Everyone knows that enterprise software is an intensely competitive business. Today “Big Data” and “Analytics” are two of the hottest topics in organizations of any size and as companies race to invest in this new space, product managers must find ways to respond. The functionality of enterprise software products must be extended to include new analytic capabilities.
Everyone knows that enterprise software is an intensely competitive business. Today “Big Data” and “Analytics” are two of the hottest topics in organizations of any size and as companies race to invest in this new space, product managers must find ways to respond. The functionality of enterprise software products must be extended to include new analytic capabilities.
While most product managers have already embedded business intelligence capabilities and up to date performance monitoring, the velocity, variety and volume of Big Data can overwhelm these traditional analytic approaches. Add the pressure to build software that responds in real-time and it should be clear that more advanced analytics are needed. Indeed I believe that the future belongs to product managers that can embed predictive analytic capabilities in their products, pushing beyond reporting and dashboards. In this series I am going to lay out a four step program for doing just that. But first, step 0: understanding the power of predictive analytics in software products.
A typical software product manages a lot of data. If customers have been using it for any length of time then it’s probably managing a lot of history. It manages years of transactions, changes, updates, customer additions and removals and much more. What it does not contain is any sense of the future. What these customers might do in the future, however, is uncertain. This uncertainty about the future limits what a product can do. Traditional reports and visualizations can help a user make predictions by helping them see patterns in the data. Yet humans struggle to consider how more than 5-10 pieces of data are varying over time. This is not a limitation in predictive analytic algorithms. As software products adapt to the world of Big Data, hundreds or thousands of pieces of data are becoming available. In these circumstances the ability of predictive analytic techniques to consider all these pieces of data to determine what matters most becomes critical.
The value of eliminating uncertainty, or at least of replacing uncertainty with a probability, is that it improves the accuracy of decision-making. There’s no value to a software product delivering a prediction unless it improves a decision—simply knowing something is not enough, a decision to act based on that knowledge is required. If a software product is to be better than its competitors thanks to predictive analytics then it must use predictive analytics to improve decision-making. The question is how to go about this.
image: software/shutterstock
