Resting on the innovation side of analytics, it’d be easy to lose yourself – and your business audience – in the data promises of predictive models.

The picture plays out something like this: You jettison the latest and greatest analytic software and a modeler somewhere between the middle cubicle and the server room, leaving behind the keys to the data and a few departmental demands for ROI. Then, during an 11th hour C-suite meeting where the business direction seem doomed for good, a frazzled modeler in a lab coat busts into the room with charts, a jumble of papers and a declarative finding that saves the quarter.

Hooray for the data model! And never mind the curious reason the modeler is wearing a lab coat (or that the CEO didn’t call security when some half-crazed individual in a lab coat erupted into their important business meeting). The real fiction behind this imaginary scenario and many others being promised in snazzy vendor offerings and scattershot mainstream media portrayals is the absence of human curiosity. More and more, data has sliced away the poor business choices at the end of the decision spectrum. But increasingly in predictive analytics, we see the importance of “the hunch” at the beginning of the process.

That thought was swirling in my head coming back from last week’s TDWI event, as, on the plane home from Las Vegas, I cracked open Eric Siegel’s new hardcover book  entitled “Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie or Die.” Siegel, who carries a background as a self-described “recovering academic” in the field of machine learning and computer science, does an nimble job of presenting PA in the “pop sci” way that has done wonders for Brian Greene and the “Freakonomics” guys in their respective fields. That doesn’t mean it’s fluff or a diversion from Siegel’s data roots. For every David Bowie lyric quoted, Siegel dives into quite a few more CART decision trees. For each reference to “Moneyball,” there are a dozen more outlines of how a range of companies can really put predictive models into play.

Early on in Siegel’s book, the word “hunch” and similar inquisitive business terms arise in case studies, both good and otherwise. In one colorful instance, he likens a marketing data exercise to “one million monkeys chucking darts across a chasm in the general direction of a dartboard.” Semantics aside, Siegel told me in a call to discuss the book that good business guesstimates, hunches and assumptions on the front end open the opportunity for analytics to move from forecasting and into influence.

“With something like data preparation: it’s the most boring topic, but it’s also the most central to the art of applying predictive modeling,” he says. “There’s a certain amount of this stuff that can be discovered through mass exploration, but there is always potential benefit from sitting back and gathering hunches. Really, there are always limits on what can be discovered automatically. The artistic side to this comes from ... the balance between setting it in the right direction and still giving it the freedom to explore.”

Today’s CIOs and data churners won’t be following through on these hunches alone. There’s a wave of business-minded data folks being prepped at the university level, according to the professors and deans in charge. And that background learning and experience is coming together most often on predictive capabilities. In its recently released “state of BI education” survey, the BI Congress III reported that the top course of study across university disciplines was predictive analytics/data mining. So, the hope is more future business leaders with a dose of advanced analytics along their coursework, rubbing elbows and rubbing off on the hardcore data churning students. I know that’s one of the overarching goals expressed by BI Congress chairwoman Barbara Wixom, an associate professor at the University of Virginia and soon-to-be MIT Sloan faculty member who has continually been an advocate of collaboration between university of business and IT.

We get so caught up in the power and promise of millions of rows of data that the information itself can appear absolute in its results. But the best examples of data implementations at the enterprise level continue to be those that include input – and support – from the business side of the house. At the start, it’s in your business interest to pass your hunches along to the modelers and data preppers. Then you’ll be ready for the decisive data you get in the end. Or, better yet, as Siegel told me, the hunches better position you to “influence the business future.” Unless you were content with waiting for a frazzled weirdo in a lab coat to burst into your next meeting with a bunch of data, which probably isn’t the best path going forward. Call it a hunch.

***

As this is my first official blog for the site, I thought it’d be a good time to introduce myself to those whom I haven’t already met and talked with. Hi! I’m the Senior Editor for Information-Management.com, handling daily trend report stories, market guru interviews and Twitter attempts at condensed cleverness. Now I’ll be adding a blog or two a month to our editorial rotation. I’ve also settled in to the “Andy Richter” role opposite Eric Kavanagh’s “Conan O’Brien” on DM Radio (where we’ve got Eric Siegel as a guest on Thursday’s episode). For guidance in much of these endeavors, I’m indebted Jim Ericson, who was a great mentor to me and everyone at our publication for the last nine years. He’s since stepped away to measure his next steps (and maybe catch some spring training foibles from his beloved Cubs) but  we won’t let him stray too far; he’s helping us with a mobility event in June in San Francisco, to be co-chaired with long-time info-mgmt.com blogger Mark A. Smith of Ventana Research.