n his recent book, The Signal and the Noise: Why So Many Predictions Fail – but Some Don’t , Nate Silver offers this warning: “Data-driven predictions can succeed- and they can fail. It is when we deny our role in the process that the odds of failure rise.” I was prompted to locate this quote in the book while watching a video of IBM CEO Ginni Rometty’s talk at the Foreign Relations Council given a couple of weeks ago. Rometty ‘s theme was that Big Data and predictive analytics were going to change how companies operate. In particular she thinks companies will change in three important ways: how they make decisions, how they create value and how they deliver value. Those are significant changes. One of her examples was risk management which an IBM study showed was most often based on “senior management intuition and experience” as opposed to hard data. But therein lies the rub. If Silver is correct, it is not simply a matter of collecting the data and running it through some magic algorithm and coming up with the right answer. Well, I guess that’s not really news after the spectacular failure of risk models that preceded the financial crisis. But this is supposed to be a new world, a world where predictions aren’t predicated on some model cooked up in room full of brilliant mathematicians, but one where the predictions are based on Big Data. But as Silver points out: “The numbers have no way of speaking for themselves. We speak for them. We imbue them with meaning.”
The number of opportunities for bringing Big Data and predictive analytics to bear on problems and challenges in the Financial Services industry will no doubt continue to grow but as McKinsey and others have noted, there will be a shortage of talent available for firms to take advantage of these capabilities. Even if the expertise can be found, anyone who has experience with the traditional methods for gathering business requirements for application development in financial services will look askance at the idea of turning over critical decision-making to the resulting software. Methods for assimilating information – for sorting between signal and noise – at the enterprise level will need to improve. More important perhaps than the talent will be the attitudes. As Rometty made clear in her talk, mindset and culture, not technology will drive successful adoption. Which all comes back to Nate Silver’s caution: Implementation of Big Data and predictive analytics initiatives will be just as dependent, if not more so, on the knowledge and experience of the people organizations rely on to get most things done. Except now the complexity has grown, the need to make the data transparent has increased while the amount of data has multiplied exponentially and it all has to be done faster to be competitive. The challenge is in synthesizing coherent output from this environment to create value. If Rometty is right, and the way your firm makes decisions, creates and delivers value are all based on the success of your predictions, the stakes couldn’t be higher.