Prescriptive Analytics: From Data to Action

In the field of business analytics, the end game is analyzing data to generate action items, advice, or next steps. Or to put it another way, the only reason that businesses spend time and money on business analytics is that they want to use data to help them make better decisions more quickly. The problem is that many businesses have not fully automated this process.

Gartner talks about business analytics in four main categories: descriptive analytics, diagnostic analytics, predictive analytics, and finally prescriptive analytics. Most companies have implemented descriptive analytics systems like Qlik, PowerBI or others, and some are even using narrative generation technology to explain the analysis in a way that everyone can understand. Diagnostic analytics are at their heart still really descriptive, and predictive analytics, while an exciting field, are often seen by business leaders as a type of fishing expedition where no solid ROI can be calculated.

This leaves prescriptive analytics, this means telling you in real-time what to do as a result of data. In many ways, this is the holy grail of analytics. Imagine, you click on a button, and an algorithm scans a real-time data stream and tells you what to do as a result of the data. It may sound like a dream, but companies are already doing this, so the question is how do we get there?

 1. Codify your best practices & regulatory framework

This is not a technological step, this is a business process management strategy. You need to understand the best practices that your best employee is following when he or she makes good decisions from data. You also need to put on paper the rules and regulations that you must follow in your industry. This is not a revolutionary step, and most businesses do something like this anyway.

 2. Analyze Your Data through the Lens of your Best Practices

Once you have your best practices on paper, the next step is to use them in the analysis of your data. This means programming those best practices into a rules engine, or better yet, a more powerful inference engine that can apply those rules to data at incredibly fast speeds. So now you have your best practices, they are being used to analyze your data, but how do you explain the results of your analysis?

 3. Articulate Results of analysis and Advice

This is the critical step, and it is where conventional BI tools fail. When you are giving advice from data analysis, you need to explain why. A graph or a simple sentence that says “you should sell ACME stock now” won’t work. You need a system that can articulate the “why” to generate something like “you should sell ACME stock now, because it’s poor performance means you are opening up your portfolio to more risk that you should.” Obviously, that is just a sentence that I made up to make a point, but software that can explain analysis and advice is already being used in Fortune 500 companies globally.

In conclusion, prescriptive analytics fulfils the dream of business analytics: turning data into advice and next steps with the click of a button. And this dream is more attainable today than ever before.