James Taylor's Decision Management

James Taylor

Decision Management Concept #1 - Definition and process

user-pic
Vote 0 Votes

It seemed like time to do a series of posts on the basic concepts of decision management. The series will look like this:

  • Definition and process
  • Decisions
  • Decision Services
  • Business Rules
  • Data Mining and Analytics
  • Adaptive Control
So, here goes.
Decision Management is an approach for automating and improving high-volume operational decisions. Focusing on operational decisions, it develops decision services using business rules to automate those decisions, adds analytic insight to these services using predictive analytics and allows for the ongoing improvements of decision-making through adaptive control and optimization.
To make decision management happen you need to follow a pretty straightforward process:
  1. Identify Decisions
    The first step is to identify and make explicit the decisions within a process or system. These decisions may be known already or hidden in the process. Known decisions, such as whether to approve a loan or identify someone as eligible for a product, may be made manually by people or embedded into systems. These are typically easy to identify. Hidden decisions are more challenging as they have typically not been identified before. Hidden decisions may include decisions which have historically been taken once for the organization but which could be taken for a specific customer or process instance. These are sometimes called “micro decisionsâ€? and include such decisions as pricing (moving to customer-specific pricing instead of generic pricing) and recommendations for next action (becoming personalized and conversation-specific).  Other hidden decisions include personalization of a process or interaction, inconsistent actions across channels and implicit decisions made by people without any awareness of the decision and thus no policies or measures.
    If you let the decision stay embedded in the system or the process then you will never manage to improve it. This does not mean disconnecting it from where it is used or ignoring how it is used by systems or processes, it just means identifying the decisions as unique aspects of a system, decision services if you will, and focusing on them as something that can be managed and improved separately.
  2. Externalize Decisions
    Having identified the decisions they must be externalized. This means removing the code that implements them from other systems – disentangling them – and explicitly identifying them in processes. For manual decisions this means documenting the policies and procedures, as well as the information, that go into making the decisions. A decision audit may be conducted to document and describe each decision within a process, a business area or even a whole organization. Such an audit identifies who makes the decision, who decides how it should be made, how often it is made, how much time is available to make it and other key attributes.
  3. Automate Using Business Rules
    The policies and procedures that drive how a decision is made can and should be represented using a declarative definition of what should be done. Business rules, either as part of a process or application environment or in a Business Rules Management System, are ideal for this. Business rules are declarative, expressive and easy for non-technical users to understand, making it possible to define how a decision is made completely and correctly. While many decisions are automated to the point of completion – the automation decides on an action and then causes it to be taken – many are not. The automation of a decision may simply identify acceptable options (or unacceptable ones) or direct a manual decision-maker to certain information resources. One consequence of this is to give business users control of how the decision is made.
  4. Improve the Rules with Data Mining
    While many of the business rules required to make a decision come from written and explicit sources – regulations, policies, procedures, user preferences etc – it is often possible to use data mining to create and improve business rules. Many rules have thresholds or limits in them and data analysis can be used to find statistically significant or historically effective values for these. Data mining can also be used to develop segmentation rules – decision trees – and these are often represented by Business Rules as well as affected by policies and regulations. Essentially one can make the rules better reflect what works, or at least what has worked in the past, by analyzing data.
  5. Add Predictive Insight
    One of the main challenges when making decisions is the uncertainty inherent in guessing what will happen in the future. To treat a customer right, set a price correctly or select the right supplier one needs visibility into the future – how will the customer react to each treatment, what price will be acceptable and profitable, which supplier will deliver on time and to budget. Without a crystal ball certainty as to the future is impossible to come by. However, we can turn uncertainty into probability using predictive analytic techniques. Information about the behavior of a specific customer, and of customers statistically similar to them, can be used to predict how likely they are to react favorably to a particular offer. Information about suppliers and their past deliveries can be used to predict how likely they are to be on time in the future. Such predictive models can be developed and then integrated into the decision making so that the rules for the decision use the likelihood of relevant future outcomes.
  6. Continually Improve with Adaptive Control
    Decisions are not stable. What makes a good decision changes over time as markets move, competitors change their behavior and as the expectations of customers change. Effective decisions then must be managed over time and continually improved. The most effective way to do this is to use adaptive control techniques to continually compare the “championâ€? approach – the one most likely to be the best – with “challengersâ€?. These challengers change some aspect of the approach – new models, new rules – and are applied to a small percentage of transactions. If one of them performs better over time than the champion it can be promoted to the champion and new challengers devised. This constant challenging leads to continual improvement and helps ensure that changes outside the organization don’t lead to a decision being made poorly for an extended period.
Tomorrow, some thoughts on decisions.

James Taylor blogs about decision-management technologies such as predictive analytics and business rules, discussing how they deliver agility, improve business processes and bring intelligent automation to SOA.

James Taylor

James Taylor blogs on decision management for ebizQ, and is an independent consultant on decision management, predictive analytics, business rules, and related topics.

Sponsored Links

Fico

Subscribe

Recently Commented On

Recent Webinars

    Monthly Archives

    Blogs

    ADVERTISEMENT