Operational Business Decisions Whose Decisions Are They Anyway?
What is an operational business decision? To answer let's start with the meaning of decision. For the purpose of business analysis the best understanding of decision comes from the Merriam-Webster Unabridged Dictionary (MWUD) definitions below. Note the prominent appearance of reasoning in the definition of determination.
decision 1b: a determination arrived at after consideration : SETTLEMENT, CONCLUSION
determination 2: the resolving of a question by argument or reasoning
We purposely have not chosen MWUD definition 1a: the act of deciding; specifically: the act of settling or terminating (as a contest or controversy) by giving judgment. When I say decision I don't mean the act of deciding; I mean something that is decided.
Big difference. Where can acts of deciding be found? They are often represented by tasks in process models.
Analyzing the something that is decided is a distinct problem. For that you need appropriate techniques — e.g., Question Charts (Q-Charts™) and decision tables. These techniques enable you to develop the decision logic needed to support the act of deciding.
Where Traditional AI and Expert Systems Differ from Business Rules
Fundamental to the meaning of decision is that reasoning is involved. As discussed in last month's column, a knee-jerk reaction to an event based on applying (testing) a behavioral business rule does not require true reasoning. Such an event is not a decision.
Aside: A behavioral business rule is a business rule that places an obligation on people, rather than a necessity on knowledge. A necessity on knowledge is called a definitional business rule. Behavioral rules can be violated; definitional rules can't.
Here's where traditional AI and expert systems of the 1980s and 1990s see things a bit differently. They argue that people can react to situations without apparent reasoning due to practice. A person, for example, doesn't need to reason about sipping from a coffee cup. It's automatic, the reasoning long since 'compiled' into autonomic behavior. Way back though when first learning, the person did have to think (reason) the problem through. So reasoning is nonetheless manifest. If there's reasoning (of whatever kind), it follows from the definitions above there must be some decision.
- A person sipping from a coffee cup would never describe what he or she is doing as making any decision. In business analysis we need to talk about the world the way business people see it, not some artificial rendition thereof.
- Just because reasoning occurred in the past (indicating a decision of some kind) does not mean that reasoning (i.e., some decision) occurs in the present. Business analysis is not the same as business operation. Obviously some thought (reasoning) must have occurred when a business rule was originally created. Business rules don't spring forth immaculately (good ones anyway). That kind of analysis (reasoning) though is very different from the reasoning needed to make operational business decisions in the present.
- Business rules are not about mimicking the intelligent behavior of some individual; they are about running a business. The former is a science initiative; the latter is an organizational endeavor. Mimicking the intelligent behavior of an individual in a generalized way is far harder (order of magnitude or more) than capturing organizational (business) rules.
Another important question for decision analysis is: Whose decision is it? Let's take an example. Suppose your organization has the business rule: An international flight must be taken in tourist class while on company business.
As an individual actor (human), you always have a choice about whether to obey a behavioral rule. Because choice is involved, that is a decision. But it's your personal decision, not an operational business decision.
I hate to break the news to you but, in general, organizations don't really care very much about how you make your personal decisions. In this example, the business intent of the organization is clearly that you not have any decision to make about your ticket class.
Business rules always take the organizational view. Although you personally can decide to violate a business rule like the tourist-class-only one, that's your business. (Of course if the company catches you, the matter does become their business. One way or another it will probably make you pay.) In decision analysis when we say decision we really mean operational business decision, never some personal decision.
Frequently Asked Questions
- When is there an operational business decision in real-world business?
Only when a choice presents itself between alternatives in day-to-day business activity.
- Do operational business decisions fall into categories?
Yes: classification, evaluation, selection, approval, assessment, assignment, allocation, diagnosis, prediction.
- Do other communities involved with decisioning look at it this way?
Predictive analytics, the closest related discipline, generally does.
- Where do business rules originate?
A great many business rules can be harvested directly from agreements, contracts, deals, regulations, laws, business policies, etc. (Or perhaps subsequently from legacy systems or sticky-note-ware, where the business rules finally came to rest.) Such business rules typically do not involve operational business decisions in the sense of an organization having a real-time choice between alternatives.
- Can decision analysis uncover all your business rules?
No, not even close. Although decision analysis helps you discover a great many business rules of a certain kind, by no means would all ever be found that way. Maintaining an exclusively decision-centric view of the world throws business analysis badly off course, almost as much as an exclusively process-centric view (or worse, just doing use cases!).
- Does it matter whether decision logic is evaluated by an automated platform (e.g., a BRE) or a human?
No. If you were to turn the machine off, the evaluation (reasoning) still has to happen (assuming the business can operate at all). How decision logic is deployed makes no difference whatsoever in determining whether you have an operational business decision.
To express rules, traditional AI / expert system people (knowledge engineers) tend to think in terms of syntax following the form condition-action or event-condition-action. These syntaxes don't work well for behavioral business rules.
From an organizational perspective, a behavioral rule involves three distinct things: the rule itself, detection of violations, and responses to violations. It's unnatural to stuff all that into a single expression. Look again at the business rule: An international flight must be taken in tourist class while on company business. That's a business rule statement in natural form. Where's any action (response to a violation)? Where's any event?
In establishing collective rules for people in society, organizations, or businesses, detection of violations and responses to violations (e.g., sanctions) are handled as separate concerns.
To prove that to yourself, have a look at any written law in your state. Examine any clause in any contract your company has ever signed. Consider any rule you've set for your kids at home. In what real-world sense does it help to see those rules as doing reasoning (and thus being decisions)?! None whatsoever. About the only way you would come to that view is by looking at them through the lens of a poorly-suited syntax or a paradigm created for a different purpose.
 Refer to Decision Analysis Using Decision Tables and Business Rules by Ronald G. Ross, a 76-page in-depth white paper available free on http://www.brsolutions.com/b_decision.php
 Ronald G. Ross, "The Anatomy of Decisions: The Business-Rule View," Business Rules Journal, Vol. 12, No. 2 (Feb. 2011), URL: http://www.BRCommunity.com/a2011/b577.html
# # #