≡ Menu

More on the relationship between Decision Management and BPM


There’s a great article over on IBM’s Good Decisions blog called “What’s Decision Management got to do with Business Process Management?” The article lays out a nice scenario for Decision Management and differentiates between business event processing, business rules management, analytics and business process management. It is definitely worth a read. There are a couple of additional points I would like to make.

While it is true that many Decision Management scenarios are identified in the context of BPM projects, this is by no means always the case. Legacy modernization, event processing and straight up analytic projects are also great places to start:

  • In legacy modernization projects it is not uncommon to find specific modules that are both high change and valuable to other systems. Examples include pricing engines, eligibility determinations and adjudication modules. These are almost always decision-centric modules in my experience and externalizing these modules as Decision Services – an service-oriented component that makes decisions, answers questions, for other components – using a Business Rules Management System and Predictive Analytics can result in a more precise and more reusable components that have much lower cost of ownership thanks to the agility and business user engagement that can result from the use of business rules.
  • Complex Event Processing or CEP scenarios are also great places to find Decision Management opportunities. While the use of business rules and analytics as part of the correlation engine is powerful, many correlation problems are tightly coupled with business decision making problems. For instance, having correlated the events that identify a particular situation as fraudulent, a business decision as to how to act must be made. The likelihood is that this business decision will be used in multiple scenarios and will have value in process settings so externalizing it as a Decision Service is going to be an effective approach.
  • I work with a lot of companies adopting analytics – often companies trying to go from using business intelligence tools for reporting and dashboards to using more predictive analytic techniques. These often find that failing to explicitly identify the decision for which the analytic are being developed severely limits the value of the predictive analytic model. Unless you “begin with the decision in mind” and understand what decision you are making, it can be hard to tell what kind of analytic model you need. In addition, wrapping the predictive analytic model with business rules to build a component that makes decisions analytically, a Decision Service, can make a huge difference to the success of these projects.

I agree also that the use of BI tools to provide decision support is important but, in a Decision Management context, I think it is also important to consider two other uses of BI tools:

  • Monitoring and reporting on the effectiveness of the Decision Service in business terms is essential. You must know which options the Decision Service selected, which rules fired, how long it took and other technical metrics but you must also know how it impacted the business. This is particularly important when experimentation is built into the Decision Service to support a test and learn strategy. Once you are experimenting with several distinct sets of business rules and analytics you will need strong BI tools to compare their business effectiveness.
  • Integrating decision support and Decision Management capabilities can be very effective. Giving a person access to decision support tools while also using Decision Services to provide the best options between which they must select or otherwise constrain their environment can give you the best of both worlds – automated compliance with policies and regulations and human insight to take make judgment calls.

Finally I would add one last comment on the value of business rules and analytics in combination. I come across lots of analytic projects where predictive analytic models work – they are highly predictive – but they are not in use. Partly this comes from problems getting the models deployed in production (something a BRMS can help with) and partly it comes from a level of mistrust of the “black box” model by business people. Because they deal with uncertainty, these models can be hard for the business to trust. They can also be counter-intuitive. Taking the insight from these models and turning it into a set of business rules – as a decision tree, a set of rules or a predictive scorecard for instance – makes the insight of the model clear and editable. As a result the business users can interact more effectively with the model, are more likely to use it and can apply their judgment to it.

What’s Decision Management got to do with Business Process Management? is s a great article and if you didn’t read it when I linked at the top, you should so here it is again.