According to 68 percent of retailers, the risk of external theft has risen in the last year. And recently, we’ve seen big name retailers reporting losses due to organized retail crime (ORC), fraud, and abuse. The recent National Retail Security Survey from NRF, the Loss Prevention Research Council, and Appriss Retail noted that 28 percent of retailers were forced to close store locations this year due to these harmful instances. What are retailers doing to mitigate this growing problem? And why isn’t it working?
Many are investing in visual AI or radio frequency identification (RFID). Visual AI gives retailers an “extra set of eyes” by automatically reviewing potential criminal or fraudulent activity captured by cameras throughout the store. Similarly, RFID monitors the movement of merchandise and can detect when an item was stolen.
In fact, the same survey found that 17 percent of retailers are piloting RFID and 31 percent are researching or testing RFID identification tagging. Additionally, the report stated that CCTV and video systems were common additions to retailers’ loss prevention strategies in the last year.
Both visual AI and RFID benefit retailers, but these technologies alone will not stop internal or external theft like ORC, or fraud and abuse. To circumvent the limitations of these technologies, retailers should implement a holistic approach that also considers data analytics from customers’ and cashiers’ behavioral patterns and anomalies.
The Shortcomings of Visual AI and RFID
Visual AI and RFID are great assets for mitigating shoplifting in the traditional sense. These tools have empowered retailers to take action when theft occurs; however, theft, fraud, and abuse occur in much more sophisticated ways today. For instance, fraudsters have access to YouTube videos and Reddit pages detailing how to remove RFID tags. And, when an ORC group relies on techniques like smash-and-grab, they may use masks, which completely undermine the capabilities of visual AI.
In addition to these blatant examples of theft, visual AI, and RFID also fall short of mitigating less flashy scenarios of fraud and abuse. For example, an employee may frequently give their friends and family discounts or intentionally ring up an expensive steak as if it were a bag of chips. This is called “sweethearting” and would go undetected by visual AI and RFID.
Even if these solutions do detect an issue, it may be a false positive. A false positive could stem from an accident on behalf of the cashier or the technology. Even if the error is immediately corrected, this momentary lapse may necessitate manager intervention, which can interfere with a customer’s shopping experience.
Imagine a new cashier confuses one vegetable for another when manually ringing up a shopper’s basket. The visual AI-driven checkout system may alert the store manager of fraud and require that they intervene. If left uncorrected, this innocent error on the cashier’s part might become a pattern, slowing down customers, and deterring their loyalty over time.
To navigate these constraints, retailers need to add the missing piece to the puzzle by integrating data-driven exception-based reporting (EBR) that can locate patterns and prevent theft, fraud, and abuse before they occur.
Data-Driven AI Is the Missing Piece of the Puzzle
The key to preventing theft, fraud, and abuse is to anticipate these scenarios better. With data analytics, retailers can monitor potentially suspicious trends coming from individuals and larger groups both internally and externally. For example, ORC often occurs in patterns. These groups may target a single retailer for months, committing the same type of fraud repeatedly from different store locations or while using different credit cards. With data analytics and EBR, this pattern can be flagged early, allowing retailers to increase in-store security or alter online policies for certain shoppers.
Similarly, EBR can also catch internal instances of fraud and abuse. The aforementioned cashier, who was prone to “sweethearting,” would be identified by the EBR system because of the ongoing pattern. Then, the retailer could contact the cashier and remind them of the policies surrounding discounts or revoke this privilege.
However, by itself, data-driven EBR has some limitations as well. This tool doesn’t have visual context to apply to an act of theft, fraud, or abuse. Yet, when combined with visual AI and RFID, retailers benefit from loss prevention with true 360-degree coverage.
A Holistic Approach to Optimizing Loss Prevention
The combination of visual AI, RFID, and EBR is powerful because it can automatically analyze every perceived instance of fraud, theft, or abuse with the background of overarching contextual data.
With the previous example about a new cashier who consistently mis-scans vegetables, the visual AI tool might flag the issue, but the data-driven EBR tools would have the appropriate context to note that the cashier has made this mistake several times. As a result, the retailer could provide the employee with a training refresh to better prepare them to work the checkout counter. This simple action can save the retailer from unnecessary losses, help the cashier feel better supported, and strengthen the customer experience.
Similarly, imagine a shopper was caught on camera removing RFID tags, stealing the now tag-less item, and returning it for store credit. Using the EBR capabilities and a real-time decision engine integrated into the point of sale, the retailer can deny the use of the store credit because the retailer has used these joint loss prevention capabilities. This denial might be the only action needed to deter that shopper from trying to defraud that retailer in the future.
Visual AI, RFID, and data-driven EBR all have limitations, but together, they form a robust, high-confidence system that can stop a multitude of fraudulent and criminal activities.
A Sophisticated Mix of Solutions for an Ever-Evolving Problem
With the threat of theft, fraud, and abuse growing, it’s more important than ever that retailers take a holistic, sophisticated approach to loss prevention. By establishing themselves as businesses prepared to deal with these scenarios, retailers can deter future problems and safeguard profits. When used together, data-driven EBR, RFID, and visual AI create the holy grail of loss prevention for today and tomorrow.
Dr. Vishal Patel is the chief technology officer for Appriss Retail. In this role he is responsible for all engineering functions, building and supporting the product suite, and helping bring new products to market that better serve Appriss Retail’s customers. He has more than a decade of software and programming experience working with a variety of different technology stacks, database solutions, and cloud providers. Previously, as the director of data science R&D for Appriss, he was responsible for building and deploying data application and machine learning models for retail customers as well as customers of Appriss Health and Appriss Insights.
