Self-Checkout Doesn’t Stand a Chance Against Fraud. Now What?

If you're only now considering methods to reduce the growing rates of self-checkout (SCO) fraud and theft, you might already be two steps behind.

According to the 2023 National Retail Security Survey Report sponsored by Appriss Retail, National Retail Federation (NRF), and Loss Prevention Research Council (LPRC), 68 percent of retailers said external theft was more of a concern than it was in 2022. Recent commentary by retailers suggests that a significant portion of this theft occurs at the SCO kiosk. In fact, Lending Tree found that 15 percent of SCO users admitted to purposefully stealing from the unmanned kiosks, and 44 percent said they’d likely do it again.

Getting rid of SCO altogether isn’t an option given the growing demand from consumers — 53 percent of young Americans prefer self-checkout. Meanwhile, many retailers are struggling to adequately staff stores amid ongoing labor challenges, fueling the need for SCO.

In response to the increase in shrink, retailers are taking control measures — from limiting basket sizes at SCO to requiring a membership to use unmanned kiosks. Unfortunately, these strategies may hinder the customer experience and decrease efficiency.

To avoid issuing resolutions that disrupt the shopping journey, retailers can turn to artificial intelligence-based fraud detection targeting SCO and the full shopper profile to mitigate risk. Revisiting standard operating procedures and reinforcing associate engagement and best practices will also go a long way.

However, before discussing the solutions, let’s take a step back and identify why SCO fraud is so difficult to detect and deter.

Self-Checkout is One of the Hardest Places to Detect and Deter Fraud

The SCO kiosk is one of the hardest places to monitor because the entire transaction is in the hands of the consumer. Even with some technological controls like scales and video cameras, retailers struggle to get a complete picture of what consumers are doing at this touchpoint. For example, some shoppers might swap price tags on items with similar weights, or they might fake scan an item and put it immediately into their bag. And, even if a retailer sees these actions on video, it’s after the fact and there's little to be done. If the bad actor doesn’t leave an identifier such as a loyalty card, then it's nearly impossible to track them down.

To combat these issues, major retailers have tried diverse methods to mitigate the risk of SCO. Target, for instance, limited the self-checkout kiosk to be used only for 10 items or less and converted more SCO lanes to traditional, staffed lanes, limiting the risk and making it easier for greeters to validate orders on exit. Meanwhile, Walmart announced that in high-risk stores, it will limit SCO usage to Walmart+ customers only. This approach is betting that an identifiable customer will be less likely to commit SCO fraud.

These methods may deter fraud, but they will have an impact on customer satisfaction and increase labor costs as transactions are migrated to staffed lanes. Other strategies exist that are helpful and minimally intrusive to the shopper.

Targeted AI Helps Identify and Mitigate Fraud at Self-Checkout and Other Critical Touchpoints

Video AI applications are helpful in analyzing consumer behavior in real time and provide the SCO attendant data and video information as they deal with SCO alerts. This is better than the typical scale approach because it allows for higher transaction speed delivering a better experience to the consumer and efficiencies to the retailer. The downside is implementation and ongoing support costs of the video infrastructure. This area will continue to improve in terms of accuracy and cost reductions over time. In the end, however, it still comes down to the SCO attendant following the process for alerts.

Another form of AI is on the data side. Using advanced analytics retailers can link SCO, shrink, attendant, loyalty and video data to identify trends and patterns that will lead to more targeted decisions. The data applications will provide the trends and anomalous transactions, enabling analysts to validate the AI assumptions using video linkage. Once verified, policies, procedures and programs can be created targeting specific stores, attendant training, and high-risk consumers.

Combining the two applications will deliver even further insights. Using the metadata from video analytics in combination with the system’s data will provide richer insights and result in better-informed decisions. Not only does this strategy help prevent intentional fraud, but it also can be used to identify process improvements and training opportunities.

Another area where SCO is driving losses is with third-party services. Particularly in grocery stores where the third-party acts like a consumer by shopping, purchasing and delivering the product to the consumer. However, these third-party services can create issues for the retailer, such as adding products to a consumer order for the shopper’s personal use or making mistakes that eventually become claims. Advanced AI solutions will identify bad consumers who file abusive or fraudulent claims and deny the claim in real time, stopping the problem in its tracks.

Advanced AI Can Bolster Security and Profitability at Self-Checkout

Self-checkout is here to stay. Consumers have come to expect it. It’s convenient, in many cases faster for the consumer and allows for the re-allocation of labor. The challenges SCO presents will continue and solutions are not one size fits all. Retailers will continue to improve the results from their SCO investments by creating programs that include advanced technologies, operational excellence, and even store design. SCO shrink challenges will evolve. The retailers that will succeed are the ones that take a holistic approach to solving the problem.

Pedro Ramos is the chief revenue officer at Appriss Retail, a company that reduces losses from retail fraud and theft while protecting the retail customer experience.