To understand the decision making process, let’s first define a few key terms:

1. Forward-looking decision making – decisions that are not only about current issues but also about positioning the organization for future success.
2. Rapid – the clock speed for businesses today are much faster than before, and agility is a key requirement for companies to succeed.
3. Constrained – this refers to an organization that does not have infinite resources at its disposal (resources like cash, inventory, time, labor, capacity etc).
4. Trade-offs – given the need to make decisions rapidly and the fact that resources are constrained, there is a need to make trade-offs across differing priorities
5. Available data – this refers to the data that companies need to make decisions. In fact, lack of trust in the accuracy of data makes companies rely on gut feeling for decision making instead of relying on analytical approaches.

Three Methods to Enable Decision Support

Now let’s take a look at the different types of analytics capabilities that can be utilized to address the need for decision support. We’ll look at three main analytics capabilities: Heuristics-based modeling, optimization and Business Intelligence / Big Data. Each has a set of advantages and disadvantages when it comes to using them in an organization. It’s crucial for companies to understand the limitations and appropriate use cases when they’re searching for the best way to answer certain questions.

Figure 1 shows a mapping of the three types of analytics capabilities we’ll discuss and their ability to address certain components of decision support.

Figure 1. Mapping of Needs to Capabilities

Heuristics-based modeling

• Logic defined by users
• Works in sequence
• Difficult to implement advanced constraints (i.e., throughput, minimum batch production, etc.)
• Leads to an answer that may or may not be feasible
• Takes effort to maintain the rules (not just the data) as product mix and throughput change
• Provides limited insight beyond the answer

Typical challenges: Does not have ability to do trade-off analysis and constrained decision making.

Optimization modeling

• Accounts for limits in sales, logistics, production, stocking, etc.
• Answers questions around what combination of activities is best?
• Flexibly maximizes/minimizes objectives (i.e. profit, demand shortage, cost)
• Leads to the best possible answer
• Typical improvement of 10-20% over heuristics-based modeling
• By definition it provides an answer that is also doable
• Provides additional insight into marginal opportunities around demand and business constraints

Typical challenges: Requires a larger amount of data than other approaches and usually takes longer, due to a larger number of combinations on which the algorithms act.

Business Intelligence/Big Data

• Provides ability to analyze large amounts of data
• Very focused on visualization and story telling
• Answers questions like ”What has been the historical performance of the organization?” or “What has the company been good at doing in the past that could potentially be replicated?”
• Does not provide forward-looking insights.
• Hard to outline business value without looking at specific use cases

Typical challenges:  Not forward looking nor focused on constraint-driven decision making and trade-offs.

Closing Remarks

All of these methods are valid approaches within organizations to apply analytics- and mathematics-based decision support capabilities.

Organizations should look at having a portfolio approach to utilizing these solutions and decide what approach suits their individual business challenges the best.

Also, a strong focus on Return on Investment is critical in determining which application of analytics is best to achieve particular company goals.

Editor’s Note: This post was originally published March 31st, 2016 and revised September 30th, 2018.