Adios, Sweetheart

It’s 11 a.m. on a Friday, and throngs of shoppers stocking up for the weekend are crowding the supermarket registers. The checker in lane six glances up and nods furtively to her friend as she guides a half-dozen items around the scanner. At the end of her shift, she’s called into the manager’s office, where the LP director is waiting: Fifteen minutes later she’s out the door for the last time.

This is a familiar scene for retailers around the country that are constantly dealing with sweethearting, the most pervasive form of employee theft. This activity alone is costing the retail industry an estimated $13 billion annually. But computer vision technology being tested by Hannaford Bros., Big Y Supermarkets and Safeway is proving to be a highly effective deterrent.

“We find that about 40 percent of cashiers at any given store are blatantly sweethearting,” says Malay Kundu, founder and CEO of StopLift Checkout Vision Systems, the Bedford, Mass.-based company that developed the technology.

“Internal theft is generally considered more of a problem” than shoplifting, says Tom Perkins, director of LP for Hannaford Bros. “There are more incidents of shoplifting, but internal theft can be far more expensive.”

Some LP and security pundits believe that as many as 30 percent of employees will steal at some point in their careers, and that as much as 75 percent of employee-related crime goes undetected.

It is being felt on the bottom line, however. A shoplifter may walk out with $100 worth of merchandise, but internal theft incidents “typically last for weeks and probably average in excess of $1,000,” Perkins says. “That’s why you have to have the right tools in place.”

Can’t ignore the problem
As with other retailers, employee theft is a sensitive issue for Hannaford Bros. The Scarborough, Maine-based chain of 165 supermarkets doesn’t want associates to feel as though they are constantly being watched, but can’t afford to ignore the problem.

“I think most retailers are now pro-actively employing exception reporting tools to identify instances of internal theft,” Perkins says. “It’s just a question of which tool works best and is easier to use. Don’t forget there are labor issues when it comes to staffing: None of us in this industry has unlimited resources to chase down everything that crops up.”

Retailers using the StopLift system are not constantly watching the video, Kundu says. Most prefer batched alerts on a daily basis.

Is there is a correlation between incidents of employee theft and the state of the economy? “I wish I had a dime for everyone who asked me that question in the last month,” Perkins says. “It’s natural for people to assume that theft will increase under these circumstances. Perhaps they’re right, but that may be too convenient.

“You have to look at how you’re operating, whether you are reducing labor or supervision in the store. And you have to look at the tools you have in place and whether you are catching more people because these tools are better than they were several years ago,” he says.

Perkins, who has been involved with LP at Hannaford for 20 years, says “you always begin with an investigation from a variety of sources. In the old days you received tips from store management, then most of us installed camera systems instead of having to watch people in real time. Then many of us began to employ exception reporting or POS data mining tools. So instead of receiving a tip from a store you had concrete data pointing you in the direction of people who might be involved.”

Perkins wouldn’t quantify just how much of a problem employee theft has become for Hannaford Bros. “I think we are seeing more incidents,” he says. “However, due to the maturity of loss prevention tools, we are catching them sooner.” Hannaford began testing the StopLift system in June 2007.

With most exception reporting tools, an associate “has to have been engaged in a certain amount of activity for them to rise to the top and for their activities to be distinguished from other cashiers,” Perkins says. “StopLift identifies them the first time they engage in sweethearting: you don’t have to wait for it to happen 25 times.”

StopLift is the brainchild of Kundu, who developed the concept while working on a retail LP field study at Harvard Business School. He eventually connected with Hannaford Bros. through a professor who used to sit on the chain’s board.

Analyzes human interaction
StopLift works with a store’s existing camera system to flag products that weren’t scanned. The heart of it is video recognition software that constantly monitors all of a store’s security video. By visually analyzing human interaction with objects, it determines what is happening in real time and immediately flags suspicious transactions for the LP department to review. The system also identifies the specific cashier, the date and time of the incident.

With a reputation as one of the most technologically progressive supermarket chains in the industry, Hannaford Bros. has an array of POS data-mining tools at its disposal, but it uses StopLift specifically for detecting sweethearting. “This was one area of the business that no one had seemed to nail down,” Perkins says.

“Most of us in the industry have camera systems that record activity [at POS], but we don’t have the labor to go through it all. This analyzes video and identifies situations like no other tool available.”

In the Hannaford Bros. stores in which it has been tested, he says, StopLift has “absolutely reduced sweethearting to zero.”

StopLift can detect when 10 items are placed on the belt “but only five are scanned,” Perkins says. “The labor savings is huge, and it really gets to the sweethearting problem in a way that no one in the industry has been able to.”

The system has also shown that sweethearting may be more pervasive than initially thought. “Prior to our test of StopLift, I would have told you the highest incidences might be for meat or cigarettes,” he says, “but we found that it was pretty much across the board.”