Decision Management Strategy





In most organizations, business process management (BPM) is a well-developed set of methods, skills and supporting infrastructure. Decision management is less so. This is understandable, since decisions have often been implicitly buried in transactional systems for as long as businesses have been developing and deploying them. It is only with the emergence of technologies and methods which greatly enhance the accuracy and efficiency of decision making that discriminating between decisions and process has become an imperative.

Decision management supports methods, technologies and processes which together provide a unified facility for designing, developing, deploying, monitoring and managing automated and partially automated business decisions. Relevant technologies include predictive analytics, business rules management, optimization and in some cases business intelligence. Methods include the newly ratified Decision Model and Notation (DMN), and those which incorporate the decision lifecycle, to be found in integrated decision management platforms.

Since it is our belief that decision automation is a separate but complementary wave to process automation, and will eventually eclipse it, so it is necessary to take an integrated view if we are to avoid the islands of automation and information that plagued transactional systems prior to the widespread adoption of ERP (Enterprise Resource Planning) platforms. In fact it is useful to think of a Decision Management platform as the ERP of decision automation.

The starting point for many organizations will be the capture of existing manual decision processes so they can be incorporated into a business rules management system (BRMS). This in itself is an immensely useful process, since the rules associated with many decisions are a combination of documented procedure and undocumented knowhow. Once rules are embedded into a BRMS they can be used by any system that needs to process them, and they become more transparent, more easily modified, and managed more effectively. The thousands of rules which are typically embedded in program code are likely to stay there for the time being, although efforts can certainly be made to extract critical rules when systems are modified.

Decisions which are generated from analytical models typically come from data mining and maybe statistical analysis. Sometimes these can be reformulated into a deterministic business rule, which can then be incorporated into a BRMS. However there are dozens of different methods used for generating such models, many of which cannot be expressed in the logic understood by a BRMS. These too need to be managed within the context of a Decision Management platform so that the lifecycle can be managed effectively. This lifecycle incorporates requirements, development, deployment, monitoring, management and modification. Rules available through a BRMS and those deployed as analytic models need an integrated environment, so they can be used collectively to address business decisions. This also applies to models which result from optimization, where the best deployment of resources is established based on various constraints and objectives.

Only when an integrated Decision Management platform is employed can business processes call upon decision models that are well understood and managed. The alternative is decision model chaos, with fragmentation of techniques, technologies and methods, since the number of decision models will number in the thousands, and for some larger organizations in the tens of thousands. These models will lend the same critical advantage to an organization that transactional systems have in the past, and as such need to be managed rigorously.

The urgency for establishing a Decision Management discipline is well demonstrated by the rise in big data analytics. The capture of increasing amounts of data, soon to be accelerated once again by the Internet of Things (IoT), permits ever more diverse analytics initiatives. The dumb business processes which have always needed human intervention for decision making, are becoming increasingly smart and efficient through access to decision models. The plethora of case studies showing how organizations are adding intelligence to their processes simply demonstrates that process efficiency is no longer enough – we have to realize new efficiencies and effectiveness through decision automation – and this is only feasible through an integrated Decision Management platform.

The previous article in this series is Decision Management Case Studies

Get the full report: