Intelligence Behind Data

Insurance companies collect and store a wealth of data about customers, to the point that they could be called “data-rich” enterprises. However, the real issue is whether this data translates into actual knowledge and therefore real business value.

Sure, insurers can provide all sorts of facts and figures, such as the proportion of policyholders with vehicles newer than 2002 or the number of lapsed policies over the past year. But what does this accomplish in terms of business decisions or profitability? In fact, when it comes to using data for business purposes, many insurers are “knowledge-poor.”

This is where “enterprise intelligence” data mining comes in. Enterprise intelligence is the process of selecting, exploring and modeling large amounts of data to uncover previously unknown patterns – which can point to critical business insights, such as customer buying behavior, geographic exposures and even fraud detection.

DATA VALUE

In today’s property and casualty insurance marketplace of single-digit return on equity (ROE), high combined ratios and paltry investment yields, companies are digging in and looking for any edge they can get. When it comes to sustained profitability, the answer might be right in front of them – in the very data they have stored across disparate systems throughout the enterprise. These customer databases, if properly managed and exploited, represent valuable corporate assets.

While some of the processes and terminology associated with data mining may seem fairly technical, the bottom-line of the technology is a completely new view of past and future customer profitability. Traditionally, insurance companies have tried to increase their customer base simply by expanding the efforts of the sales department and broadly targeting potential clients who meet certain policy constraints. These marketing campaigns often yield a low hit rate. At some point, sales become more difficult and higher marketing budgets lead to diminishing returns.

In contrast, enterprise intelligence data mining strategies enable analysts to tighten the marketing focus. For example, the criteria of a marketing campaign could be refined by maximizing the lifetime value of policyholders – the profits expected from customers over an extended period of time. To sharpen the focus further, advanced data mining techniques can combine segmentations to profile the high lifetime-value customers and produce predictive models to identify those in the group most likely to respond.

CUSTOMER RETENTION

Most insurers know that a customer holding two or more policies with the same company is more likely to renew than a policyholder with just one policy. But, how do you measure which customers are likely to buy bundled policies? Using a data mining technique called “association analysis”, insurance firms can more accurately select which policies and services to offer to which customers.

Notably, Chubb Insurance uses such a process in its U.S.-based “Pinpoint program”, a data mining and marketing application building off the company’s personal lines “Masterpiece products”. Chubb’s objective was to cross-sell excess liability and valuable articles coverage to its traditional customer base of home and auto owners. Through the Pinpoint application, Chubb measures the value of each customer to determine which ones are more likely to purchase specific products. They then use “segmentation” and “cluster analysis” capabilities to differentiate customers and develop a relative “score” to determine a customer’s propensity to buy.

Since the program was introduced in May 2001, Chubb has expanded Pinpoint to commercial lines, citing positive feedback from agents in terms of qualified leads and future sales. Further plans for Chubb include expanding enterprise intelligence data mining applications in areas of underwriting, risk assessment and customer relationship management.

This is one example of the potential of enterprise intelligence data mining. There are many more uses, such as establishing accurate rates, developing new product lines, estimating outstanding claims provisions and assisting regulators in understanding a company’s rates and models. Fraud detection in its own right has become a significant application target of several data mining applications.

METHODOLOGY

There are, of course, obstacles in the way of true enterprise intelligence data mining solutions. Some companies may have data in several different, largely inaccessible formats. In addition, many software solutions do not integrate well with current IT infrastructures. In some cases, highly trained quantitative experts spend more time trying to access and manipulate data from disparate sources than actually solving business problems.

These roadblocks demonstrate why a proper methodology for implementing data mining techniques is crucial for the success of any enterprise intelligence project. A thorough approach to enterprise intelligence can be divided into several areas:

Plan – a set of proven, best practice roadmaps supported by industry-specific data models, project methodologies and consulting expertise;

ETLQ – extraction, transformation, loading and data quality functions which allow reading of all data formats;

Intelligent storage – providing analysis-ready information compatible with your existing query/reporting tools;

Business intelligence – keeping control over consistency and reliability of source data, while providing analysts with interfaces and functionality according to their needs; and

Analytic intelligence – a versatile, integrated platform for a wide range of analysis, with capabilities for predictive and descriptive modeling, forecasting, and optimization.

THE BOTTOM-LINE

Exploiting the results of data mining is where the “rubber meets the road” for insurance companies. Information obtained from data mining can be incorporated into executive information reports or online analytical reporting systems. Executives and senior managers throughout the organization can then rely on the data to clearly see what impact various company activities are having on the overall company strategy – thus enabling them to make critical business decisions and answer vital business questions such as: “How we can we increase the return on investment of our marketing campaigns?”