The phrase "big data explosion" is quickly becoming a cliche in the business intelligence space. By now, those who are familiar with how quickly data is being generated are well aware that the phenomenon is not so much a momentary explosion as it is a steady, powerful stream that continues to build momentum and magnitude. Increasingly, businesses are not only looking for solutions to help them analyze the massive amount of data their customers are continually producing, but also to assist them in making predictions based on the available information, and transform those predictions into triggers for the right course of action. With growing demand for more sophisticated, forward-looking solutions, vendors are taking basic descriptive analytics to the next level with predictive and prescriptive solutions. But where do predictive analytics end and prescriptive tools begin? Answer that, and you'll know what came first, the chicken or the egg.
On the surface, there is a linear nature to the relationship between predictive and prescriptive analytics. "Predictive analytics can help you forecast what might happen in the future based on data that was gathered in the past. Prescriptive analytics take it a step further by suggesting what to do next based on that future insight," Stefan Schmitz, vice president of product management at MicroStrategy, a business intelligence platform provider, explains. "So while a predictive analytics model might tell a retailer that they are likely to sell 5,000 green shirts to women next month, prescriptive analytics might inform the retailer how to best position and market green shirts to meet that future demand," he adds. Schmitz's definition is a classic take on the difference between the two.
The Shortest Distance to Personalization Is a Straight Line
Many of the solutions on the market today promise to take data through a similar progression and offer a variety of measurement, visualization, and personalization tools along the way. For example, Adobe Analytics, which was recently named the leader in the Forrester Wave report on Web analytics, offers an intuitive and logically progressing analytics workflow that leverages the joint potential of both predictive and prescriptive analytics solutions, according to the report.
Adobe's analytics process begins with monitoring tools that watch for any fluctuation in data—these are reporting, or descriptive, analytics. For example, when conversions noticeably dropped for one of Adobe's retail customers, an outdoor and sports equipment store, the analytics tools within Adobe's Marketing Cloud picked up on the change and alerted the customer in an insight report. Everything that happened past this point required predictive and prescriptive tools, John Bates, senior product manager for data science and predictive marketing solutions at Adobe, explains.
Once descriptive analytics identify that there is reason for concern, the predictive capabilities kick in to diagnose the problem. The solution uses clustering algorithms to identify the audiences that are driving the conversion problem by evaluating individuals across different criteria rather than evaluating audience segments formed through subjective cluster definitions, Bates explains. For the sporting goods store, the predictive clustering revealed two distinct audience groups—those who had a high probability of conversion and those who were still likely to convert, but were "on the fence," according to Bates. The predictive tools also enabled the retailer's marketing team to score each shopper on his or her likeliness to return to the page to complete the purchase.
The marketers can then transition into the prescriptive analytics phase, which takes place inside Adobe Target, another solution that