Big Data: Can you tell the signal from the noise?

Last month, Google’s SVP of people operations Laszlo Bock gave a much-discussed interview to the New York Times on the role of Big Data in the recruitment process. On the one hand, he says, hard data can help you separate your feelings from the facts. While almost everyone thinks they’re leadership material, for instance, few people actually are.

That’s where Big Data can come in handy:

If you go back to somebody and say, “Look, you’re an eighth-percentile people manager at Google. This is what people say.” They might say, “Well, you know, I’m actually better than that.” And then I’ll say, “That’s how you feel. But these are the facts that people are reporting about how they experience you.” You don’t actually have to do that much more. Because for most people, just knowing that information causes them to change their conduct. One of the applications of Big Data is giving people the facts, and getting them to understand that their own decision-making is not perfect. And that in itself causes them to change their behavior.

On the other hand, lots of seemingly-relevant data can actually turn out to be useless noise. Unless you graduated in the last two years, for instance, your college GPA—or even if you went to college, period—has no correlation to the quality of your work.  But what recruiter wouldn’t be impressed by a 4.0 GPA (or deterred by a 2.3)?

Similarly, those annoying “how many piano tuners are there in Chicago”-type brainteasers turn out to be irrelevant, too. So are your test scores. And even the recruiter you interview with (turns out none of Google’s interviewers are any better than the others at picking the best candidates). With so much irrelevant data being collected during the interview process, how can you possibly zero in on what matters?

The key is to approach the process like a scientist. Instead of relying on your gut, collect as much information as possible, and then—this is the key part—decide what actually matters. If you find that GPAs don’t correlate to performance, then stop collecting them. But don’t make the decision to stop until after you’ve examined the data.

By the same token, you may find that data that seemed irrelevant is actually a great predictor of future success. Maybe 80 percent of your best engineers are model train enthusiasts, and you should start probing candidates about their hobbies. Or maybe you’ll discover that you’ve never made a bad hire from a particular training program. Whatever the result, the key is to analyze the data without prejudice. You may have relied on certain criteria for decades, but if the data shows it isn’t helping you today, have the courage to throw it out.

Looking for more recruitment best practices? Check out our latest e-book, The ultimate guide to employee referrals: How to engage employees, save millions, and drive Employee Success™.