A great deal of what passes for a ‘business process’ is actually decision logic, and so large numbers of decisions are already embedded into processes in program code. As such they lack transparency, accessibility, quality control mechanisms and are extremely difficult to orchestrate. By treating decisions as more than an appendage to the process, it becomes possible to implement them more easily, embrace greater complexity and measure efficacy.
Decision tables and decision trees are often the first mechanisms used to explicitly define decisions. These are deterministic in nature, and once a set of conditions has been met the corresponding decision can be made. In loan approval for example, data such as age, salary, number of dependents, other loans – and so on, will be fed into a table or tree, and a ‘accept’ or ‘decline’ determined, with little indication of whether the decision is borderline or firm. Of course this can be emulated by finer granularity in the logic. Business rules management systems (BRMS) employ these techniques, and bring many benefits. Where transparency is required (often demanded by regulators) a BRMS provides an ideal environment for implementing decisions.
The recent rise in profile of predictive analytics means a whole new battery of techniques is available to create decision models. Many of these are probabilistic in nature, assigning a probability to a decision. And so a loan approval for example might have a 51% probability of belonging to the ‘approved’ class – which is clearly borderline. Decision trees can produce probabilistic models as do many other techniques (Bayesian methods for example). Where transparency is not critical other mechanisms are available, neural networks being one of the most popular. It is important that decision models can use innovative new techniques since this is often where an edge might be found.
Finally a whole set of decisions is associated with resource deployment. How many people in a call center for example, should be placed on various campaigns? Or how should manufacturing resources be used to maximize profitability? It is here where optimization techniques (prescriptive analytics) offer powerful solutions that can significantly improve operational efficiency.
Business rules management, predictive analytics, optimization, and ultimately, decision management, are all capable of addressing complex business decisions. They provide mechanisms for the automation of numerous decisions that form a large part of most business processes. Of course the decision made during one activity within a process often impacts other decisions, and so there is a need for them to be orchestrated will facilitate such orchestration and allow managers to monitor the effect of automated decisions. It really is very important that organizations put a proper infrastructure in place and use proven methodologies, since the number of decisions which can be fully or partially automated is often be measured in the thousands or tens of thousands in a large business.
Most industries already use some form of decision management technology, but as new technologies broaden the application domain, so there is a need to be more rigorous.
Examples of decision management application are numerous and include:
- Banking – authorizations, self-service web enquiries, account management
- Insurance – underwriting, claims processing, fraud management
- Retail – marketing/campaign management, behavior scoring, order configuration
- Telecom – call center/CRM, problem resolution, personalization
- Government – compliance, collections, self-service web enquiries
- Manufacturing – product/service recommendation, sales commission calculations, process improvements.
This is really just the tip of the iceberg, and many businesses are going well beyond these well known applications. The improved efficiency and efficacy associated with effective decision management is both tangible and substantial, and rests almost entirely on the ability to create, modify, coordinate, monitor and manage business decisions effectively. This is a very rapidly evolving set of capabilities with big data, new analytics techniques and management methods, changing the way businesses use technology.
The previous article in this series is Intelligent Business Process Methods
The next article in this series is Decision Management Case Studies
Get the full report: