In the data-mining race engulfing the insurance industry, American International Group Inc. (AIG) may not have started near the front of the pack. But the company thinks it's gaining ground. AIG has invested heavily in a data-analysis team that's looking to harness new sources of information so the company can make better decisions and increase profits.
The chief science officer of AIG's property-casualty operation, Murli Buluswar, has assembled an analytics team of about 100 economists, physicists, data scientists and other specialists--mostly outside hires who came to AIG this year--to handle the task.
The team is beginning to delve into reams of in-house data--and, just as important, an ever-growing pile of outside information--in search of new information about its products, customers and the atmosphere in which they operate.
"There's no lack of data around us, but there is a lack of clarity around what questions to ask and why that matters," said Mr. Buluswar in an interview. "With all this data that is being transmitted and created every day, the winners ... will be firms that can understand that information and connect it to problems they are trying to solve."
In many cases, the analyses Mr. Buluswar's team is attempting had been impossible until recently, but advances in computing power and an explosion in the amount of digital data available for study have provided the company with a chance to uncover new insights about its business.
AIG's property-casualty unit is hardly alone in the insurance industry in these efforts. According to McKinsey & Co., one study estimated the amount of new digital data increases 40% annually across the globe, and insurers large and small are looking for ways to benefit from it.
But AIG executives--including Chief Executive Robert Benmosche in a talk with investors earlier this year-- acknowledged that until Mr. Buluswar joined the company late last year, the insurer hadn't been doing enough to try to understand its in-house data, much less to analyze outside information.
The brutally competitive insurance marketplace, where companies constantly try to cut prices, contributes to a thirst for fresh insights, said Russ Schreiber, the head of the insurance practice at Fair Isaac Corp. (FICO).
"They're grabbing on to the analytics stuff with both hands," Mr. Schreiber said. "The market has become so much more competitive today, the consumer is so much more price-sensitive, that the carriers are really looking for ways to bring down that cost."
One of the widely acknowledged data-collection leaders in the insurance industry is auto insurer Progressive Corp. (PGR), which spent years developing a device policyholders could volunteer to install in their cars to measure how and when they drive. The device allows the company to price its coverage and potentially give discounts using information most other insurers still haven't gathered in any significant quantity.
Such usage-based insurance is now considered the future of the auto-insurance industry, and Progressive has analyzed more than five billion miles worth of driving data and offers the product in more than 40 U.S. states. Allstate Corp. (ALL), a fierce rival, said in July it had collected 11 million miles of data and offered the product in three states.
Mr. Buluswar once worked for Progressive, and more recently was a vice president at Farmers Insurance Group, a unit of Zurich Insurance Group AG.
At AIG, Mr. Buluswar's team is studying information about its marketing efforts, customer-service practices, underwriting assumptions and fraud-detection programs-all with an eye on uncovering new insights hiding in the data. The company aims to improve loss forecasts, reduce costs and devise more accurate pricing for insurance policies, even for the biggest clients in its commercial-insurance operation, where such analysis is relatively rare.
FICO's Mr. Schreiber said rapidly improving data-processing capabilities and the quickly multiplying amount of data are proving especially helpful in spotting insurance fraud. FICO sells software that helps insurers detect potential fraud.
The next step in the process of rooting out networks of people working together to defraud insurance companies: using websites like Facebook and LinkedIn to spot otherwise-hidden connections. FICO has developed the technology but hasn't yet sold it to any insurers, Mr. Schreiber said.
"The technology is moving forward very quickly," he said. "You'll see the nerds and the former law-enforcement guys working shoulder-to-shoulder in ways they never had before."