Adam Jacobson,
VP of Operations
101 Continental Boulevard, Suite 1666
El Segundo , California , 90245
(310) 591-8688
adam@adengage.com

Description

AdEngage ( www.adengage.com ) is an online text-ad brokerage that displays over 250 million text ads per day on over 1400 websites. Payment is made for an ad to be displayed using credit cards or PayPal. Around 700 credit card transactions occur per month and over $80,000 are processed through the credit card gateway.

Credit card theft is common and ad purchases are frequently attempted using stolen credit card information. AdEngage delivers a virtual product so it is a natural target for criminals. The credit card processing gateway often does not detect any problem with the transaction as the true owner of the card is not aware that his or her card has been stolen. As a result AdEngage has to use a manual process to review all the credit card transactions and declines or investigates suspicious transactions. If the fraudulent purchase is not caught quickly then AdEngage has to pay the publisher for displaying the ad and refund the money to the real credit card holder. This can cost thousands of dollars per month. As AdEngage has grown the manual process of reviewing each transaction has become impractical.

AdEngage uses ColdFusion 7 and Microsoft SQL database on 15 HP ProLiant DL380 servers operating on Windows Server 2000 and 2003.

The project scope is to (1) develop new ways to identify credit card theft, (2) create scoring systems for the existing and new credit card fraud identification systems, (3) combine them into a credit card scoring system and (4) design a display to show the results in way that allows an operator to easily and quickly identify and refund fraudulent transactions.

Some of the challenges associated with this project are (1) creating algorithms to define the existing rules used to identify fraudulent purchases, (2) creating new rules to identify fraudulent purchases, (3) ability to compress the data down into summary information so that the system operator can quickly spot fraud and (4) creating a solution that can be maintained by the AdEngage staff.

Given the nature of this project certain information about the fraud detection algorithms can not be publicly posted on the Internet.