Month: June 2013

“Good enough” data analysis will lead to good data analysis

Someone on Quora asks, “What are the most promising disruptive innovations for the next decade (2011-2020)?”

The always incisive Venkatesh Rao offered a long list. At least one of his predictions brings to mind the decision support industry’s many do-it-yourself tools.

Deprofessionalization/democratization: if you thought blogging taking down professional writing, and amateur photography threatening professional photography were big, you ain’t seen nothing yet. Tons of problems that previously required professional expertise to solve can now be re-engineered so that amateurs can solve them. Movies and education are next.

Look around. If the tools don’t exist, but the high-margin professionals and eager amateurs do, try and invent them.  A big opportunity here is for amateur animated video shows. There are some primitive things that get animated characters speak out scripts, but we’ve barely scratched the surface.

Note the absence of data analysis. He could have listed it along with movies and education. But he didn’t, most likely because he knows too much about it to accept “good enough.”

People tend to go along with the good-enough except in their own field. There they see exactly why “good enough” just isn’t. Professional writers and photographers still scoff today at amateurs’ work even 20 years into that takedown.

When you need something done well, you hire a pro. But for many jobs, you can get by on “good enough” — until you’re skilled enough to know what you’re missing or curious enough to want something better.

That path played a big part in the California wine industry’s recent history. Back in the early ’80s, I talked to the finance officer of one of the many new craft brewers in California. He had seen how mass tastes work a decade before with the rise of a bad jug wine, Italian Swiss Colony. (“Made by that little old winemaker, me.”)

Sure, the stuff was awful by today’s standards. But the marketing got Americans drinking wine. Before long, many of them wanted to know how the good stuff tasted. So throughout the ’70s and even today, small wineries thrive thanks to that new thirst.

So it will be with data analysis, not to mention movies and education.

Two more execs on data analysts in the executive suite

An insurance executive I quoted a few weeks ago in this space spoke heresy: Not only do most data analysts he’s known lack the polish for presentations to execs, even worse was too narrow a point of view for leading an organization. It’s the Analytics Gap.

To that, the prolific tweeter and data scientist @data_nerd, Carla Gentry, reacted for many others in the decision support industry: “Huh?”

“I think a logical mind is perfect for business,” she tweeted to me when I inquired. “You must see the whole picture to run a successful business. Having an analytical mind helps.”

All true. Even the insurance executive and two others of that level — none of whom agreed to be named — would agree. The difference comes in the contrasting views of the “whole picture.”

Carla seems to assume that the whole picture can show itself in the data. The execs argue that it takes experience and judgement to see the real thing. “The conceit of BI,” emails a retired veteran of multiple technology startups and longtime acquaintance, “is that it has intrinsic value in its own right.”

Data analysts are narrowly focused, and rightly so. They’re looking for the certainty — or at least the illusion of certainty — that numbers provide. Execs — at least the good ones — know they are dealing with a messy and uncertain world. Human beings don’t behave as reliably as numbers. And executive decisions must deal with programs that stretch over many quarters, or even longer. In that time all manner of unplanned events may interfere and require solid executive leadership.

I still think that data analysts will someday find their profession to be a good route to the top. Sales and finance are good paths now, and each tends to breed limited candidates. But until data analysis evolves, it looks like the door to the executive suite may not open as readily as I’d thought it would.

Execs vs. data analysts gives us the Analytics Gap

One insurance company’s chief data scientist is “terrific at what he does,” says his boss, even “brilliant.” The company is lucky to have him. But he never sees the inside of the executive suite.

“He is a quirky, quirky guy,” says the boss, “and he is a super powerful dude in what he’s doing for us.” But upper executives would judge him harshly. “There would be a terrible outcome.”

Let’s call it the Analytics Gap. At least in this company and probably in many others, some of those who interpret the data don’t stick to rigid, unspoken rules of behavior. And those who with the greatest executive authority can’t quite appreciate what these people can tell them.

“These [analysts], including me,” says the analyst’s boss, “we’re different.” The brilliant data scientist — who as a hobby builds computers from video cards because “‘there’s a lot of power in video cards'” — makes a bad impression in person. He wanders off topic and, for example, has been known to digress into world politics and weird analogies.

“When I hear something come out his mouth that’s a little off color,” the boss says, “I chalk it up to his brilliance.” The company’s top executives, on the other hand, are “a harsh crowd.”

This crowd judges the off-color analyst harshly and fails, for example, to appreciate the real meaning of his risk models and other aspects of advanced analytics. They isolate themselves from the abstract evidence in data analysis that could correct or at least be reminded of their own biases.

The executives tell the analyst’s boss they want that. “I say that if you want that, you have to realize you’re dealing with people whose skill sets are completely different from anything you’ve dealt with. In a perfect world, my chief executive officer would say, ‘I understand how his analysis is applicable.’ No. Nothing.” They just don’t respect what he has to offer.

We’ve seen the work of harsh crowds before. Didn’t we just witness something like this on Wall Street? Those who knew the data saw the mortgage meltdown coming. But those who could have prevented it either dismissed the warnings or misunderstood them. What idiots, we say now. They should be jailed or made to work at Panda Express the rest of their working lives.

Potential consequences of the Analytics Gap are mostly minor by comparison but more widespread and chronic.

“I don’t think it’s a stretch to suggest that micro-versions of [the meltdown] could happen in product development, pricing, marketing, risk, and other domains” — unless data analysts and executive leadership come to understand each other. 

Why not train more quants? It won’t work, he says. Generally, those interested in data analysis tend to be too technically minded for management.