The predictability of big data analytics had a miniboom starting a few years back, almost at the same time that erstwhile baseball statistician Nate Silver found sudden success predicting elections. Then, unpredictability came knocking.
Like many seers before him, time has taken Silver down a few pegs, principally because he underestimated the chances of upstart candidate Donald Trump. Silver has been far from alone in missing that one. Meanwhile, a carnival of surprises has marked 2016.
The political rise of a rank outsider, the exit of Britain from the European Union, the late nose dive of the Golden State Warriors -- these and other elements have combined to give 2016 the aura of "Year of Unpredictability."
That could take at least a little air out of the big data analytics balloon. Why? If successful strings of prediction like Silver's gave some CEOs a warmer feeling about big data and analytics, a seeming chain of flops may give them pause. Who is to say that such bumps on the road to data-driven predictions won't affect some CEO's trust in data and analytics?
Data distrust in the C-suite?
Some data suggests trust in data and analytics is narrow. Although data managers have worked over the years to improve the quality of corporate data, leaders in C-suites continue to have doubts. According to a KPMG survey of 400 U.S. CEOs, 77% have some level of distrust toward the quality of the data on which they base their decisions. Hand in hand with that, only 33% of the CEOs said they had a high trust in the accuracy of their data and analytics efforts.
"You can come in and do the best data science in the world, but if the leader doesn't trust the results, it isn't useful," said Wilds Ross, principal of data and analytics at KPMG.
Ross described the gap as a ''trust deficit.'' There may be good data quality and data analytics work going on, but often it is not being used by the business, he said.
Predictive analytics needs to become friendlier to the C-level executive, according to Ross.
The gap between the concerns of the data technologists and the concerns of the CEO can be a wide one. The problem is often in communications, he said.
"Sometimes data is presented in 'black box' structures," he said. "The message from the analytics expert is 'just trust me, this is fine.' On the other side, senior people are wondering 'how do I know this really is as they present it?' or 'if the coding was free of errors?'''
Ross said to bridge the gap, one, the data itself has to be well managed; two, the experts who deliver the analysis need to be able to walk executives through the entire process that led to the analysis. As well, it is not helpful for the analytics messenger to not act as though they have ''smartest person in the room'' syndrome.
Analytics have to be done in the context of how decisions in the organization are made. "That is a group effort," Ross said. "It's not about being the smartest person in the room; it's about a group effort."
Start with decisions in mind
Another take on the predictability as it relates to analytics and decision making comes by way of James Taylor, CEO and principal consultant at Decision Management Solutions in Palo Alto, Calif.
For many years, his practice has focused on decision making because he has seen so many instances where technologists -- and the business players -- focus on the newest technology while ignoring the way a company is making its business decisions. He sees this repeated new big data and predictive analytics technologies today.
Taylor said he regularly sees analytics presented in ways that don't help decision makers. His advice is ''start with the decision in mind'' -- a variation on Steven Covey's advice to "begin with the end in mind." Without that objective, the data scientist is unlikely to really ask the right questions of the data.
"Too often, the data is analyzed and presented by people who don't understand what the business is," Taylor said. "Members of the analytics team too seldom talk to the decision makers about how they make decisions -- that is a disconnect."
To be successful with big data and analytics, Taylor said, companies have to understand the process of decision making, and he employs decision modeling to address this issue. It is not just CEOs that benefit from connecting analytics to decision modeling either, he said, pointing to call centers as a place where many decisions are made that affect a business.
Maybe it was predictable -- that trust in predictability will ebb and flow like tides. But it is everyone's job to keep an even keel. If people on the data and analytics side can become better data curators and more compelling data storytellers, they are doing their part. It is a step toward building trust with their co-workers on the line and in the corner office.
Review David Loshin's take on decision-making models
Learn the latest on big data analytics
Look at the environment of decisions