From Add-on Analytics to Enterprise-Wide Decision Management

(Posted by Guest Blogger Dan Dolberger)

As someone who has patiently followed the slow percolation of predictive analytic methodologies into the financial services over the past five years, I couldn’t help noticing the relative surge of recent articles and posts on the subject.

Gib Basset’s December 2007 post on the EDM blog discusses the merits of enterprise-wide Decision Services in Financial Service Companies. Gib’s longer article in the DM Review presents this concept in more detail. A November 2007 article in Insurance & Technology Magazine, titled “Predictive Analytics and Complex Event Processing Technology Move to Cutting Edge of Financial Services Industry raises several examples of how leading financial services firms have successfully deployed predictive analytics by taking an enterprise view. Not to mention Stephen Swoyer’s July 2007  article titled Predictive Analytics: Slow Adoption despite Big Benefits, and Ian Turvill’s more optimistic March 2007 EDM post citing 39 North American Insurance Companies recognized by Celent as “Model Carriers” for their optimal use of technology, including nine companies which specifically used enterprise-wide predictive analytics.

Most of these articles and posts point at what seems to me a critical issue. True Decision management happens when predictive analytics meets business logic on an enterprise wide scale, as opposed to isolated silo based solutions. You can’t staple the predictive analytics concept to the side of enterprise. You can’t launch predictive analytics as an “isolated, bolt-on solution”. As Daniel Bravo, SVP of Selective Insurance’s strategic operations group puts it, you need to ” bring [predictive analytics] into the regular insurance operations of the company and not disrupt those operations while still taking full advantage of the guidance and granularity the models provide”.

Or, in the language used by Insurance and Technology magazine’s Nathan Conz and Melanie Rodier, you need to immerse the company in predictive analytics.

That is easier said than done. “Immersion” is a big word, invoking visions of yet another frightfully disruptive IT overhaul. So how do you bring predictive analytics into regular mainstream operations avoiding both stapling on the one hand and disrupting on the other?

One possible good approach would be to find a solution which can be initially applied within a particular department or area of activity and then gradually scaled further to eventually encompass the entire organization.

To me Price Optimization is a good indication of where to start a financial service’s journey towards “total immersion” in predictive analytics. Pricing is one of those activities that intimately link an organization’s day-to-day tactical decisions to its overall financial performance in a clear and measurable fashion. It connects the tactical to the strategic at any given time point, as well as short term KPI’s to long term customer lifetime values.

Price Optimization is exactly the type of predictive analytics solution which can start in one department (pricing) and then be scaled and extended to address the question of which product to offer which customer through which channel at what price – a question which requires the concerted analytical effort of no less than the entire organization.

Of course, this type of solution goes hand in hand with an enterprise oriented architectre, one that emphasizes grid computing, collaboration and standards such as IBM Websphere.

Which really ties really well into a post titled Beyond the Golden Niche published on this blog By Barak Melnik.