For psychological safety to emerge among a group, teammates don’t have to be friends. They do, however, need to be socially sensitive and ensure everyone feels heard. “The best tactic for establishing psychological safety is demonstration by a team leader,” as Amy Edmondson, who is now a professor at Harvard Business School, told me. “It seems like fairly minor stuff, but when the leader goes out of their way to make someone feel listened to, or starts a meeting by saying ‘I might miss something, so I need all of you to watch for my mistakes,’ or says ‘Jim, you haven’t spoken in a while, what do you think?,’ that makes a huge difference.”
In Edmondson’s hospital studies, the teams with the highest levels of psychological safety were also the ones with leaders most likely to model listening and social sensitivity. They invited people to speak up. They talked about their own emotions. They didn’t interrupt other people. When someone was concerned or upset, they showed the group that it was okay to intervene. They tried to anticipate how people would react and then worked to accommodate those reactions. This is how teams encourage people to disagree while still being honest with one another and occasionally clashing. This is how psychological safety emerges: by giving everyone an equal voice and encouraging social sensitivity among teammates.
Michaels himself says the job of modeling norms is his most important duty. “Everyone who comes through this show is different, and I have to show each of them that I’m treating them different, and show everyone else I’m treating them different, if we want to draw the unique brilliance out of everyone,” Michaels told me.
“SNL only works when we have different writing and performing styles all bumping into and meshing with each other,” he said. “That’s my job: To protect people’s distinct voices, but also to get them to work together. I want to preserve whatever made each person special before they came to the show, but also help everyone be sensitive enough to make the rough edges fit. That’s the only way we can do a new show every week without everyone wanting to kill each other as soon as we’re done.”
IV.
By the summer of 2015, the Google researchers working on Project Aristotle had been collecting surveys, conducting interviews, running regressions, and analyzing statistics for two years. They had scrutinized tens of thousands of pieces of data and had written dozens of software programs to analyze trends. Finally, they were ready to reveal their conclusions to the company’s employees.
They scheduled a meeting at the headquarters in Mountain View. Thousands of employees showed up, and many more watched via video stream. Laszlo Bock, the head of the People Operations department at Google, walked onto the stage and thanked everyone for coming. “The biggest thing you should take away from this work is that how teams work matters, in a lot of ways, more than who is on them,” he said.
He had spoken to me before he went onstage. “There’s a myth we all carry inside our head,” Bock said. “We think we need superstars. But that’s not what our research found. You can take a team of average performers, and if you teach them to interact the right way, they’ll do things no superstar could ever accomplish. And there’s other myths, like sales teams should be run differently than engineering teams, or the best teams need to achieve consensus around everything, or high-performing teams need a high volume of work to stay engaged, or teams need to be physically located together.
“But now we can say those aren’t right. The data shows there’s a universality to how good teams succeed. It’s important that everyone on a team feels like they have a voice, but whether they actually get to vote on things or make decisions turns out not to matter much. Neither does the volume of work or physical co-location. What matters is having a voice and social sensitivity.”
Onstage, Bock brought up a series of slides. “What matters are five key norms,” he told the audience.
Teams need to believe that their work is important.
Teams need to feel their work is personally meaningful.
Teams need clear goals and defined roles.
Team members need to know they can depend on one another.
But, most important, teams need psychological safety.
To create psychological safety, Bock said, team leaders needed to model the right behaviors. There were Google-designed checklists they could use: Leaders should not interrupt teammates during conversations, because that will establish an interrupting norm. They should demonstrate they are listening by summarizing what people say after they said it. They should admit what they don’t know. They shouldn’t end a meeting until all team members have spoken at least once. They should encourage people who are upset to express their frustrations, and encourage teammates to respond in nonjudgmental ways. They should call out intergroup conflicts and resolve them through open discussion.
There were dozens of tactics on the checklist. All of them, however, came back to two general principles: Teams succeed when everyone feels like they can speak up and when members show they are sensitive to how one another feels.
“There are lots of small things a leader can do,” Abeer Dubey told me. “In meetings, does the leader cut people off by saying ‘Let me ask a question there,’ or does she wait until someone is done speaking? How does the leader act when someone’s upset? These things are so subtle, but they can have a huge impact. Every team is different, and it’s not uncommon in a company like Google for engineers or salespeople to be taught to fight for what they believe in. But you need the right norms to make arguments productive rather than destructive. Otherwise, a team never becomes stronger.”
For three months, Project Aristotle traveled from division to division explaining their findings and coaching team leaders. Google’s top executives released tools that any team could use to evaluate if members felt psychologically safe and worksheets to help leaders and teammates improve their scores.
“I come from a quantitative background. If I’m going to believe something, you need to give me data to back it up,” said Sagnik Nandy, who as chief of Google Analytics Engineering heads one of the company’s biggest teams. “So seeing this data has been a game changer for me. Engineers love debugging software because we know we can get 10 percent more efficiency by just making a few tweaks. But we never focus on debugging human interactions. We put great people together and hope it will work, and sometimes it does and sometimes it doesn’t, and most of the time we don’t know why. Aristotle let us debug our people. It’s totally changed how I run meetings. I’m so much more conscious of how I model listening now, or whether I interrupt, or how I encourage everyone to speak.”
The project has had an impact on the Aristotle team, as well. “A couple of months ago, we were in a meeting where I made a mistake,” Julia Rozovsky told me. “Not a huge mistake, but an embarrassing one, and afterward, I sent out a note explaining what had gone wrong, why it had happened, and what we were doing to resolve it. Right afterward, I got an email back from a team member that just said, ‘Ouch.’
“It was like a punch to the gut. I was already upset about making this mistake, and this note totally played on my insecurities. But because of all the work we’ve done, I pinged the person back and said, ‘Nothing like a good Ouch to destroy psychological safety in the morning!’ And he wrote back and said, ‘I’m just testing your resilience.’ That could have been the wrong thing to say to someone else, but he knew it was exactly what I needed to hear. With one thirty-second interaction, we diffused the tension.