of Retail Technology for Diebold Nixdorf, a retail and banking technology company. “Regardless of the magnitude, it’s a really complex problem because the types of offenders and their methods vary widely.”

Some thefts are opportunistic, motivated by rising inflation and the impulse to steal small items. Then there are professional shoplifters targeting specific stores and brands. And some are driven by desperation, with people stealing to feed themselves or their families due to financial hardship. It’s also important to remember that not all losses are intentional; customers might accidentally forget to scan items or experience payment processing errors.

With such a diverse range of factors contributing to retail shrink, how can retailers address the issue? “Retailers are having to level up in regard to the technology that they use, the sophistication of that technology, and optimization of that technology across the store and its employees,” Redwood explains.

Combating Retail Shrink at Self-Checkout

One of the most obvious places to start is at the self-checkout area. Here, it is presumed the easiest place to steal since it is technically unstaffed.

To protect this area, technology solutions like Diebold Nixdorf’s Vynamic Smart Vision come to the forefront. This AI-powered system detects potential theft or unintentional errors by identifying behaviors such as missed scans, item manipulation, and product concealment.

Once an incident is flagged, the system can alert store staff or directly prompt the customer to correct the issue. This proactive approach is particularly effective in addressing unintentional errors where a customer may be distracted and leave items under their cart.

“What we’re finding is that with the use of customer nudging—such as audio or visual alert that says, ‘There’s items left in your basket, do you want to scan those?’—about 85% to 95% of customers will self-rectify,” Redwood explains.

If the customer does not do so, the solution provides store staff with detailed information about the incident, including what happened, how it happened, and where it happened so they can decide how best to intervene.

To effectively implement a system like the Vynamic Smart Vision, Diebold Nixdorf recommends installing a dedicated fixed camera above each self-checkout device rather than using existing CCTV networks.

“Technology is constantly evolving, with equipment being moved around stores, new units being added, and the layout of POS lanes and self-service areas changing,” says Redwood. “Every time the camera’s position relative to the checkout area changes, the AI requires retraining. By using fixed cameras, deployment is much quicker and easier, even in dynamic store environments.”

“#Retailers are having to level up in regard to the #technology that they use, the sophistication of that technology, and optimization of that technology across the store and its employees.” – Matt Redwood, @DieboldNixdorf via @insightdottech

CCTV networks can still play a valuable role by providing a secondary viewpoint of the self-checkout area and the overall solution. This additional data can enhance the solution’s accuracy through data triangulation.

Intel is crucial to the success of this solution. According to Redwood, whether a business opts for an attended POS, self-checkout, or kiosk, each operates on a single Intel platform. By harnessing Intel technology, devices not only gain sufficient computing power but can also execute several use cases directly on the solution itself.

“Our partnership with Intel unlocks many retail benefits. Not only does it help retailers reduce the expense and maintenance associated with large, space-consuming servers but it also enhances energy-efficiency through lower power consumption,” says Redwood.

Retail Loss Prevention Across the Store

Redwood envisions a broader application for Vynamic Smart Vision beyond self-checkout areas.

“We are only seeing 30% of total store shrink happening at self-service checkout, which means the rest of the shrink is happening elsewhere in the store,” he explains. “This is a ‘whack-a-mole’ type journey where we must explore one loophole and then try to limit that loophole as much as possible with technology. That will either stop shrink from happening or it will drive malicious theft to other areas of the store, and then we will have to deploy the technology elsewhere.”

Diebold Nixdorf is developing a comprehensive AI-powered solution that tracks shrink and customer behavior throughout the store, including aisles, entry, and exit points. This technology can be integrated into any device with a camera, creating a connected store where data from various sources feeds into a central AI model.

“We can identify suspicious actions in the aisles, follow the customer to the checkout, and intervene at the appropriate moment, such as when we detect concealed items,” Redwood emphasizes.

In the future, a scalable solution will be essential for retailers to experiment and learn while building a robust, store-wide AI system capable of handling multiple endpoints and use cases.

This article was edited by Georganne Benesch, Editorial Director for insight.tech.