Google's infamous brain-teaser interview questions don't predict performance
They're supposed to test, in theory, how you tackle a big problem analytically. In the haircut example, you would likely start with the population count and multiply by how often you think people get a haircut every year.
However, those questions that Google was once known for asking don't actually predict how well an employee will do at Google.
"Everyone likes to ask case questions and brain-teasers. It turns out our data shows that doesn't actually predict performance. There's no correlation with your ability to do that," said Laszlo Bock, Google's SVP of People Operations, in an interview with Kleiner Perkins Caufield & Byer's partner Beth Seidenberg.
"Part of the reason is that those are tests of a finite skill, rather than flexible intelligence which is what you actually want to hire for," Bock continued.
Google has moved away from the famous brain-teaser questions, Bock has previously noted.
Instead, better questions for Google are structured interview questions that "are not rocket science." To screen for problem-solving, a better question is "Give me an example of a hard problem you solved" so the interviewer can then drill down for specific examples.
"Those kinds of questions are actually predictive of how someone will perform versus what we typically do," Bock said.