For underwriting, predictive analytics have evolved from the traditional methodologies to utilizing the current advances in predictive analytics to predict pricing and future losses with uncanny accuracy. Predictive analysis allows the underwriter to "filter out" the normal underwriting cases and focus solely on the exceptions to the human underwriter. By automating this process, underwriters are spending less and less time screening potential applicants that would pass the underwriting screening. Instead, they are focusing on potential cases where an application would disqualify an individual, thereby resulting in only quality applicants that meet the underwriting guidelines making the cut.

Statistically, 100% of P&C companies with more than $1 billion in earnings use predictive analytics. This might attribute to the reason their earnings are much higher than companies that do not. Smaller companies would also benefit significantly by adopting these practices; however, many companies mistakenly believe that predictive analytics are complex and very costly.

What many smaller companies don't focus on is the ROI that naturally comes with a good predictive analytics strategy. It is more about execution than tools: the most expensive tool in the world can fall in the hands of the wrong analyst.

There are three considerations for predictive analysis:

1. Data sources
2. Data mining
3. Deployment.

Many companies begin with data warehousing and OLAP (Online Analytic Processing).

The most efficient data warehousing architecture will be capable of incorporating or at least referencing all data available in the relevant enterprise-wide information management systems, using designated technology suitable for corporate data base management. Most companies already leverage some form of Microsoft SQL Server in the enterprise.

OLAP, which is also referred to as FASMI (Fast Analysis of Shared Multidimensional Information), allows users to create real-time "views" from multiple data sources.

OLAP facilities can be integrated into corporate environments as part of a data warehouse strategy to provide a tool for managers and analysts to view various metrics, not only across the enterprise, but also within the market and potential markets. Views from the same data source can be tailored to the use case; for instance, a marketing department will receive actionable data that may differ from a view that is presented to a claims department.

Although data mining techniques can operate on any kind of unprocessed or even unstructured information, they can also be applied to the data views and summaries generated by OLAP to provide more in-depth and often more multidimensional knowledge. In this sense, data mining techniques could be considered to represent either a different analytic approach (serving different purposes than OLAP) or as an analytic extension of OLAP.

Next page: It's your data, so why not put it to work for you?

It's your data, so why not put it to work for you?
There are many benefits to implementing a predictive analytics strategy as part of a company's overall data strategy. The advantages far outweigh the disadvantages, and the investment can be minimal if you do your research.

The pros are obvious:

• Increase sales through analyzing buying patterns of existing and potential customers
• Limit the underwriting cycle to make underwriters more efficient by focusing on the risks, not the normalcies
• Create predictive scores and rating systems
• Decrease the likelihood of fraudulent claims and score claims by the likely size of the settlement to determine which claims are high priority.

The cons are a little less obvious but can be very costly:

• Inconsistencies in your predictive model
• Cost of implementing predictive analytics techniques
• Resistance to changes in reporting within the organization
• The need for clean and accurate data sources, which generally don't exist at face value in most companies.

For this reason, even small to mid-size companies should begin to consider and embrace data warehouse technologies as a means of moving toward predictive models. The biggest argument against predictive analysis is it can never determine human behavior like a seasoned underwriter is able to; however, predictive analysis is meant to augment, rather than replace the "seasoned" underwriter. It helps the underwriter, marketing or claims executive to make informed decisions about the direction they need to take, thereby enabling the enterprise to make intelligent, informed decisions based on historical data or trends that have occurred in the past.

It has been said that "history is bound to repeat itself." With properly-implemented predictive analytics, you can be sure that if it does, your company will be prepared for it. 

Read Nieuwendam's previous article in this series, "Put your data to work."