By Nathan Needel, PRD International
For years, surveillance systems have recorded thefts and other incidences, including those at POS location, for future investigation. Now, video can be used to stop incidences as they happen. To do this, users need to quit thinking of surveillance systems as simply repositories of what’s been seen and recorded and, instead, think of cameras as sensors in a system. Why should you have to search through recordings to locate the incident when, if fact, they can stop it as it happens?
For most retail organizations, any cuts in shrinkage will typically go directly to the bottom line. That’s because, as reported by the University of Florida’s Nation Retail Security Survey, the total shrink percentage of the retail industry in the United States is 1.52%. It’s budgeted for. Thus, any lowering of that figure adds to the profit line – and retailers operate on very thin margins.
To make sure that we’re all on the same wavelink, let’s agree that the shrinkage we are discussing is only that caused by employee actions at the POS (point of sale) terminal, not in the warehouse, at the docks or elsewhere. It’s the source for major shrinkage problems and not that difficult for employees to do. There are different ways to manipulate a POS system, such as a cashier giving customers unauthorized discounts, creating fraudulent returns, manually entering lower values in the system or making a no-sale, which means that the cashier opens the cash counter without registering a sale.
Traditionally, POS fraud is fought by surveillance staff monitoring a POS terminal or by manually searching through surveillance video recordings. Modern POS systems can have automatic alerts when specific exceptions are detected. Also, exception reports and listings based on employees, refunds and terminals are possible to detect with modern systems. Modern networked-based POS systems can also include network video to POS exception listings, giving quick access to detailed information of what has happened.
Today, most retailers still have to search through their recorded video to catch an incident. As they will tell you, it can be extremely time-consuming to view video in a search to catch a clerk pilfering. Even in cases where the retailer has associated video and text for faster searches, the incident is still discovered after the fact. In these cases, using high resolution cameras, extremely sharp digital images capture the detailed faces, numbers and small objects while also capturing the text that appears on the customer’s receipt. It then associates it with the corresponding video and records all on internal hard drives.
To investigate a suspicious transaction, loss prevention personnel enter receipt text, such as “no sale,” “void,” “return,” or certain brand names. The surveillance system retrieves the associated video and transaction text very quickly, eliminating the afore-mentioned practice of viewing hours of recordings. Personnel can search by register, time, date or motion in a target area and view the POS text as they review images. Although this is a much better alternative than formerly undertaken, it is still after-the-fact.
Camera + video analytics = greatly increased loss prevention results
There are a variety of ways that a camera, which is often already installed, can really add to its return on investment by acting like a sensor for the entire system with the help of video analytics to stop shrinkage incidents as they happen.
Here’s a very obvious example. Perhaps a retail jewelry store wants an alert whenever a certain object, such as the very expensive necklace in the front showcase, is moved. With a simple motion detector, if the necklace is moved, the system sends an alert in any manner that the store management wishes.
The system could also detect suspicious activity for immediate follow-up such as people entering low traffic areas that could be used for tag switching, undue motion around the POS stations and out-of-hours activity around the POS stations. At the changing room, the system can automatically compare what merchandise was carried into the changing room versus what is carried out.
Theft can involve cash, credit cards, gift cards or sweethearting, when employees give free or discounted merchandise to friends or family. This is accomplished through cancellations, including "as is" items, returned items, discounted items, void/no transactions or manual entries at the register. Sweethearting has been easy to accomplish at POS terminals and particularly difficult to detect – until now.
Using cameras as sensors, the system can quickly determine whether a person returning an item entered the store carrying that item. It works by detecting and tracking customers at entrances and customer service desks and associating the two events. Such a solution only requires cameras at the store entrances and returns counters, so it is simpler than an more customary approach where the customer is tracked throughout the store (requiring many cameras and very reliable camera hand-off algorithms) and must be continuously monitored to determine whether items are picked up.
Another video application could target employees that ring up a “return transaction” without a customer present. If so, the instant the return transaction occurs, the camera looks for a customer and, seeing none, alerts management. Many organizations would like to better enforce a two-person rule for such types of transactions. The system operates in the same manner. If two people aren’t present at the time of the transaction, an alert is instantly sent.
Besides catching incidents and helping the retailer immediately take action, such systems will also help the organization improve customer service at the registers. For instance, if company policy is to have no more than four customers in line, once the fifth customer queues up, the system sends an alert to the store manager to get another cashier at the station.
How to create such systems
In many cases, there are already legacy cameras and video systems in place. Nobody wants to tear out their old systems to add new features. Hopefully, such a system is based upon an open system so that the new video analytics will work with the present cameras, VMS and rest of the system. Hencefort