The rapid adoption of predictive analytics in retail and in insurance
There was a New York Times article from February of this year that made a brief splash in the zeitgeist and then faded away. You may remember the blurbs saying that Target had found a way to infer that a woman was pregnant before she made it public. But the full article (linked below) gives us a lot more to think about.
Predictive Analytics in Retail
The basic idea is that “predictive analytics” allow companies to study data to predict consumer behavior and can therefore find ways to get them to buy more of something. The idea raises concerns about consumer privacy, but it’s fair to say this genie isn’t going back in the bottle.
Predictive Analytics in Small Business Insurance
The Small Business insurance industry has been engaged in this analytical effort for several years now, but we aren’t interested in whether you’re pregnant. We have a different goal than selling you more products: We’re trying to figure out how risky of a customer you are.
Insurance is a unique industry in that we sell a product for a price without knowing what our actual costs for that product are. More specifically, we don’t know how many claims an individual customer will have, so we don’t know how many dollars we will eventually spend on those claims. With more data and more analysis, we can get a better sense of whether a particular plumber or office building or auto fleet is a better or worse risk.
What’s the Goal of Using Predictive Analytics In Insurance?
With an improved understanding of each customer’s potential for having claims, we can more accurately price the risk. And yes, that means some businesses will pay more for their insurance than others. Said differently, it means that businesses with good risk characteristics will not be paying extra to subsidize businesses with less favorable characteristics. This isn’t an outcome that everyone likes, and I recognize that I’m glossing over a lot of the details here. But, in my experience, predictive analytics has lead to more availability of insurance because companies are able to price for an exposure rather than decline to cover it. It also leads to less expensive coverage for the majority of customers, more expensive for a small minority.
So, What’s the Problem with Using Predictives?
A jaundiced view of this effort would say that it’s all about more profitability. Of course profit is a motive; insurance companies aren’t charitable organizations after all. But they are also fierce competitors who want to sell more insurance than Company X any day of the week. This is another tool that helps them do just that.
In the end, I think the problem is that the industry hasn’t told this story well. In the NY Times piece, there was a sound bite that made the headlines, but there’s a far more nuanced and interesting story behind it. That’s true for predictive analytics in insurance as well, and all of us in the industry need to do a better job of telling that story.
Acadia is pleased to share this material for the benefit of its customers. Please note, however, that nothing herein should be construed as either legal advice or the provision of professional consulting services. This material is for informational purposes only, and while reasonable care has been utilized in compiling this information, no warranty or representation is made as to accuracy or completeness. Recipients of this material must utilize their own individual professional judgment in implementing sound risk management practices and procedures.