Click to learn more about author Kimberly Nevala.
Descriptive. Predictive. Prescriptive. Perceptive?
Artificial Intelligence (AI) provides exciting new opportunities for exploiting data and information. But is AI a revolution or natural evolution of existing BI/Analytics use cases? Certainly some of the core techniques that power AI (advanced Machine Learning and Deep Learning in particular) represent a fundamental shift in algorithmic programming. The application of AI, however, answers a need previous analytic generations could not. To see why, let’s start at the beginning.
First, there was Descriptive Analytics: aka Business Intelligence (BI). Utilizing information generated as a byproduct of operational systems, Descriptive Analytics provided a rear-view window on business performance. How many sales did we have? Which patients suffered a medical complication? How many vehicles were recalled? What was our revenue?
As these questions were answered the next logical query raised its hand: “What will happen next?” Enter Predictive Analytics: projecting future outcomes. Predictive Analytics began with simple forecasting models in which past performance was used to project future outcomes. More advanced techniques and improved access to data allowed predictions to become more sophisticated. Thus taking into account not only historical actions but also elements such behavior, location and time. Predictive Analytics answered questions such as: What is our projected revenue for next quarter? Which patients are likely to be non-compliant with their prescribed treatment? What shows are you likely to watch next? Which products might you also need or want based on what “people like you” have bought?
Knowing what might happen next is nice. Knowing what to do about it is even better. Cue Prescriptive Analytics: Assessing the likely outcome of different actions to inform the best course of action to achieve an optimal business result. With Prescriptive Analytics, optimization is the name of the game. Questions addressed by Prescriptive Analytics include: Which offers is this customer most likely to respond to? What is the best intervention for this patient? Which combination of what proffered to whom optimizes outcomes for the individual, the cohort and/or the corporation?
At this point, there are seemingly no new questions. Analytics can address what has happened, what might happen next and what action to take. So what’s left to do? Despite much ado about democratizing access to data delivering consumable, contextually appropriate insights remains a challenge. Therein lies the rub: The best insight is for naught if it does not lead to action. Enter Perceptive Analytics (my term). With Perceptive Analytics, systems do not merely passively analyze but actively engage and interact with their environment. As such, the business questions Perceptive Analytics systems (AI in particular) answer are not different than those defined above. What is radically different is how AI systems are developed and deployed to enable action.
With AI, machines (aka algorithms) learn from, monitor, and respond to events (aka data input) in their environment with less direction and supervision. This has allowed many tasks previously thought to be the sole purview of humans to be automated. More importantly, AI provides the ability to deploy analytics systems that communicate in the same ways we do – in “plain English”. Armed with the ability to hear, see, speak and write AI systems facilitate an active dialogue between human decision makers and information that focuses attention on: What insight do I need right now? What else should I consider? In this way, AI solutions bridge the gap between creating and deploying impactful analytics. It is this capability which may, ultimately, deliver on the promise of a data-driven organization: one in which access to contextually appropriate insight is not limited by an individual’s technical expertise or background.
Each Analytics progression has created new opportunities for competitive advantage and meaningful engagement. There is no question AI’s transformative potential will eclipse its predecessors. Yet the seeds of that development and, more critically, the appetite for AI were planted in the work that came before. Even so, adding a new floor to your Analytics house will take an architect with vision and an investment of time and money. But if your organization is practicing analytics today you have unwittingly begun to lay a good foundation to build toward AI.