Cameras at the point of sale have become a routine feature of brick-and-mortar stores. They help ensure that items do not get stolen or accidentally passed through checkout lines without being scanned. But the “tale of the tape” by itself is not always obvious when it comes to detecting and preventing loss at the POS.
“I’ve been a cashier, so I know a cashier can miss something,” said Doug Haworth, director of loss prevention, Woods Supermarket. The Bolivar, Missouri-based independent chain, which operates nine supermarkets and two convenience stores, has an honesty and integrity program for all cashiers. Even so, when Haworth saw a demonstration of the ScanItAll computer vision solution from Stoplift Checkout Vision Systems at an industry conference in the first quarter of 2011, it caught his interest.
Driven to Distraction
Woods Supermarket tested the ScanItAll server in one store, connecting it to the existing POS camera system. It started getting results almost immediately. The system uses mathematical algorithms to analyze video of cashiers handling items at a pixelated level, detecting motions that indicate an item was not properly scanned, such as covering a bar code with a hand. It also analyzes items in a grocery cart, including hard-to-see areas such as underneath the carriage, to see if they have been scanned.
In the first 10 days of the pilot, the retailer detected 40 items that had not been scanned and would not have been detected through cameras alone.
“There was no ‘sweethearting’ [employees purposely not scanning items] or customer theft,” said Haworth. “Cashiers were distracted talking to customers and ran items, not noticing whether or not they beeped, or the bagger would be in ‘bagging’ mode and grab whatever items were on the conveyor belt and put them right in the bag before they could be scanned.”
Success Breeds Success
After running the initial pilot for 30 days, Woods had already saved a significant amount of money in recovered losses at the POS, and decided to implement a full rollout across the chain.
“We got the most hits in the first three months of installing ScanItAll in a store,” said Haworth. “Newer cashiers didn’t realize they had to slow down and look at the register. We calculated that we would achieve ROI within two months of implementation.”
As part of the solution, Woods also has a secure Web 2.0 interface running a live feed from its NCR Advanced Checkout Solution POS server to a remotely hosted video analytics platform that matches up the number of items detected passing through the checkout area to the number of items that have been scanned. Woods can obtain store-level performance reports and also view checkout video and receipt data, broken down by individual incidents at individual stores.
Since the initial rollout, Woods Supermarket has compared the data it gets from ScanItAll with data from its existing Shrink Trax POS exception reporting tool. And in the one store where the retailer runs a self-checkout terminal, it has implemented StopLift Self Checkout Accelerator that is designed to immediately flag unscanned merchandise and alert the attendant.
“We had eight cases in the past year where customers were caught not ringing an item up,” said Haworth. “If they showed intent, we called the cops.”