≡ Menu

First Look: Valen InsureRight


Valen Analytics has been around since 2004 and is focused on providing proprietary data, analytics and predictive modeling to the Property and Casualty market with 25 customers in production. To help carriers manage and segment their portfolios to drive underwriting profitability, Valen develops productized models based on pooled consortium data for carriers. Valen’s InsureRight platform allows them to collect data from multiple carriers, develop consortium models with this anonymized data and external data, and then tune and deploy these models for each carrier to use in production. The platform has a series of layers:

  • A base layer stores industry data and external data as well as providing a data warehouse for reporting against their customers’ systems of record. Predictive analytic models are then built with this data.
  • There is a layer that supports data governance, scoring of the predictive models and portfolio management. It also provides BI capabilities for reporting on the data as well as rules and compliance capabilities.
  • The products themselves are developed for worker’s compensation, homeowners, commercial auto, Business Owner Policies and commercial packages.

Valen works with carriers in one of several ways.

  1. Small to mid-sized carriers can take an industry standard model and deploy it on the Valen InsureRight platform.
  2. Carriers can calibrate one of these models to their own portfolio and deploy that on the Valen platform.
  3. Custom models can be delivered on the same platform if there is no industry model or if a carrier really wants a custom solution.

In principle the platform could also be used to deploy third party or internal models. The platform itself is cloud-based allowing for easy integration and distribution of results. The models are generally delivered as part of a report used by an underwriter for instance. The scores and decisions are also available through an API, however, so they can be used in automated systems.

The homeowner’s models are focused on helping carriers know what they are insuring, on improving profitability and aligning strategies with field agents/managers. Today carriers focus all their energy on inspecting new business but Valen found that 22% of first renewals have an issue they did not have when first insured, demonstrating that this effort might be better used if spread between new policies and renewals. Meanwhile profitability is tough in homeowners with the loss ratio exceeding 100 most years.

The homeowner’s database that drives the models has 18M policies along with detailed and recent inspections. All of this data is anonymized and linked to external data – demographic, regional and weather data (climate and catastrophe data). This data drives two scores – a Condition Hazard Score and an Insurance to Value Score – as well as one automated decision to identify the correct inspection type (outside only, inside only etc) to maximize the return on inspection investment. The hazard score predicts how likely an inspection would be to find something material while the ITV score predicts the likelihood that the rebuild cost of a home exceeds the amount it is insured for.

In addition the platform provides portfolio management capabilities. These summarize the overall book of business in terms of the scores – average hazard score for instance. A geographical view shows how these scores vary for a specific state. This is combined with profitability information, premium and policy growth and retention and other metrics. Data validation uses external data to compare policy characteristics, such as year built or square footage, to county data to validate the book of business. The tool provides an overall view and drill-down capabilities for each element.

The workers compensation business is focused on market visibility, pricing and inertia to adopt advanced analytics within underwriting – what is a carrier’s portfolio of risk and where are the pockets of poorly priced business as well as how to avoid adverse selection. The workers comp platform has 4M detailed policies (anonymized) as well as an additional 2.3 million audit policies and a wide range of external data.

The platform is used to deliver three scores. The core is a risk score for a policy (a business) that is delivered with the top 5 risk factors for the policy. This is supported by a misclassification score showing how likely it is that the policy is misclassifying the people covered and a premium impact score that shows how much impact on the premium such a misclassification is having. The reports can be drilled into to see how similar policies are priced more accurately as well as supporting explanations and reasons. The portfolio management capability takes these same predictions and applies them to the whole book of business including geographical distribution, etc.

More information on the Valen products can be found here.


Comments on this entry are closed.