James Taylor's Decision Management

I saw this article on decision trees over on Intelligent Enterprise - Analysis: Decision Trees Boost Web Site Performance. While this article gave some good advice on decision trees, and showed why using a technique like a decision tree to devise a better treatment for a customer is effective, it had two holes.

Firstly it assumes that only web data should be analyzed this way. In fact most companies have lots of other data besides their web traffic that can be analyzed to build decision trees. Indeed decision trees that combine data from several different sources are likely to be more predictive. Don't assume analytics = web analytics.

Secondly it assumed that the website was the consumer of these decision trees. But what about your ATMs, your call center, your branch staff, your marketing mailers? In fact a better approach is to combine this kind of predictive analytics with business rules into decision services and have those decision services drive better decisions everywhere.

To misquote George Orwell,"Decision Trees Good, Decision Services Better"

Technorati Tags: analytics, decision tree, web analytics