Listening to the Day 2 Keynote at the recent ACORD/LOMA Systems Forum, we were reminded of a quote from Abraham Kaplan: “Give a small boy a hammer, and he will find that everything he encounters needs pounding.”
Ian Ayres, the author of the widely cited book, Super Crunchers, was providing example after example of how analytics, especially predictive analytics, are changing how business is conducted and, more importantly, what the results of the business are.
Ayres was preaching to the converted for many of the attendees, present company included. In January, we shared our belief that the intelligent use of data and analytics is the overarching issue for insurers. We have have blogged frequently on the impact of analytics on aspects of the insurance industry, such as risk management, marketing, and the use of Telematics for underwriting and claims management.
Ayres added some elements that went beyond an exhortation to proceed. First, Ayres noted that while the insurance industry is ahead of others, in some respects, because of its familiarity with the importance of data, most insurers tended to use analytics for pricing, but not for other critical elements, such as marketing and customer service. Ayres noted that retailers (such as Walmart) tech firms (Google) and Social Media (Facebook) are setting new standards in customer centric utilization of analytics.
More significantly, Ayres noted that senior managers in many industries (and we believe that insurance is among these) lack a firm understanding on the proper uses of various analytic vehicles. Ayres indicated that there is an over-reliance on correlation analysis in business modelling.
Correlation is important for identifying association of variables, but does not test causality (for example, credit scores are statistically correlated with insurance claims behaviour, but do not indicate that a particular credit scores elements cause claims activity, or vice versa). To determine causality (which can be important in understanding customer behaviour), some form of randomized testing is usually required.
This is not an analytics issue. It is a management issue. To intelligently make such decisions, managers must have sufficient grounding in the use of analytics. While we don’t have data for the insurance industry, our going in hypothesis is that this area needs work.
So, what do you think? Are analytics on your radar? Does your organization have a plan for anayltics in all aspects of its activities, or is it restricted to the actuaries for pricing? If you are using analytics, do you have a variety of tools, or are you pounding everything with a correlation tool?
Leave your comments here, and let’s start a discussion.