We all remember what happened to Shakespeare's Julius Caesar when he ignored the warning: "Beware the ides of March."
By contrast, in most industries today, people would not only not dismiss any insight into the future, they’d pay top dollar to anyone who could give them a slightly better glimpse into the fate of their business. It's here that in the last few years big data analytics have played the role of the soothsayer, culling data into meaningful patterns with probable futures. But the point of the Industrial Internet isn't merely to predict --it's also to advise. After all, what good is information if it doesn't influence decision making? Analytics are evolving into the kinds of oracles who don't just exclaim, "beware the idea of March," they recommend when, where, and how.
This class of analytics, known as prescriptive analytics, is a big focus of the Industrial Internet. Taking predictions derived from data and turning them into smart business strategies seems like an industrial holy grail. And yet there are larger questions here: What causes ineffective decision making in the first place? Can more information really stop it? Will decision makers actually heed the advice of analytic software? What new opportunities can prescriptive analytics open up?
Good Choices, Bad Choices
If you look at any one of several studies and insights into business decision-making, common themes emerge. First, good decisions almost always stem from well-structured decision-making processes. The best decisions weigh factors properly, balance short-term and long-term objectives, anticipate the ramifications of each choice, and advance the cause of an organization as a whole, not just a segment of it.
Prescriptive analytics are designed to consider not only past, current, and forecasted future data, but also overall business goals and objectives. The "prescription" comes as much from understanding the values sought by businesses and customers as it does from the interpretation of the data. The influence of this prescriptive power really shows in situations that are highly complex, involving multiple variables, incomplete information, competing objectives, and high opportunity cost.
Take a potential application of prescriptive analytics in oil and gas. Analytics software tracking data across an entire field could take the fluid flow information from hydraulic pumps in every well across the field, combine it with geological and seismic data about the entire site, assess the productivity of specific wells, and finally recommend choices on whether or not to stop tapping certain wells and open up new ones in specific locations, or whether to make small adjustments to increase production from the field as it is.
Decision makers are likely to make use of prescriptive analytics in cases like these because the situation is so complex it prevents an easy read on it by any one single person. But what about in situations where it's a bit easier to dismiss the soothsayer? Are prescriptive analytics bound to be an amazing tool sidelined by bad business habits?
Everyone's An Analyst
The Industrial Internet benefits not only industrial processes, but crucially, workers too. The flow of intuitive, actionable information into the hands of workers is geared toward enabling workers to do their jobs more easily and effectively. In doing so, the Industrial Internet is also giving the power of decision making to more workers than ever before. Workers are poised to be more knowledgeable about the technologies and projects they are working on, along with the way their work integrates with that of others.
If this kind of decision making, made by looking at real-time information, becomes routine in everyday tasks, then not only will more workers be actively making decisions, but each decision will have a higher degree of visibility than ever before. More visibility decreases the likelihood that people would make decisions in the face of contradictory information. In this sense, the Industrial Internet has the power to democratize decision making for the benefit the whole organization.
Prescriptive analytics are crucial to that process, since they provide the alternative choices, opportunity costs, and feasibility insights that good, structured decision making requires. But is there a need to go even farther than presenting comprehensive alternatives? Should software, factoring in prescriptive analytics, go ahead and actually make those decisions for us?
Digital Decision Making
At Melbourne's Alfred Hospital, a piece of software is actively saving lives. The Trauma Reception and Resuscitation Project is a decision support system designed to help emergency room staff treat trauma patients within the crucial first 30 minutes of their arrival. It factors in numerous variables -- such as blood pressure, chest tube insertions, shock, air entry, injury -- at highly specific levels and recommends actions or pages personnel to best ensure patient survival. It's been incredibly effective, reducing errors made my emergency staff by 21% over 33 months.
But it's one thing for software to aid in decision making, it's entirely another for it to make decisions directly. But that's exactly what's happening in safety domains like collision avoidance systems in cars, which use radar and laser sensor data to alert drivers of upcoming front-end collisions. If a driver takes no corrective action, the car automatically breaks to avoid the collision. Through collision avoidance, we can uncover a tacit belief about the relationship between prescriptive analytics and human decision makers, at least in the area of safety. If analytics detect a potential problem, their first and generally considered best course of action is always to alert a human. But in an emergency when time is critical, if humans don't respond, the software itself has the capability to interface with control systems and take corrective action.
These systems could eventually scale to become commonplace in other areas of transportation and industry, producing smarter autopilots that help not hinder in emergencies, or trains with automatic breaking systems that prevent derailments. Does this mean that people, as decision makers, will eventually become less skilled and less valuable? There are still several technical and procedural challenges in enabling prescriptive analytics software to actually take action. Prescriptive analytics, even when they factor in tremendous data of all sorts and source, still rely on people's willingness to embrace them and make changes for the better. The Industrial Internet, like many transformative technologies of the past, is still ultimately only as effective as we make it. Information, like the cryptic warning about the ides of March from Shakespeare's soothsayer, will never be 100% clear in depicting the world as it is or as it will be. But it's there, impelling us to take a look and consider. Whether we do, or whether we ignore it is ultimately up to us.