Cloud Computing

Combating retail theft & fraud


Shrink in retail, and more specifically, theft & fraud components, have long been accepted as a cost of doing business. In the last few years, we have seen both consumer and associate theft & fraud explode. To the point where big-name retailers have closed higher risk stores that have been deemed unsustainable, due to the levels of loss. 

Some retailers, determined to stay in a particular market have taken the approach of placing nearly all products behind plexiglass, which results in a truly horrible shopping experience for consumers, which cannot help flagging revenue and requires additional staff to open and re-secure displays. In a time where labor is one of the largest issues in the industry, working in an environment like that cannot help to attract or retain associates.  

Speaking at a retail industry event recently, I quipped that I need to stop referring to Loss Prevention, because it really doesn’t prevent anything. Instead, we should refer to the traditional approach as Loss Recording, which evoked chuckles in the room.  

Today, Loss teams may invest tens of hours reviewing video in an attempt to confirm a suspicion that something nefarious may have happened – it is costly and time consuming, and it has very little effect on the problem. Less than 1% of all video recordings are even reviewed, so retailers are paying to store all that content too. Loss teams are not immune to the labor challenges either – loss is tracking exponential growth and the staff to investigate is reducing, creating an ever-widening gap. 

Active loss & fraud detection

So how do we move from passive loss & fraud recording to active loss & fraud detection? While we will never eradicate loss, we can take a bite out of certain types of loss – Integrating SmartCameras running video analytics with the Point of Sale allows us to use AI to review higher risk transactions, such as returns, voids, refunds, and gift card activations, amongst others, and “look for” suspicious behaviors at the time and location of the transaction. For example, flagging these transactions if there is no customer present in front of the counter. Could there be cases where there may not be a customer right there at the time? Of course. However, if the system can identify the exact moment the suspicious transaction occurred and flag that for human review, this allows the loss teams to be more effective and targeted with the limited labor available. This approach can also be used to look for other actions, such as under-scan, by reconciling the number of items placed on the counter or belt with the number of on the POS T-log to look for discrepancies. 

While most retailers will not interfere when the theft is being perpetrated by a consumer, for obvious safety reasons, the intelligence it provides may allow the retailer to take mitigating actions, such as limiting self-check use during periods of the day or week, which have shown to have a higher loss rate, forcing consumers to use a staffed lane.  

While this will not prevent all theft and fraud, we can start to lessen the impact in some key areas and ensure staff is being effectively leveraged – any measurable reduction in what is a $100Bn problem, could significantly move the needle on this problem. 

 

Check out our use case on Portfolio Explorer for more information. 

Share: