Having expanded to more than 1,350 remote offices and 19,500 employees by 2007, Fidelity National Financial (FNF), a provider of title insurance, specialy insurance and claims management services, needed to streamline its manual auditing processes. "We couldn't efficiently and economically visit every office to conduct audits frequently enough," recalls David Riddell, FNF's associate VP and manager of automated audit solutions. "And we had no automated, centralized way to monitor activities within offices."
To reduce the business and regulatory risks, Jacksonville, Fla.-based FNF sought a continuous monitoring solution to proactively identify questionable transactions. As the core analytics engine, FNF selected AuditExchange by Vancouver-based ACL Services. "It was a proven platform that was deployed worldwide and had been adopted by the big four accounting firms," Riddell says, explaining the solution's selection.
In August 2007, relates Riddell, who was hired in part for his programming skills, he began familiarizing himself with AuditExchange and mapping out a development strategy. "The ultimate goal was creating a system modeled on credit card transactions, where real-time fraud protection occurs," he reports. "We decided to call the system 'FAST,'" for Fidelity Analytic Software Technology.
Next, Riddell divided system development into two phases. The first phase included building the FAST framework and rolling out internal corporate transaction monitoring. The second phase -- referred to by FNF as "HUDs," or Housing and Urban Development transactions -- consisted of external title and escrow-related transactions. "With multiple external feeds involved in real estate closings, numerous internal software systems and thousands of new escrow orders a day, it made sense to save HUDs for the second phase," Riddell says.
Riddell built the FAST framework from early 2008 through early 2009, including a workflow subsystem and end-user dashboards to complement the AuditExchange engine. For the workflow and dashboards, he says, he leveraged Microsoft (Redmond, Wash.) .NET, Visual Studio and SQL technologies. For data crunching, a Windows-based server was obtained.
Although static data was available for development and testing, to go live Riddell needed regular feeds from FNF's various corporate systems. When Riddell ran into roadblocks with getting IT resources dedicated to the project, he gained executive sponsorship from the CIO. During the negotiations with IT, however, it was decided that data delivery would occur as nightly batch files, Riddell says. But later, as FAST evolved, the company would migrate toward real-time data delivery, he adds.
By August 2010 data was flowing into FAST and the system went live successfully. Within a few weeks, Riddell reports, it began returning actionable information. "We've found no suspicious employee behavior at our head-quarters," he says. "However, FAST has detected irregularities at our remote offices that required investigation."
Phase 2 development occurred through the end of 2010 and rolled out on Jan. 1, 2011. According to Riddell, the resulting data volumes proved so large that FNF added a second server. When both servers struggled, load-balancing capabilities were added to push work to additional servers as needed.
Further, FAST soon began generating thousands of audit alerts, requiring modification of the alerting scheme to efficiently prioritize them. "We developed a ranking system that places the most urgent detections at the top," Riddell explains. "This permits our auditors to triage them quickly. We built the ranking system to automatically evolve as FAST evolves, minimizing further development tasks."
In addition to identifying audit-worthy transactions, FAST is creating unexpected efficiencies, Riddell notes. "For example, in accounts payable and payroll it has assisted with reducing errors," he says. "In fact, accounts payable is already asking us for additional analytic tests."
Moving forward, the next FAST build-out will include a data warehouse with a reporting database. "This will enable management to utilize the completed analytic results in their decision-making, resource-planning and risk-assessment exercises," says Riddell. "It's all possible with AuditExchange as the high-performance analytic engine at FAST's core."