Chartis' Lean, Multidisciplinary Analytics Team Delivers Big Results

Insurers have struggled to adopt analytics in an efficient and consistent manner across the enterprise, not least because they have tended to use isolated teams of "quants" who connect on an ad hoc basis with business and IT executives. Chartis ($30.7 billion in 2009 net written premium), the rebranded P&C arm of AIG (New York), has shown how a multidisciplinary approach not only can deliver advanced analytics solutions faster and more efficiently, but it also can deliver big results: Since its inception in 2005, Chartis' analytics team has delivered more than $750 million in benefits to the insurer.

The team originally came together when John Savage, VP of strategic analysis, and David Lee, assistant VP of strategic analysis, were tasked to build a predictive model for the company's executive liability business. Senior executives, Savage recalls, "were concerned about rising losses in the D&O [directors and officers] insurance line ... and wanted statistical modeling to boost the profitability of that business."

Today, Savage reports, the team tackles projects principally in three areas: predictive modeling for commercial underwriting; broad use in claims; and high-value, finance-related projects emanating from the office of CFO Rob Schimek, to whom Savage reports. When tasked by Schimek to create the team, Savage says, he thought the team's relevance within the larger organization would depend on depth of expertise. "You don't want 11 Eli Mannings on the team," he explains, referring to the New York Giants' quarterback. "You want position players, too."

Chartis' approach to building the advanced analytics team was to assemble high-level achievers from various disciplines to create an elite, multidisciplinary body of problem solvers, Savage adds. The composition of the team, he says, was driven by a strategy of creating a small and self-sufficient group of leaders dedicated to four founding principles -- the team and its output needed to be relevant, accurate, institutionalized and nimble. Today the team consists of seven individuals, including Savage and Lee, two certified database administrators, two data analysts, and an actuary with predictive modeling expertise. "The team is evenly split between the data and presentation sides," Savage comments.

The team's position within the CFO's office ensures that the projects it takes on have high value, visibility and management support, according to Savage. "I think of my role as that of a consultant," he says. "David [Lee] and I get involved at a high level and seek to understand problems and map out solutions."

The team's dedication to the principle of "relevance" translates into an unwavering focus on profitability, and its emphasis on institutionalization ensures that it will deliver models in support of strategic priorities, Lee relates. "Analytical teams are too often occupied with creating models that are not being used to support business decisions," he observes.

Data Masters

Work within the Chartis team begins with mastery of key data sources. "It's a task of extracting potentially 20 years worth of historical internal and external data expressly to build statistical analysis to implement a business solution," Lee says. "We take data as input and then run algorithms and tasks to determine what's relevant to the end question, then we synthesize patterns and data into a model that can be leveraged into a presentation layer -- a GUI -- that end users can interact with."

The backbone of Chartis' analytic capabilities is technology from SAS (Cary, N.C.). According to Savage, the team has SAS licenses for each member's desktop, as well as access to Unix client/server platforms with large data sets. SAS Scalable Performance Data Server (SPDS) is used for the largest data sets. Savage adds that Chartis uses open source solutions, when possible, for presentation work and the automation of data collection tasks. And the team accesses SAS' OnDemand team to ramp up staff resources according to need.

The nimbleness of the advanced analytics team accounts in significant measure for its ability to deliver, Savage reports. "There's great value in simply expediting what would otherwise be slow projects," he says. "We can take on high-stakes projects that we can't afford to have last two years. We can come in, quickly solve the problem and avoid an investment in teams upon teams of consultants, internal IT professionals and cross-functional teams."

Anthony O'Donnell has covered technology in the insurance industry since 2000, when he joined the editorial staff of Insurance & Technology. As an editor and reporter for I&T and the InformationWeek Financial Services of TechWeb he has written on all areas of information ... View Full Bio