Google’s top executives long believed that building the best teams meant combining the best people. They had other rules of thumb too. “It’s better to put introverts together,” was one, according to Abeer Dubey, a manager in the company’s People Analytics division. But, Dubey said, “it turned out no one had really studied which of those were true.”
So in 2012 Google set out to study it. The project was code-named Aristotle, after the line about the whole being greater than the sum of its parts. Researchers picked 180 teams, 115 in engineering and 65 in sales, some strong and some weak.
Even deciding what “effective” meant was hard. Leaders first wanted hard numbers, then saw the problem: more lines of code is not always better, and more bugs fixed means more bugs were made. The researchers settled on ratings from executives, team leads and team members, plus sales against quota. Then they ran more than 35 statistical models on hundreds of variables, from personality traits and skills to tenure, seniority and whether teammates sat in the same office.
The roster barely showed up. “We had lots of data, but there was nothing showing that a mix of specific personality types or skills or backgrounds made any difference,” Dubey said. “The ‘who’ part of the equation didn’t seem to matter.” Two teams could have nearly identical makeups, even overlapping members, and radically different results. In Google’s analysis, team size, extroversion and even the individual performance of team members were not significantly connected to how effective a team was.
What did connect was how people worked together. The researchers named five dynamics and put one first: psychological safety, the confidence that no one on the team will embarrass or punish you for admitting a mistake, asking a question or offering a new idea. The Harvard professor Amy Edmondson had described the idea in a 1999 study. Dependability, clear structure, meaning and impact came after.
In late 2014 the researchers began presenting their findings inside Google. After one talk, a manager approached them. He had once been on a SWAT team; at Google he led engineers who respond when the company’s sites or servers go down. His previous team hadn’t gelled well. “There was one senior engineer who would just talk and talk, and everyone was scared to disagree with him,” he said. Outside meetings, everyone liked the man.
He gave his new team the project’s survey. He had thought of them as a strong unit, but they rated the clarity of the team’s role, and whether their work had impact, middling to poor. He took them off site to talk it over, and asked everyone to start by sharing something personal. He went first. He had Stage 4 cancer, he told them. They had worked with him for ten months. No one had known.
Another teammate then spoke about her own health. Someone else talked about a hard breakup. After that, the small frictions that had been bothering people came out more easily. The team agreed on new norms: the manager would show them how their work fit Google’s larger mission, and they would all try harder to notice when someone felt left out or down.
Nothing in the survey had told him to talk about his illness. “Googlers love data,” he told the Times. Putting empathy into charts, he said, made it easier to raise.
“It’s easier to talk about our feelings when we can point to a number.”