CASE STUDY: Using AI to Detect Fraudulent and Abusive Employees

Artificial intelligence tools are now becoming a seamless part of our everyday lives. Just ask Alexa. Or Siri. Or the customer support chatbot for Lyft, Spotify, or Whole Foods.

It stands to reason that retail loss prevention should be leveraging this new, valuable technology for its own toolkits as well.

Technology beats the bad guys: Find out how by downloading our FREE Special Report, Retail Technology: Electronic Article Surveillance, RFID Inventory Management and Trends in Omni-Channel Retailing

That’s just what Appriss Retail did in a recent partnership with The Bon-Ton department stores. The partnership aimed to detect fraudulent and abusive employees and collusive organized retail crime (ORC) in stores via the application of a complex machine-learning algorithm to five years’ worth of transaction data and employee files. The case study is detailed in a sponsored article in the October 2018 issue of LPM Online by David Speights, PhD; Vishal Patel, PhD; Alan Gillette, PhD; Sruthi Shyamsunder, MS; Adi Raz, PhD; and Ping Yang, PhD, with Appriss Retail. From the article:

To detect employee deviance, Appriss Retail identified more than 3,000 variables that describe an employee’s pattern of activity. The resulting model was applied to recent employee activity, and many cases were recommended to The Bon‑Ton as candidates for further investigation. Two rounds of testing were performed to evaluate the system’s ability to identify previously unidentified cases and to determine the quality of the cases being provided by the model. There was at least a 25 percent increase in fraud‑related terminations above what would normally be captured by EBR alone and a 25 percent decrease in time required to track down the fraudulent activity within many of the cases.

Check out the full article, “A Department Store Case Study of Artificial Intelligence in LP,” to discover the results of the modeling research and a more in-depth examination of the use of dynamic learning and feedback when it comes to detecting suspicious behavior.

If you’ve missed any of our previous LPM Online editions, go to the Archives. Be sure to become an LPM digital subscriber, so you are the first to know when new issues are available. If you haven’t already, SUBSCRIBE NOW for free. If you would like a free subscription to our award‑winning print magazine, go to LPMsubscription.com.

Comments

Leave a Reply

Enter Your Log In Credentials
This setting should only be used on your home or work computer.

×

Send this to a friend