It may happen in a variety of ways.
A contracting officer awards a lucrative deal to a friend or relative. Several contractors who realize they're always among the winning bidders decide to compare notes and take turns splitting profitable contracting jobs among them. Or perhaps internal staffers collude with one or more outside businesses or individuals to subvert the competitive bidding process and gain unfair advantage.
However it happens, it's called fraud.
And in an organization the size of the US Postal Service, with its estimated 800,000 employees, 37,000 facilities, $70 billion in revenue, and $30 billion awarded in outside contracts annually, fraud spells big trouble. Cost is one issue; lack of resources for locating and investigating suspicious situations is another.
Enter Elder Research, Inc. (ERI), a data mining consulting firm the Postal Service commissioned to help identify fraudsters inside and outside its ranks. With the organization contracting out everything from trucking services to printing to telecommunications and even some property management services, the possibilities for suspicious activity are almost limitless, as Antonia de Medinaceli, director of business analytics and fraud detection at ERI, described during a presentation yesterday at Predictive Analytics World in New York.
"This is the quintessential needle in a haystack problem," she said. However, she added, ERI had focused on a solution that removes 90 percent of the hay, metaphorically speaking, of course.
Though being deliberately vague on details, Medinaceli discussed some capabilities of the Postal Service's customized fraud detection tool. Using more than 30 indicators to search a wide variety of data, the fraud detection tool flags and ranks instances of suspicious activity, allowing investigators in the Postal Service's Office of the Inspector General to decide which leads to pursue.
During her presentation, Medinaceli showed off a dashboard created by the fraud detection tool depicting a US map with 250,000 Postal Service contracts represented by circles of different colors and sizes. The color-coding helps immediately identify potential risk factor while the size of each circle signals the contract's dollar value.
In a separate presentation Thursday, Jennifer Boyce, senior manager at Deloitte Financial Advisory Services, discussed some of the key indicators used to flag potential fraud and other abuses in forensic analytics.
Deloitte uses analytics to predict likelihood of fraud, improper payments, and money laundering in a variety of organizations. It mines internal data as well as external data collected from blogs, social media, and other sources. Flagged activities often include instances of increased expenditure with decreased time put in on projects or an increase in the number of vendors or instances with no segmentation -- in other words, situations in which one person is in charge of all functions or vendors in a particular department.
Fraudsters, of course, are a tricky bunch, always modifying their techniques with the hopes of evading detection. With that in mind, Medinaceli noted the importance of regularly adjusting fraud detection tools.