Self-Checkout is a Must-Have for Retailers: Advanced Fraud Detection Can Further Improve Benefits

Shoppers have always valued time and convenience, and the last thing they want is to wait in a line. Offering a number of checkout options, such as self-serve kiosks, is critical to ensuring consumers can be in and out of a store quickly.

However, after the introduction of self-checkout, retailers commonly face increased fraud and loss. Between entering the wrong PLU, skipping scanning altogether, or scanning a barcode from an item of lesser value, 40 percent of retailers said the largest increase in fraud occurred in their stores. What’s more, nearly one in five shoppers said they have stolen something from a self-checkout lane.

Artificial intelligence-based algorithms can help improve fraud detection processes. Currently, retailers are using standard, non-AI-based algorithms and data analysis to reduce fraud in their self-checkout lanes to trigger re-scans. However, it can be difficult to maintain rule sets to ensure only fraudulent activities are flagged. As a result, grocers often force customers to re-scan unnecessarily. This has a negative impact on the customer experience — and also requires expensive staffing.

With advanced, AI-driven fraud detection, retailers can improve accuracy in their self-checkout processes, expanding profitability and benefits for both themselves and consumers.

Analyzing Customer Behavior Can Identify Outliers

Modern, AI-driven fraud detection practices monitor customer behavior as they go through the self-checkout process. The AI ingests accurate and robust data at the point of sale and creates an analysis that can determine suspicious transactions. Then, triggers are created when a purchase needs to be audited. These triggers appear when there are outliers in the data.

Examples of outliers analyzed include:

• time it takes to scan;
• scanning speed;
• number of items;
• number of item types;
• item weight; and
• value total.

For example, AI algorithms can flag when the scale determines above-average weights for produce. Grocers refer to this as “the banana trick,” where shoppers key in the SKU for a lower-priced item while placing a higher priced item into their bags. When this happens, an alert is then sent to self-checkout attendants to resolve the issue. With improved algorithms, retailers reduce unnecessary triggering of a visit from a self-checkout attendant during an otherwise contactless journey.

AI Algorithms Allow for Continuous Learning

An important factor that advances fraud detection is the ability to implement self-learning features. The AI system automatically adapts to new fraud strategies based on programmed and learned best practices. In other words, after a purchase is made or a re-scan is triggered, the system improves its algorithms for future purchases.

For example, every time a customer scans a box of cereal, the transaction provides new data pertaining to accurate item weights, scanning times, waiting times between items, typical basket sizes, and item combinations. With continuous learning models integrated into fraud detection, the process is always improving and allows retailers to prepare for future scenarios and audits. 

Advanced Fraud Detection Can Reduce Human Error Risks

No solution is perfect when it comes to customer service, as human error must always be taken into consideration. As a result, self-checkout will always bring some level of risk. However, the benefits far outweigh the risks.

Shoppers have come to expect this level of convenience and autonomy in stores, and it also reduces the time spent waiting in line. What’s more, retailers can benefit from labor savings and can re-allocate store associates to other customer service roles. By implementing AI processes into fraud detection, retailers can reduce the number of re-scans to improve the overall customer experience, while also increasing accuracy in suspected fraud cases to ensure their bottom line is protected.

Michael Jaszczyk is the CEO of GK America, a leading omnichannel solutions company.