The Future of Fraud Prevention:

No More Binary Policies

By Pedro Ramos, Chief Revenue Officer, Appriss Retail

With shrink and retail loss on the rise, what should retailers do to stem the tide? Eliminating binary policies and increasing the use of AI within loss prevention strategies are two key ways as retailers take serious steps toward developing stronger fraud prevention plans.

According to a report from the National Retail Federation (NRF) and Appriss Retail, shrink totaled $112 billion in retail losses in 2022, up from $93 billion the year before. Additionally, the report added that theft represented nearly two-thirds of total shrinkage and that fraudulent behavior is becoming more sophisticated. 

Case in point, in 2023, card-not-present fraud accounted for nearly $9.5 billion in U.S. losses, and incidents of return fraud and abuse surged from 10.2% to 13.7% in 2022, per an NRF report. Retailers sit at a pivotal point and must address fraud prevention today to protect tomorrow’s profits.

Loss prevention policies can be tricky. On one hand, retailers need to do all they can to protect goods and profits but, they also don’t want to create policies that ostracize loyal shoppers.

Oftentimes, binary policies, or strict rules-based regulations around purchases in-store and online, as well as returns processes, can upset loyal consumers. Enforcing a limited returns window or not accepting a return without a receipt are examples of binary policies.

The rules don’t factor in subtle nuances with each transaction. A hardline policy treats all returns and transactions the same, but that policy can alienate some loyal shoppers by lumping them in with consumers who exhibit red flags. 

Similarly, when retailers put controls in place that stop a transaction at a staffed lane due to a rules-based policy, they frequently just cause inconvenience to the shopper. What is meant to be a control — such as requiring a manager to review a purchase — usually just ends up slowing down the buying process and not stopping fraud.  

For retailers looking for fraud or other shrink-related issues, the initial work begins with reading through tons of data for anomalies that are indicative of fraud or other margin-eroding issues. Exception-based reporting (EBR) helps streamline the process and identify problems early. 

One challenge is the level of expertise required to get the maximum value from an EBR system. Another challenge is the application remains largely dependent on the user’s bias toward certain behaviors, ignoring emerging behaviors that may be far more valuable.

Today, AI-enabled EBR systems take the process to a whole new level. For example, where the application may identify issues leveraging AI, it presents the user with an additional set of correlating behaviors that shed more light on the problem. 

Consumers engage with retailers in an omnichannel world. So why are retailers implementing single-channel solutions which provide a limited view of the consumer and the associate to stop omnichannel issues? To stop fraud and abuse in all channels, the solutions need to integrate all online and in-store, or integrate with other solutions that fill in the gaps.

In the EBR example above, the retailer needs to integrate in-store, digital, and other transactional systems into the application to leverage the full benefits that AI can deliver. Integrating only the point-of-sale system limits the value the application can bring and leaves doors open. Another example: video AI applications are very effective at preventing self-checkout fraud, but they, too, should be integrated into the EBR system to provide the nuances that may inform future deployment strategies or design. Using an EBR application as the control point, retailers remove all barriers from the transaction, improving labor utilization and the consumer experience.

The next phase of fraud prevention is in the use of AI for real-time applications. Reporting solutions are effective at identifying trends and behaviors that require attention, but the real loss prevention comes when retailers can stop them as they happen.

Refund fraud and abuse continues to grow across all channels along with claims and appeasements fraud. These can be addressed with applications that use AI to create consumer-specific risk models that will stop the transaction in real-time. This not only eliminates the loss at its root cause but also provides a better experience for the in-store associate and consumers.

AI can adjust to different types of behaviors and adjust based on the retail vertical. For example, wardrobing is conducted more often with fashion retailers. Just like there are several forms of fraud, consumers vary, too, with some posing more challenges in-store, online, or at specific types of retailers. There are nuances to understand to best protect against fraud. This provides retailers with the ability to better understand a consumer’s behavior and provide the best experience while controlling risk.

AI is at a point where it can provide real-life applications and drive value. As it matures and expands to a broader set of use cases, the retailer can expect better results in all areas of the business as a result. But for that to happen, the foundation of good data, integrated technologies, and strategic planning needs to take place. Otherwise, the AI applications will be just another point solution.