Most supply chain analytics tell you what happened or what might happen next. But prescriptive analytics takes it further – it’s like having a seasoned logistics expert who not only predicts problems but hands you a detailed playbook to solve them. Think of it as the difference between a weather forecast that predicts rain and one that tells you exactly when to reschedule your shipments, which alternate routes to take, and how to adjust staffing to minimize disruption.

In this article, I’ll look at how prescriptive analytics can transform supply chain management. Indeed, it goes much further than other types of analytics like descriptive, diagnostics, and predictive. Further, I’ll explain why it’s more effective than other decision support and recommendation systems. Also, I’ll show how prescriptive analytics fits into the broader business analytics landscape. This will include real examples to highlight the key differences between predictive and prescriptive approaches. By the end of this article, you’ll understand exactly how prescriptive analytics helps supply chain leaders make better decisions.

1. Prescriptive Analytics Defined.

At its core, prescriptive analytics is a decision-support powerhouse that leverages historical data, knowledge-based inputs, and real-time data. Based on these digital inputs, it uses advanced algorithms to recommend a specific course of action, explain why it is the best, and details how to implement it. Moreover, businesses can use predictive analytics to automate decision flows, when appropriate. Below is Gartner’s definition of this type of analytics.

Prescriptive Analytics Definition

“… a form of advanced analytics which examines data or content to answer the question “What should be done?” or “What can we do to make _______ happen?”, and is characterized by techniques such as graph analysis, simulation, complex event processing, neural networks, recommendation engines, heuristics, and machine learning.”

Gartner

Indeed, unlike its analytical cousins, prescriptive analytics doesn’t stop at telling you that your warehouse efficiency might drop 20% next month. No, it can go much further.  For instance, continuing with the warehouse example, it can outline precisely which picking strategies to modify, how to reorganize your storage layout, and calculates the expected ROI for each suggested change. To sum it up, prescriptive analytics operates as a problem-solver that comes with its own action plan. Further, it can either provide a recommended action plan or through automation, actually implement the recommended solution. See below for a graphical depiction on how prescription analytics work.

Prescriptive Data Analytics Model

2. Prescriptive Analytics: More Than Just Another Recommendation System.

Research organizations and businesses started using the term “Prescriptive Analytics” back in the 1980s. So, this type of analytical tool is fairly new when you consider that human problem-solving itself goes back to the beginning of time. Additionally, prescriptive analytics is still evolving as it leverages emerging AI capabilities, new methodologies, and increasingly faster computers. As prescriptive analytics continues to mature, it is important to understand what it is and is not, Hence, it is useful to compare it to other types of analytical capabilities that focus on problem-solving and making recommendations. For instance, let’s first look at expert AI systems and recommendation engines. 

a. What Exactly Are Expert AI and Recommendation Engines?

First, below are definitions for these recommendation systems.

1) Expert System Definition.

“… is a computer program that is designed to solve complex problems and to provide decision-making ability like a human expert. It performs this by extracting knowledge from its knowledge base using the reasoning and inference rules according to the user queries.”

JavaPoint

2) Recommendation Engine Definition.

“… is a system that gives customers recommendations based on their behavior patterns and similarities to people who might have shared preferences. These systems, also known as recommenders, use statistical modeling, machine learning, and behavioral and predictive analytics algorithms to personalize the web experience.”

TechTarget

So, in both cases these solution-oriented systems are more limited, geared to providing recommendations only. Also in most cases, these systems are designed for one-on-one human-machine interface. Hence, they are not really designed to work in an organizational setting. Next, let’s compare these systems with a prescriptive analytics capability.

b. How is Prescriptive Analytics Different from AI Expert and Recommendation Engines?

Now, for comparison purposes, I’ll identify the key difference between prescription analytics and these other recommendation systems. See below:

1) Expert System Vs Prescriptive Analytics. 

Essentially, an expert system focuses on replicating human expertise through deterministic knowledge-based rules. Basically, you ask the AI expert a question, and it gives you an expert answer. Note: An expert system may seem similar to an AI Large Language Model (LLM) like ChatGPT, but it isn’t. This is because LLMs are not designed to be expert systems. Instead, they are more general-purpose using statistical pattern recognition and prediction to generate responses.

Now on the other hand, Prescriptive Analytics does provide advice like an Expert System. Further, it also geared to do a lot more to include providing the why behind its answers and detailing how to implement its recommendation. To summarize, a full-fledged Prescriptive Analytics capability will recommend a course of action, explains why it is the best course of action, and details how to implement the recommended solution.

2) Recommendation Engine Vs Prescriptive Analytics.

Now, let’s look at recommendation engines. Indeed, most of us are familiar with recommendation engines such as the system that provides movie recommendations on NetFlix. Other examples of these types of systems include product recommendations on Amazon, and on YouTube, it provides personalized selections of videos to watch next. 

So, a recommendation engine as we know them are usually narrowly focused on product or content.  Whereas prescriptive analytics is a lot more versatile where it can support a full range of decision-making scenarios and use cases. Also, as discussed previously, it not only recommends a course of action, but will tell you the why behind the recommendation and prescribes an action plan.

c. Prescriptive Analytics:  Business Decision Support that Answers What Should Be Done, Why, and How to Do It.

So to illustrate, Prescriptive Analytics empowers decision support systems to go to a new level. For instance, take inventory management – an expert system or a recommendation engine may only recommend restocking when levels are low. However, prescriptive analytics strives to go much further considering supplier reliability, seasonal demands, transportation costs, and warehouse capacity constraints to generate detailed procurement and distribution plans. Further, a business may have a predictive analytics capability that can automatically trigger orders, adjust safety stock levels, and reroute shipments based on real-time conditions.

For more details on expert AI systems, see JavaPoint’s article, What is an Expert System? Also, for recommendation engines, see Appier’s article, What Is a Recommendation Engine and How Does It Work?

3. How Prescriptive Analysis Works with other Analytics to Make Recommendations and Action Plans.

Prescriptive Analytics has unlimited potential to empower both business analytics and decision-making. At its best, imagine it as the conductor of an orchestra where descriptive, diagnostic, and predictive analytics are the musicians. When a supply chain disruption looms, descriptive analytics spots the initial signs, diagnostic analytics identifies the root causes, predictive analytics forecasts the impact, and prescriptive analytics composes the response. For instance, when a port delay threatens operations, the system doesn’t just predict delivery delays. Indeed, it automatically recalculates optimal inventory levels, suggests alternative ports, estimates cost implications of each option, and even initiates communications with affected customers and suppliers.

a. Prescriptive Analytics Techniques.

Prescription Analytics is not necessarily tied to any particular analytical model, technique, or methodology. One reason for this is because analytics is an evolving discipline as new innovations emerge. Also as with any data analytics capabilities, it is dependent on the data available, type of use case it is supporting, resources allocated, and skills of analysts available. To illustrate, below is a sampling of techniques that could be used as part of a predictive analytics capability.

Examples of Techniques Used in Predictive Analytics 

• Probabilistic Models. For example, use these models to assess financial risks in terms of percentages by estimating the likelihood of various investment outcomes.
• Machine Learning (ML) / Data Mining. In this case, ML can help optimize supply chain operations by predicting demand and prescribing actions to minimize costs and enhance efficiency. What’s more, it learns over time to improve its recommendations.
• Mathematical Programming. For instance, use these techniques to determine the optimal allocation of resources in production to maximize profit while minimizing waste.
• Simulation. Here, analysts can create virtual environments to test different strategies in order to predict and measure their outcomes.
• Logic-Based Models. Lastly, software developers can use these types of logic-based models to automate decision-making processes in complex systems like smart grid management, ensuring efficient and reliable energy distribution.

See Alexandros Bousdekis’ listing, Classification of the Methods for Prescriptive Analytics, for a comprehensive roster of prescriptive analytics techniques. Also, see AIMMS’ article, Prescriptive Analytics technology achieves recommended actions using optimization modeling. AIMMS is also an example of a company that offers a prescriptive analytics tool oriented on supply chains.

b. How Prescriptive Analytics Fits In With Other Types of Analytics. 

Most modern companies are familiar with or use Business Intelligence (BI) dashboards. However, there is a lot more to business analytics than BI dashboards and even prescriptive analytics. Indeed, there is a full range of business analytics tools available for decision-makers. Specifically, the major types of analytics include descriptive, diagnostics, predictive, and prescriptive. Further, each of these types of analytics focus on answering a specific question as listed below.

Primary Types of Business Analytics and The Question They Answer

• Descriptive Analytics: What Happened?
• Diagnostic Data Analytics: Why Did This Happen?
• Predictive Data Analytics: What Is Most Likely To Happen?
• Prescriptive Data Analytics: What Action Should We Take?

Also, each of these types of analytics can employ a full range of analytical tools and methodologies to help answer the questions they are designed to answer. At the same time, decision-makers do not necessarily need to use all these types of analytics. It depends on the situation. For instance, a decision-makers or business does not always need the full power of prescriptive analytics. At other times, decision-makers need a full range of analyzes to include descriptive, diagnostic, predictive, and prescriptive. In fact, in some cases a decision-maker will not only need the full range of analytics, but they will need each analytics type to mutually support each other. 

A useful chart for understanding how prescriptive analytics fits within other forms of analytics is the Gartner’s Analytics Continuum depicted below. This chart shows how each form of analytics supports both business decision-making and implementation as well as how they can mutually support each other, tied together by a common goal.

For more discussion on different types of data analytics, see my article, A Data Analytics Perspective To Better Empower Supply Chain Managers. This article will also highlight more forms of analytics to include on-demand analytics and AI-powered analytics.

4. Examples to Clarify the Differences Between Forward Looking Analytics: Predictive Vs Prescriptive.

As predictive and prescriptive analytics are forward looking, I think it is important to better understand the differences between them. To illustrate, think of these two types of analytics as two different GPS systems for your supply chain. One tells you there is likely to be heavy traffic on the road ahead (predictive), while the other provides you recommendations on how to take an alternate route (prescriptive). While both these forward-thinking analytics sound similar, you can see that they serve distinctly different purposes. 

To further clarify these differences, let’s look at some specific examples. Below are several excellent examples from Gurobi Optimization’s article, How Can Prescriptive and Predictive Analytics Work Together?, that clearly illustrates the differences in these two forward thinking analytics types. Specifically, one provides a prediction and the other one provides a decision (recommendation) with an action plan.

Examples of Predictive and Prescriptive Analytics Working Together

Predictive Analytics Types: The Best Opportunities For Supply Chains

a. Supply Chain Management.

• Prediction: Machine learning predicts potential supply chain issues with specific suppliers in a region.
• Decision: Mathematical optimization determines the least costly way to reduce shipments, considering various constraints from the supplier. This combination allows for proactive and cost-effective management of supply chain disruptions.

b. Investment Strategy.

• Prediction: Machine learning predicts investment opportunities.
• Decision: Mathematical optimization allocates limited cash across investments, considering constraints on investment amounts. This combination ensures optimal investment strategies.

c. Customer Engagement.

• Prediction: Machine learning predicts customers’ propensity to buy.
• Decision: Mathematical optimization decides how many coupons to offer to maximize revenue or profit. This synergy allows for targeted and effective customer engagement.

Supply Chain Analytics Types and The Way They Work To Better Empower Decision-Making

More References on Prescriptive Analytics.

• Gurobi Optimization’s link, FAQs Prescriptive Analytics, provides a comprehensive overview of prescriptive analytics.
• ThroughPut Inc’s article, Prescriptive Analytics for Supply Chain Optimization: 6 Powerful Use Cases, for more on prescriptive analytics and supply chains.
• Build In’s article, Top 14 Predictive Analytics Tools to Know, provides an overview to different software vendors that offer predictive analytics tools.
• For an example of how analytics works within a business decision-making framework, see my article, OODA – Enabling Business Agility: The Best Way To Disrupt Competitors, Seize Opportunities, And Overcome Obstacles.
• For more discussion on different types of data analytics, see my article, A Data Analytics Perspective To Better Empower Supply Chain Managers. 
• This article, An Agile Decision Platform to Empower Executives For Superior Supply Chain Performance: Here Are The Best Attributes,  illustrates how prescriptive analytics would work within a decision intelligence platform.

An Agile Decision Platform to Empower Executives For Superior Supply Chain Performance: Here Are The Best Attributes

For more from SC Tech Insights, see the latest articles on Data Analysis and Decision Science.