Is video analytics the best solution for retail loss prevention?

Preventing theft, breakage, or wastage is critical to any retailer. But this is even more important in the current scenario when the pandemic has hurt economies, and eCommerce threatens to kill brick-and-mortar stores. Realizing this, several video analytic companies are highlighting the advantages of using their solutions to help identify incidents like theft, either to prevent them or for investigative purposes.

This article explores how video analytics helps in loss prevention, how retailers can use them best, and their limitations.

Advantages of using video analytics for loss prevention

These days, most retailers have video surveillance cameras installed in different parts of a store. Traditionally they have been used for monitoring purposes. Some retailers are now willing to explore running analytics on the footage for security and business intelligence purposes. However, many are still not sure how to go about doing it. Naturally, systems integrators do have a role in making the customer understand all this and hence it is important for them to have a clear understanding. 

1.Analytics at POS

One of the primary use cases that we see for video analytics, specifically for loss prevention, is to refine and streamline video searches to decrease investigation time. This is especially true of incidents that happen at POS. But looking for a specific incident of theft, fraud, or suspicious behavior at POS could be like searching for a needle in a haystack. This is where analytics would come into play.

“If you’re using a unified system, you can also leverage motion detection or object detection type of video analytics, to search for all refund transactions that are made by the cashier where no motion is detected on the other end,” explains Kerri-Anne Beech, Industry Marketing Manager at Genetec. “So, you’re really leveraging information that goes beyond the POS data to filter down what would be searching through hours and hours of video to detect theft or suspicious behavior in a few clicks.”

2.Analytics at perimeter

Retail stores often have intrusion detection systems that will alert you if someone’s trying to break into the store. But some retailers can also leverage the power of analytics to be more proactive and be alerted before an incident.

“This is often known by different names perimeter detection, intrusion detection, etc., where you can have a specified zone to monitor,” Beech continued. “So maybe for a warehouse, it’s a fence line. For just a typical retail store, it could be the area right in front or the parking lot, and you’d be notified if someone breaches that zone. This is typically overnight when there isn’t supposed to be someone coming in.”

Beech added that they had seen more of such incidents at the beginning of the pandemic when stores had to remain closed for the entire day due to lockdowns. Retailers had to monitor their stores around the clock, and there were many more break-ins happening at that time. Video analytics became an extra layer of protection that they could use in such situations.

3.Analytics within the store

Amol Kulkarni, Vice President and Country Head of Dragonfruit AI, pointed out that footage from surveillance cameras placed within the stores is also an excellent raw material for analytics to work on. The software can take video footage from existing CCTV infrastructure and provide crucial insights like occupancy, movement patterns, dwell times, and engagement metrics to help retailers focus loss prevention resources optimally.

Best practices for maximum advantage

Systems integrators can follow certain steps to make the most of any video surveillance solution. According to Sean Foley, Senior VP at Interface Security Systems, the first step is ascertaining each customer’s unique challenges.

“Is the customer experiencing an internal shrink problem with their with their current employees?” Foley said. “Or are they dealing with organized retail crime from the outside, like robberies? Or is it just general customer disturbances? Or are they just looking to gain more valuable knowledge about customer behavior? We begin with those challenges, ascertain what is the most pressing to the customer and go from there.”

Once you figure that out, there’s the question of architecture, like where the processing power will come from. Customers have a choice here. There’s edge-based analytics that reduces the bandwidth consumption with the analytics running on the edge device itself. But the downside with this is that you need specific cameras. For a server-based approach, the processing power is on the server. If you don’t want to rip and replace your system altogether, you can leverage existing cameras and apply the analytics on the server-side.

Challenges and limitations

As a systems integrator, when starting a project with video analytics, it is essential to manage the expectations to be sure that you’re using it for the right reasons. Therefore, you need to know the problem that you’re looking to solve and not use it as a solution in search of a problem.

1.Size of the store

The size of the site where the video analytic solution is to be deployed makes a difference to the overall result.

“Something that’s important to consider is that enterprise analytics are very different than smaller deployments,” Beech said. “So, you need to be sure that your security system is built for an enterprise type of project, meaning that it can manage multiple sites with large numbers of sensors, including failover, different things like that.”

2.Variations in accuracy  

It is also essential to know that accuracy levels could fluctuate depending on the conditions. Especially because in retail stores, there’s always so much movement that can affect accuracy. Hence a retail analytic solution is not something you can set and forget.

“For the best accuracy rates, you really need to adjust it continuously; you need to look at what you’re getting, and if outside factors are impacting it,” Beech continued. “Because it can be affected by so many things like lighting, positioning, resolution, and quality. You need to make sure that you’re checking up on all that. And, of course, you need to define your parameters properly. If it’s too wide, you’re going to get false alarms and so many different notifications, which can lead to alarm fatigue, and if it’s too narrow, you could miss things. So, it’s crucial to just continuously work on it until you make the most out of it for your scenario.”

3.Technical concerns

Foley explains that although video analytic solutions are advancing rapidly, the infrastructure to process them may not be available everywhere. You can have high-end solutions that claim to provide maximum accuracy, but you won’t be able to take advantage of them if your network cannot support them.  
“These high-end solutions often require more broadband and more capable networks,” Foley said. “The challenge that we do come across at times is that a customer wants to take a very robust high-end video solution that requires a lot of broadband and a lot of network capability and install it at a location that currently does not have quite that level of broadband capability. Sometimes achieving everything that a retailer wants out of their video analytics solutions requires some level of network upgrade.”

Conclusion: is your store protected against loss prevention?

Ultimately, retailers need to have multiple tools at their disposal to help with loss prevention, according to Dragonfruit AI’s Kulkarni. Some of these tools help minimize or prevent losses and shrinkage before they occur, while others help analyze incidents post-facto. 

But it is essential to begin with a customer’s problem when approaching a project while setting the right expectations. Although video analytics is a sophisticated system that has the potential to provide a lot of benefits, they work best with the proper infrastructure and a continuous reviewing system.