Why You Should Move From Predictive To Prescriptive Analytics

By Dave Smith




With all the hype about Big Data and predictive analytics, we are seeing a plethora of technology providers tout how great their analytics capabilities are. Analytics is the new black! Predictive analytics will not completely replace human decision making, and it shouldn’t. It will ultimately come down to how prescriptive applications can be used to augment and provide insights, suggestions and recommendations. People will still have to make decisions, but having the applications get smarter is a revolutionary trend impacting all business disciplines.


Applications can now recognize human or machine patterns and make relevant recommendations in context. However, the applications will have to be tuned and trained over time to start making correct and effective recommendations. Also, it is not enough for machine-learning algorithms to enable applications to predict what will happen; it has to prescribe specific actions deeply related to the business domain. It has to be in the context of the particular business process.



I’m not trying to wage a semantics war between predictive and prescriptive analytics. What I am saying, though, is that where predictive analytics ends with forecasting the future based on data modeling, data mining and machine-learning techniques, prescriptive takes over to literally prescribe an action that a business leader or decision maker can act on. At the end of the day, the business decision maker has to act.


Having better analytics and understanding about customer behavior, their issues and how and why they’re contacting the company can provide insights that can lead to better decisions about improving the customer experience across all business operations and channels.

Predictive analytics will help you to predict multiple possible outcomes of what might happen based on the insights from data. Prescriptive analytics will take that predictive model and the loop of decisions taken with past outcomes and prescribe the best course of action to take for a particular business scenario. The prescriptive model will enable you to understand what the possible consequences are for particular actions. Think of the Google self-driving car—based on an innumerable amount of data points, calculations and predictions, it can determine which route to take and whether to go left or right at an intersection. That’s prescriptive analytics. The car will know the consequences of going left or right in traffic.


With a prescriptive model, data linking is important to provide context. No matter how “Big” the data is, without context, that data literally has no soul. Context is where the meaning and value comes from and what leads to better decisions.



For the business decision maker, prescriptive analytics is priceless. For the human resources professional, who needs to make quality hires quickly, insights from analytics that are prescriptive on who to hire can dramatically improve the talent acquisition process. I recently spoke with a sales executive who explained how he is applying prescriptive models for lead assignments. Sifting through all the variables and possible outcomes to recommend the right leads to go after is a game-changer in sales.

MORE: Solving the Leadership and Employee Engagement Conundrum


Imagine what this could mean for customer interactions. Having better analytics and understanding about customer behavior, their issues and how and why they’re contacting the company can provide insights that can lead to better decisions about improving the customer experience across all business operations and channels. Having better prescriptive analytics in talent management can help understand talent readiness, spot leaders and suggest the best course for engagement.



I’m not trying to oversimplify prescriptive analytics here. It’s a complex approach that will require work and investment. It will involve lots of data and the right algorithms that have to be programmed to adapt to changes without human control. These algorithms will be automatically optimized, so that over time, they will produce better predictions that will be the basis for better prescriptions or recommendations. This is disruptive and will impact business decision makers in all business applications and processes.






Dave Smith is the research director and lead analyst for collaboration at Aragon Research. Previously, Mr. Smith was a research analyst at Gartner, where he covered collaboration and web conferencing. Follow him on Twitter @DaveMario.