Developing business decision-making models that really work

David Loshin defines weighting factors and offers advice for beginning a decision-modeling effort that ultimately leads to the best business outcome.

We have long cast various aspects of reporting, analytics and business intelligence in terms of "decision support." In fact, the early data warehouses and data marts used for BI applications that targeted senior management reporting were called decision-support systems.

But what does it really mean to "support" a decision? Are we employing commonly used business words to attempt to describe something that, in some ways, eludes description? By examining the process of decision making, we can consider two different aspects of what could be meant by the word support -- guidance for the decision before it is made and reinforcement of the process after the decision is made.

There have been many attempts to develop models that represent the business decision-making process, and it would be naïve to believe that these simple models are comprehensive. However, even simple decision-making models can highlight where potential gaps in knowledge can impede optimal decision making. Understanding what type of information is needed to make a particular decision is one step in considering the variables that are relevant for improving the process, as well as ways that the right amount of information can streamline or optimize those variables.

Decision-making models abstract some of the key facets of a decision-making scenario. Let's begin with a straightforward description of what is meant by a decision: a choice or determination made after considering a set of presented alternatives. The quality of that choice could be weighted along a number of dimensions, such as:

  • Informed-ness, a term we can use to describe the degree to which the decision maker has the right amount of information available at the point that the decision is to be made.
  • Timeliness, which measures the degree to which the decision is made with enough time for the desired actions to be taken.
  • Completeness, which reflects the degree to which the decision maker is confident that the appropriate decision has been made.
  • Precision of directed action, which characterizes how well-defined the list of triggered actions is, as a way of ensuring that the desired outcome is attained in a streamlined manner.
  • Optimality of outcome, which, as the ultimate measure of decision quality, describes the degree to which the decision has led to the best business outcome.

In turn, there are characteristic features of each decision-making scenario:

Business use case. This is a description of the steps of a process and the interactions between a set of actors. The use cases provide the context for the decision points. An example is the selection process for choosing a mobile phone and corresponding voice and data plan. There may be multiple use cases, including an in-store use case, a Web-based use case and one via outbound call-center marketing.

Actors. These are the participants in the decision-making scenario. There may be multiple actors, they may be humans or systems, and each may contribute to more than one decision. To continue our mobile-phone example, the actors include the salesperson and the prospective customer.

Business question. This specifies the question or questions that are the basis of the decision. In our example, the question is "What mobile device, voice plan, data plan and contract duration should I select?"

Inputs. This encompasses the information that is available to the decision maker at the point that the decision is to be made. When choosing a mobile device, the inputs might include the device's operating system, the speed of the device's processor, the availability of software for the device and whether the device has a slot for additional storage.

More expert insight from David Loshin

Find out why visualizing data in BI dashboards requires a measured approach

Read about Hadoop 2's expanded ability to support big data analytics

Get advice on how to manage an effective BI data quality program

Decision points. These are the points in time within the business process at which the business question is asked, choices are provided and an actor is pressed to make a decision.

Choices. These are the options available for selection as the object of the decision. In our example, the choices for a data plan may be $40 for 1 GB a month or $60 for 2 GB. Other questions would have their own sets of choices.

Desired outcomes. There may be multiple sets of desired outcomes, such as what is desirable for the business, what is desirable for the customer, what is desirable for affiliated vendors, etc. This enumerates the most coveted results. For example, the mobile service provider may want the prospect to opt for a higher-bandwidth data plan with the understanding that in many cases customers don't use their entire bandwidth allotment each month.

Performance measures. These are the variables used to classify how good the decision was, such as profitability, limitation of risk or improved customer experience.

I've provided a starting point for identifying the decision points within a business process flow and assessing what types of decisions are properly informed and will guide users to make choices that lead to the most-desired outcomes. In an upcoming article, I'll examine the application of decision-making models in greater detail and discuss how to adapt the process to understand what types of information can lead to better overall business performance and what types of analyses can provide that information.

About the author:
David Loshin is president of Knowledge Integrity Inc., a consulting and development services company that works with clients on big data, business intelligence and data management projects. He also is the author or co-author of numerous books, including Using Information to Develop a Culture of Customer Centricity. Email him at [email protected].

Email us at [email protected] and follow us on Twitter: @BizAnalyticsTT.

Dig Deeper on Business intelligence strategy