FOR THOSE interested in doing a better job of managing people – supporting them, inspiring them to greatness – there is plenty of advice out there. The problem is that it is not clear which advice to follow. Management philosophies are like apple pies; they all sound good.

The issue is a vital one, and not just for companies that want to improve their bottom line. A large part of what people do is done in groups. If there are indeed bedrock principles of management, then their wide application could do a lot of good for a lot of people.

Now MIT’s Sandy Pentland has an answer to this apple pie conundrum, and the answer is “data.” In this month’s Harvard Business Review, he describes years of work outfitting people with electronic badges to track their office interactions. The results suggest that successful teams have a clear signature – a DNA of team brilliance – and that lower performing teams can be readily improved with simple changes.

“This is the version of apple pie that actually seems to work,” says Pentland, who directs MIT’s Human Dynamics Laboratory.

Pentland is certainly not the first person to try to turn the study of management into more of a science. That’s an effort that goes back many decades. But he’s created a rich, new source of data. And in the 21 organizations he’s studied over the last seven years, one central idea has emerged: What really matters is the pattern of communication. It is so powerful, Pentland argues, that the communication pattern – who is talking to who, when, how – is a more significant factor in excellence than any of the more obvious possibilities, such as the intelligence of team members, their personalities, and skills.

In a healthy team, all the individual members talk to each other, not just the boss. Everyone listens as much as they talk. There is frequent communication, but it tends to be fairly fast. And people regularly make forays outside the team, learning new things, and then share when they come back.

Pentland and his colleagues generate graphics with their data and the results are striking. They draw a circle of everyone in a team, with lines between them showing the intensity of communication. In dysfunctional groups, you see a few heavy lines – the boss issuing orders to his lieutenants, say – and lots of light lines. People don’t talk among themselves. But in bastions of creativity and productivity, the boss-man does not dominate discussions and everyone is talking to everyone else.

This certainly rings true for me. When I think of the time in my career that I was in the most creative, effective working environment, it was a place where there were plenty of impromptu, free-flowing conversations. They tended to be brief and didn’t interfere with focused work. There were virtually no formal meetings. We all had frequent, important contact with people outside our department.

Pentland tells the story of a large bank he worked with, which was trying to improve its call centers. He gave them a tip which turned out to be worth millions: change the coffee break schedule. The bank was doing the logical thing, which was to have people on the same team stagger their coffee breaks. But when they had the whole team break together, they found that employees’ performance and job satisfaction jumped. People were talking to each other more. The bank is going to make the change in all its call centers and predicts an annual productivity boost of $15 million, according to the Harvard Business Review article.

Pentland’s data offer clear lessons for the bosses of the world. The best ones, Pentland says, are “charismatic connectors.” They talk to everyone, not just the bigwigs. They listen as much as they talk. They put people in touch and understand that the good ideas are not going to just pop into their heads, but are a product of people sharing what they know.

Pentland offers a great way of thinking about this: a beehive. When a bee gets back to the hive after a successful foray, she does not write a memo to the queen. She does a little dance, sharing the news with her coworkers.

“You need to think of the group as an information harvesting and winnowing machine,” says Pentland. “You want to get a bunch of ideas, bounce them off of everyone and see what sticks.”