Decision Management Loops Back to Decision Support Systems

September 29, 2012
207 Views

Manual DecisionsJames explained the difference between Decision Managemen

Manual DecisionsJames explained the difference between Decision Management and Decision Support Systems in a recent post.  Although it was a solid picture of manual versus automated decisions, I think it is missing half the story…  Let me elaborate.

Current state

In summary, James makes the point that repeatable decisions, that cannot afford human intervention, should be handled by a Decision Management System.  There is no argument there.  It was the very reason that those systems were created in the first place.

The issue I see in such an abrupt dichotomy between automated and human decisions is that, as long as we keep doing it this way, the 2 systems will not leverage each other.  At best, automated decision services document the identified red flags or opportunities identified by the business rules before deciding to refer the decision to an expert.  From that point on, the manual decision ignores the body of knowledge accumulated in the form of decision logic, and turns to other sources like wikis or simply expertise.  There is hardly any investment happening today in looping back these manual decisions into the repository of decision logic.

Why they should eventually merge

If, alternatively, we considered the world of decisions as a single “decision support system” in the literal sense, we would be able to better connect the dots.

  1. From Automated to Manual: If the case worker could see more than stored rule execution results, and explore the business rules that triggered those flags “How high risk this high-risk applicant?”.  Rules execute or they don’t.  Seeing that the application was borderline may influence further the case worker, and help him or her make the best decision.  Furthermore, in isolation, a manual decision for one customer or one transaction can seem harmless, but assessing how it would impact the business if operationalized can yet again help guide the manual decision.  Business Intelligence tools provide some of this insight of course, when they are well-integrated.  The only provide exploration tools on the distribution of your portfolio though.  It would be much harder to assess how many dollars would be involved if you granted a discount to all applicants with such and such characteristics.  This is where Decision Management shines, allowing the operator to actually run decision logic and simulate the results, including any post-processing calculations or predictions.  In the hands of this expert making manual decisions, it would him/her further.
  2. From Manual to Automated: Once the decision has been made, the case worker moves on to the next decision, the next case.  So what happens to the intellectual effort that was put together for this manual decision?  Gone…  To be fair, it will linger for a little while in the head of that case worker for a while.  Unless that person is extremely diligent and documents every case carefully and then shares them with colleagues, it is unlikely that this knowledge will be propagate beyond that one person.  Manual decisions are delicate decisions, often applied to the high-risk / high-value cases.  That decisioning logic, although it may only be applicable to one exception, should be captured though in ‘decision logic’ format (complete or incomplete) such that It can become a candidate for automation down the road.  Peers would be able to benefit from the insight that “according to how we have been processing similar applications recently, the informal decision logic recommends to approve/decline”.  Management would be able to review how each case worker or how the group has been manually making decisions to operationalize those that seem the most repeatable and the most effective (meaning more revenue, less risk, better customer satisfaction, less complaints or any other set of metrics that are important to your business).

The main point I want to make it that decisions span manual and automated processing, but it is not an either-or situation, there are shades of grey, and an integrated approach to both can deliver huge benefits.

That reminds me of the discussion on Business Rules versus Business Intelligence, the rear view mirror story.  The key benefit is in combining both for better decisions.  Having a dashboard readily available when you are eliciting business rules gives you a tremendous advantage, compared to most organizations that can produce the equivalent report in a matter of weeks.  The more insight, the more effective.