From tonight's New York Times feed on the Google hiring process:
As a result, [Laszlo] Bock, who joined Google from General Electric last spring, has been trying to make the company’s rigorous screening process more efficient. Until now, head hunters said, Google largely turned up its nose at engineers who had less than a 3.7 grade-point average. (Those who wanted to sell ads could get by with a 3.0 average, head hunters said.) And it often would take two months to consider candidates, submitting them to more than half a dozen interviews.
Unfortunately, most of the academic research suggests that the factors Google has put the most weight on — grades and interviews — are not an especially reliable way of hiring good people.
“Interviews are a terrible predictor of performance,” Mr. Bock said.
Replacing GPAs and interviews is a new survey instrument designed by Todd Carlisle. The initial prototypes of the survey checked to see if you owned a dog (not a good indicator of performance) or if you started a club in high school.
But Dr. Carlisle was able to create several surveys that he believed would help find candidates in several areas — engineering, sales, finance, and human resources. Currently about 15 percent of applicants take the survey; it will be used for all applicants starting this month.
This post from Todd Raphael on "Horsing Around at Google" refers to Google's "staffing analytics" department:
He'll be looking at what traits successful current employees have, and what lessons can be learned from those people (Did they attend Purdue and work at Microsoft for at least five years? Did they major in music?) that will help Google select new ones.
"We have more jobs, more locations," Carlisle says. "How do you know if they fit?"