≡ Menu

Decision Management: Orchestrating Consistent Enterprise-Wide Decisions


As companies move to this more changeable, uncertain world that requires a coordinate extended enterprise, it is essential to manage decisions as well as processes – not by using process management to manage decisions but by managing decisions alongside processes. These operational decisions – micro decisions – are the front line in driving business agility and improving your business processes (by making your processes simpler, smarter and more agile as I like to say). These decisions include pricing and promotion decisions for a single customer, identifying a transaction as being a fraudulent one, determining how much to pay on a claim.

IBM defines Decision Management as an approach combining software and expertise to automate, improve and govern operational decisions across the enterprise (as does Decision Management Solutions, my company). This drives agility, customer-centricity and better risk management. They gave an example of a claims adjudication system that combines rules for claims validity and payment with predictive analytic models for predicting fraud risk. Another was a bank that used the approach to empower their front line staff to make better, more targeted offers to customers – the rules management element allowed them to operationalize the decision and transmit best practices while the analytics predicted risk and opportunity for particular product offers to customers. All brought into a single decision environment. This company improved their frequency of its offers by 2.5x, 10x improvement in cross sell acceptance and $15M in just 10 weeks!

IBM uses the same hierarchy of decisions that Neil and I introduced in Smart (Enough) Systems

  • Strategic decisions like “we are going to differentiate from the competition with customer experience”
  • Tactical or Managerial Control decisions like “how many people do we need in the call center to maintain service levels”
  • Operational or “micro” decisions like “how much do we pay on this particular claim”

IBM also like the Observe-Orient-Decide-Act model – the OODA Military Decision Loop. Different products support different pieces of the loop – a business rules management system, for example, supports the Decide and Act elements as do executable predictive analytic models while business event processing supports Observe while business intelligence and business activity monitoring support the Observe and Orient elements.

Decision Management, says IBM, enhances business processes from design time to run time. Primarily by removing decisions from the process so that the decision can be analyzed, assessed, changed without impacting the process. Examples include days not months to change pricing, 50% increases in quotes generated, eligibility from 6 weeks to 7 minutes, 10x offer acceptance.

From a practical point of view IBM has been integrated its products to make Decision Management more straightforward:

  • Cognos 10 and SPSS Decision Management 6 have been integrated
  • JRules has been embedded into the new IBM Business Process Manager 7.5
    Exposing the rules engine and rules syntax to process people
  • WebSphere Decision Server brings together event processing and rules
    For real-time decisioning in particular
  • JRules has a Business Monitor support pack to integrate it with business monitoring
    Bring rules-based decisioning to the monitoring environment.
  • JRules has an analytics support pack that brings in PMML models from SPSS
    This allows you to bring data mining to bear on rule creation (bringing, for instance, a decision tree from SPSS into JRules) and to bring predictive models into an integrate rules/analytic environment – check out this presentation I did at Predictive Analytics World on exactly how to do this and this white paper on Putting Predictive Analytics to Work in Operations.

Finally they discussed some adoption patterns and in particular how you might look at a process to see how Decision Management could help:

  • Customers externalize a decision from within a process so it can be managed more easily. Then they might use dashboards and reports to see how effective this decision is. More formal data mining and predictive analytics might follow to apply data to the decision directly.
  • They might also begin with decision support tools – business intelligence tools like Cognos 10 for instance – and then start to codify the expertise of those using those tools using business rules.
  • Some customers have not automated their process yet but have, for instance, a portal where they go to find the various system components and forms they need. A decision-making component could be developed to plug into this portal without needing a automated process context.

They see customers adopting Decision Management tactically, incrementally and strategically:

  • Tactical efforts involve fast pathing regular or normal transactions (by automating the decision to fast path a claim say) or coordinating decisions across multiple processes
  • Incremental efforts involve externalizing decisions currently embedded in legacy applications or business processes (where legacy code or spaghetti processes “manage” the decision today) or where a manual decision limits straight through processing
  • Strategic efforts involve building reusable decision services in a systematic way, customer-centric process redesign (webinar here) or by making processes adaptive.

Check out these webinars on applying Decision Management to legacy applications, on simplifying complex processes and on smarter enterprise applications like ERP for some suggestions on how/why to do this.