Multiple decsioning engines – a reader asks

Neeraj asked me an interesting question the other day – how would a decisioning product like Oracle RTD and a business rules engine co-exist? Rather than answering this specifically I thought I would try and generalize it. After all there are products like Unica and Chordiant that also offer decisioning engines that are not general purpose business rules platforms so this is a question that others might have.

The first piece of advice I would give someone considering this is that they should know what decisions are important to them. Having a sense of the types of decisions that matter, who owns those decisions, how they are measured etc is a critical first step. A broad and shallow view is probably all you need to start with – a decision landscape or decision audit. This is what I cover in Decision Discovery, the first phase of Decision Management.

The second thing is to understand the characteristics of the engines you are considering. Some rules engines, for instance, do not handle analytic models well while others do. These decisioning platforms are often optimized for specific kinds of decisions such as marketing or customer treatment decisions. The relative strength of each for the various types of decisions involved is something you will need to understand. For instance, some decisions are compliance-oriented and a rules environment with good rule governance and management features will be critical while analytics will not matter at all. Others, like marketing decisions, might have few rules but will need not only analytics but adaptive analytics to do well.

The third and final thing is to get good at handling multiple engines. The reality of most companies is that they will end up using multiple business rules/decisioning platforms. Some decisions will be embedded in applications, some built using a platform, some perhaps handled as a service. It is easy to foresee both multiple platforms (through acquisition or because a platform is focused on a particular technology stack or style of decisioning) as well as multiple applications that include decisioning or are pure decisioning applications. Good impact analysis tools, an ability to reuse sets of rules and analytic models in multiple decisions (using standards like PMML or just by calling web services) and an awareness / record of which sets of rules or analytic models are being reused in which decisions will all be key.