The business case for reducing fraud is compelling for property and casualty insurers. With fraud on the rise and accounting for an increasing number of claims, reducing its incidence by as little as one or two percentage points can push millions of dollars to a carrier's bottom line. An estimated 10 percent of all U.S. P&C claims are fraudulent, and we estimate that fewer than 20 percent of those are detected or denied.

Predictive analytics can be a powerful weapon in the battle against fraud. By modeling past incidences of fraud, and pairing those results with social-network analysis, predictive analytics can help insurers understand what it is about a claim, claimant or insured party that correlates with a higher propensity for fraudulent behavior. Predictive analytics can help insurers discover sophisticated fraud rings and help them intervene earlier in the claims process to prevent payment for a fraudulent claim.

Predictive analytics is the process of analyzing historic and current data and generating a statistical model to help predict future outcomes. At the heart of this approach is the concept of predictors: one or more variable factors likely to influence an outcome that can be measured or scored to predict probable results.