Traditional BI tools, which use descriptive and predictive analytics, provide insights based on accumulated data. Prescriptive analytics, which is the latest stage in the field of business analytics, goes far beyond by proposing actions that the business can take to maximize profit or mitigate risks. It can be considered as the apex of a cutting-edge analytic capability. Prescriptive analytics can assist leaders in making critical decisions by providing an unconventional roadmap that not only identifies the optimal decisions, but also the impact of each decision.
The Analytical Decision Model
Before we look at the different methodologies of prescriptive analytics, we must understand the significance of an analytical decision model. The model represents the components involved in the making of a decision:
- The decision and the choices: What is the decision that we need to make? What are the possible choices against the decision? A simple example of a decision and possible choices is as follows Which offer should we give to the customer (Offer A, Offer B or Offer C).
- Information relevant to the decision: Do we have information that will help us make the right decision? For example, does the call center rep, who is a decision maker, have access to the customers profile?
- Objective of the decision: What is the outcome we would like to see from a decision? For example, can the decision help us increase revenue for the current quarter?
Different methodologies of prescriptive analytics
There are several methodologies of prescriptive analytics such as optimization, game theory, simulation, decision-analysis methods and control systems, to name a few. It is observed, however, that the following two types are used by most organizations:
A: Predictive Analytics Plus Rules
All prescriptive methodologies tend to follow an analytical decision model. In other words, they have the three components involved in the making of a decision. The Predictive Analytics Plus Rules methodology is simpler, however. It combines predictions with business-defined rules and assumptions without considering all conceivable outcomes. Thus, it may not suggest the best possible action always. This methodology tends to be adopted to make operational decisions.
Consider an example: The action point is to make an offer to the customer. To identify the best offer that can be proposed to the customer, a predictive model can be made for each offer. This will measure each customers inclination to respond to the offer. A rule can be made to propose the offer with the highest customer response to every customer.
Although business-defined rules make this methodology prescriptive, the use of data and analytics makes it prescriptive analytics. This combination makes the rules smarter and the analytics actionable.
B: Optimization
The second methodology of prescriptive analytics is optimization. Usually used in making strategic and tactical decisions, the methodology is also applied in making operational decisions. It follows an analytical decision model to determine all conceivable outcomes of each alternative and evaluates the scope of compromise among numerous objectives. It determines the ideal usage of limited time and resources in situations that have different levels of uncertainty.
A problem in optimization has three key elements:
- Objective The metric that needs to be optimized. For example, an objective could be to maximize profit for the current fiscal year.
- Decisions The decisions that need to be made to achieve the objective. For example, a decision could be made on the amount of financial investment.
- Constraints They are restrictions or limits that can influence a decision. For example, the size of the budget can limit the scope of a decision.
Consider the example used to explain the first methodology. Against the action point (make an offer to the customer), the analytical decision model would provide all possible choices (send offer A to customer today by email, for example) and all possible outcomes (response probability, for example). The user can then compare one combination (action point, choice and expected outcomes) against all other combinations to identify the ideal action point.
As the complexity increases (more action points and more outcomes), so does the need for sophisticated algorithms that can compute efficiently. This need is addressed, thanks to major developments in the field of optimization algorithms.
Some of the common applications of this methodology in the retail sector are as follows:
- Optimize promotional campaigns Organizations can decide which campaigns to execute and for which products while considering business constraints like budget, channels, inventory, resources, etc.
- Optimize product assortment Organizations can decide the ideal proportion of premium, medium and low-range products in order to generate the maximum value. The model considers factors like product cost, demand, and substitution effects.
- Optimize layouts Some retail chains are using the prescriptive method to determine the ideal store layout in order to increase sales.
- Optimize pricing As the prescriptive method helps organizations to identify and understand patterns and insights, they can be confident in making pricing decisions.
Use Various Analytics Techniques to Augment Your Decision-making
All analytics capabilities are part of the process to improve decision-making. When they are used in concurrence, they provide support for various parts of the decision-making process.
You start with the present where you need to make a decision. Using descriptive capabilities on historical data, you attempt to understand the scope and background of the decision. Before you take a final call, you will want to review all the conceivable outcomes. You can achieve this effectively and accurately with predictive analytics. The final step is to identify the ideal course of action. You can achieve this based on your gut feeling, business-defined rules that embody best practices or prescriptive analytics. Insights gained from descriptive and predictive analysis should be leveraged in this step.
The cycle does not end when you make a decision. On the contrary, it repeats itself. After a decision is made, you will want to evaluate the impact of the decision. For this, you will use descriptive analytics. Thus, we see that the use of analytics capabilities in concurrence at various stages in the decision-making process can help organizations achieve the following:
- Bring various business processes under comprehensive control;
- Address a larger set of business problems more effectively and in greater detail;
- Create and promote a decision-making process driven by data;
- Improve the organizations analytics maturity.
Conclusion
Early analytics helped organizations understand the reason for events that took place in the past and predict events yet to occur based on accumulated data and business-defined rules. Prescriptive analytics enables organizations to go a step further by suggesting actions that can produce the desired results. As a result, organizations of all hues are embracing the opportunity that prescriptive analytics presents.
It is important, however, that organizations engage in due diligence before committing to any implementation. The prescriptive analytics methodologies described in this article differ from each other significantly. In order to identify the appropriate type of prescriptive analytics for their organizations, leaders must assess the nature and complexity of their problem.
