Location Intelligence | Pitney Bowes
Property and Casualty Insurance Company
Insurer targets location accuracy to lower policy risks and improve pricing.
- A national, top 25 property-and-casualty (P&C) insurance company
- Provides a diverse range of commercial and personal insurance products
- Fiscal year 2017 revenue exceeded $30 billion
- Consolidate and validate customer geographic, risk, and property attributes data to provide greater accuracy in assessing risk and pricing policies
- Enrich internal data with externally sourced precision geocodes, risk and property attribute data to support more detailed analysis
- Integrate enhanced data lake into new policy management system to streamline customer quotes
- Increased proportion of parcels for which location information is precise down to the parcel centroid and building centroid level, from 50% to nearly 100%
- Expanded data underlying policy-pricing decisions to include factors such as landslide risk, distance from the coast and drive time over a street network to the nearest fire station
- Demonstrated a 0.3% to 0.6% improvement in loss ratio for test cases, representing a prospective multimillion-dollar savings
- Estimated return on investment (ROI) in 14-17 months
When this property-and-casualty (P&C) insurance leader replaced its outdated policy administration system with a third-party cloud solution, it needed a better way to determine exact property locations. Whether for a home, auto, or commercial, geographic precision down to the parcel or building level would allow the insurer to more accurately price policies based on proximity to different kinds of risk. This information also needed to seamlessly integrate into the new policy administration system.
The insurer’s legacy geolocation software vendor was able to provide information at the parcel (lot boundary) level only about 50 percent of the time. Often the solution assigned a location only to the block or ZIP code level. Without more precision, the insurer could either underprice a policy, which resulted in undue risk, or overprice policies, which might result in lost business.
The insurance company needed its new policy administration system to incorporate highly precise location information. It also wanted to enrich its internal data with externally available data sources to support more accurate risk assessments.
The insurer projects a complete return on investment in 14 to 17 months.
The insurer had previously implemented the Pitney Bowes® Spectrum® solution to improve data quality and enhance risk models for its actuarial group. When it needed to incorporate Location Intelligence into its pricing models, the company again turned to Pitney Bowes.
The P&C insurer rolled out the Spectrum Technology Platform to consolidate and validate data from its disparate internal systems. Now the solution determines specific geographic coordinates for each location and assigns each a unique identifier, known as a pbKey™, that remains persistent with the address location even if the latitude/longitude co-ordinate moves or improves up to 50 feet.
This pbKey unlocks a wealth of information associated with each location (such as previous storm or flooding events), and because each pbKey is linked to specific geographic coordinates, all associated information is preserved over time, even if the policy or customer changes.
The result is a data set with 198 million properties (each one with a pbKey) that Pitney Bowes® then enhances with third-party information such as wildfire scores, distance from the coast and flood zone attributes. The Spectrum® Spatial bundle simplifies data analysis by creating graphic representations of factors such as wildfire, landslide or flood danger. The module also calculates driving time between locations, such as a fire station and a particular address — an important consideration in assessing risk and pricing policies.
All this information is now integrated into the insurer’s new policy administration system, improving the quality and quantity of data underlying pricing decisions. The Pitney Bowes solution has also decreased the insurer’s “time to quote” by enabling a higher volume of quotes to go through an automated process that returns all risk attributes associated with a particular address in just one-tenth of a second.
The insurer re-ran a number of risk assessments that it had previously used in pricing decisions. With the Spectrum data, it could have improved its loss ratio between 0.3 and 0.6 percent.
By validating and enriching its data through Spectrum, the insurer increased the proportion of properties for which it has parcel-level Location Intelligence from just half to nearly 100 percent. This provides invaluable data that the company uses in assessing location-specific risks. In addition, the persistent pbKeys enable the insurer to view claims and catastrophic events for a specific location over time regardless of the current policyholder. This provides a unique and more permanent perspective in evaluating risk by geographic location, in contrast to how most P&C companies evaluate hazards only by policy or the individual associated with the policy.
After implementing the Spectrum Technology Platform, the insurer reevaluated risk and pricing models previously used in pricing decisions. The company found that with the Spectrum data, it could have improved its loss ratio between 0.3 and 0.6 percent, representing millions of dollars in savings.
The insurer projects a complete return on investment (ROI) in 14 to 17 months. Deployment speed is a crucial factor in achieving that ROI: The entire project spanned less than five months, from initial data consolidation, through validation and enhancement, to integration with the policy administration system. As effective as it has already proven to be, the solution continues to evolve. With almost 200 million U.S. properties tagged with unique pbKeys in place, the insurer can now layer on other data, creating additional insights into geographic risk and can also add demographic data to provide more accurate, customized products for its policyholders.