By Deep Work Club · Last updated: August 2026 · 8 minute read
The intuitive answer is that the best team is the one with the best people in it. Google spent several years testing that assumption against its own teams and found something else. This article covers what actually separates high performing teams, why psychological safety is the mechanism rather than the mood, and the four leadership behaviours that build it. It draws on episode 3 of the Deep Talk podcast (Spotify) with Lea Mersch, psychologist and psychotherapist, and Till Westermann.
Key takeaways
Talent, experience and IQ predicted team performance surprisingly weakly. How a team works together explains more than who is on it.
Google's Project Aristotle studied roughly 180 internal teams and identified five factors: psychological safety, dependability, structure and clarity, meaning, and impact.
Psychological safety is the strongest of the five. Amy Edmondson defines it as a shared belief that the team is safe for interpersonal risk taking.
It is not harmony. Teams high in psychological safety often have more conflict, but about the work rather than the people.
The behaviour of the highest-status person in the room sets the standard. That makes leadership the most direct lever available.
What is a high performing team?
A high performing team is one that consistently exceeds its goals because its working process holds, not because its members are individually exceptional. Decades of research point at process rather than composition. Steve Kozlowski and Daniel Ilgen reviewed roughly 50 years of team research in 2006 and concluded that team effectiveness is driven substantially by coordination, communication and mutual support.
That is an uncomfortable finding, because it undercuts a common hiring logic. Assembling a team of A-players does not reliably produce high performance. Sometimes it produces the opposite.
Why talent alone does not build a team
Google launched Project Aristotle in 2012 expecting talent density and experience to explain performance. It examined roughly 180 internal teams, with over 200 interviews and around 250 variables. The expectation did not survive contact with the data: talent, experience and IQ correlated surprisingly weakly with team performance.
The central finding, in Google's own words:
"Who is on a team matters less than how the team members interact, structure their work, and view their contributions."
— Google Re:Work, 2016
One caveat worth keeping: these are findings from inside Google. Other organisations may differ. The broader research tradition points the same way, which is why the result is credible rather than anecdotal.
In the episode, Till describes what the gap feels like from inside a team where the credentials were excellent and something was still missing:
"I didn't really know what would happen if I made a mistake. How bad is this, actually? Should I just cover it up and keep acting like I have everything under control?"
— Till Westermann, Deep Talk episode 3
Lea's reading: when everyone claims similar status and pressure arrives from above, you do not get collaboration. You get elbows. Output can hold for a while. It does not hold for long.
The five factors
Factor | The question the team is really asking |
|---|---|
Psychological safety | Can I admit mistakes, ask questions and disagree without fearing consequences? |
Dependability | Does everyone deliver what they committed to, at the agreed quality and time? |
Structure and clarity | Are roles, goals and decision paths clearly defined and understood? |
Meaning | Does each person find a personal reason to consider this work significant? |
Impact | Does each person believe their work actually changes something? |
These are not independent switches. They form a system: structure without safety produces compliance, safety without structure produces drift.
Structure and clarity is the factor most often underestimated. In Gallup's Q2 2025 survey of 17,660 US employees, only 47 percent strongly agreed with the statement "I know what is expected of me at work." Fewer than half, on a question that sounds almost too basic to ask.
Meaning has a solid theoretical foundation in the self-determination theory of Edward Deci and Richard Ryan. Their point matters for leaders: the goal is not that someone loves the work. It is that each person finds their own reason to treat it as significant. That can be actively supported.
Psychological safety: the strongest and most misread factor
Psychological safety is the shared belief among members of a team that the team is safe for interpersonal risk taking. Amy Edmondson, professor at Harvard Business School, defined it that way in 1999. The word "shared" carries the meaning: it is a property of the team, not a trait of individuals.
The concept predates the current conversation. Edgar Schein and Warren Bennis described it in 1965. Edmondson has studied it empirically as a team construct since the late 1990s.
Her foundational study covered 51 teams in a manufacturing company, and the pathway she found is the real point: psychological safety predicts learning behaviour, and learning behaviour predicts team performance. Safety is not a soft addition to performance. It is the mechanism that produces it.
Her hospital research makes this concrete. Edmondson examined medication errors in clinical teams, expecting better teams to report fewer. The opposite showed up. As Lea puts it in the episode:
"They didn't make the most mistakes. They communicated those mistakes. The other teams lifted the carpet and swept the errors underneath. Which means they never had a chance to get better."
— Lea Mersch, Deep Talk episode 3
In 2001, working with Richard Bohmer and Gary Pisano, Edmondson found the same pattern across 16 hospitals adopting minimally invasive cardiac surgery. Comparable reputations, dramatically different outcomes. The successful teams were the ones where people could speak openly in the operating room. Even in steep hierarchies where lives are at stake, communication decides.
What psychological safety is not
This is where most of the misreading happens. Psychological safety is:
not harmony. It does not require that everyone likes each other.
not the absence of conflict. Edmondson notes that teams high in psychological safety often have more conflict, but task-focused rather than interpersonal.
not a comfort culture. High standards and safety are not opposites. They depend on each other.
When people fear social consequences, that is not a character flaw. There is neuroscientific evidence that stress responses impair prefrontal cortex function, the region responsible for complex thinking. Amy Arnsten demonstrated this in laboratory work. The direct transfer to team dynamics is plausible but not settled.
Four leadership behaviours that build psychological safety
From more than 20 years of field research, Edmondson identifies four:
Frame the work as a learning problem, not an execution problem. "I expect you to deliver this" punishes uncertainty. "This problem is complex and I need what you know" makes it normal.
Show your own fallibility. Name your mistakes and knowledge gaps. Not as weakness, as modelling. Brené Brown's qualitative work suggests leaders who do this are perceived as more trustworthy rather than weaker, though the evidence base is qualitative and quantitative replications vary.
Show real curiosity. Ask questions that do not already contain their answer. Not "we all see it that way, right?" but "what am I missing?"
Reward speaking up quickly and visibly. When someone raises a problem or admits an error, acknowledge it immediately and in front of the team. Not later, not privately.
The fourth is the one leaders find hardest, and the reason is often biographical. Till describes it directly:
"The leaders we had back then were never praised either. And the interesting part was that I then didn't do it myself. I had never experienced it."
— Till Westermann, Deep Talk episode 3
Three things you can do this week
1. The one-minute disclosure
In your next team meeting, take one minute to name a mistake you made in the past week and what you took from it. Edmondson describes this as one of the most effective signals a leader can send, precisely because it models the behaviour.
2. A single diagnostic question
Ask your team, anonymously if needed: "On a scale of 1 to 5, how safe do I feel voicing a dissenting opinion in this team?" You do not need an elaborate survey. One question is enough to start, and the trend across several months tells you more than the first score.
3. The pause before the no
When someone offers an idea and it meets resistance, stop and ask: "What could be interesting about this idea?" Not because every idea is good, but because the person learns they are safe even when they are wrong.
One practical note: rituals like this take time to work. Till introduced a standing Friday feedback round in threes, after five minutes of silence. It took several weeks before anything substantial came out of it. That is normal, not a sign the format is failing.
Frequently asked questions
What makes a high performing team?
Not talent density but the way people work together. Google's Project Aristotle identified five factors: psychological safety, dependability, structure and clarity, meaning, and impact. Psychological safety is the strongest of them.
What is psychological safety at work?
The shared belief within a team that interpersonal risks are safe: asking questions, admitting mistakes, disagreeing. The term comes from Amy Edmondson, who defined it as a team construct in 1999.
Does psychological safety mean avoiding conflict?
No, close to the opposite. Teams high in psychological safety often have more conflict. The difference is that it concerns the work rather than the people.
How do you measure psychological safety in a team?
A single question on a 1 to 5 scale about how safe it feels to voice disagreement is enough to start. The direction of travel across several measurement points is more informative than any single score.
How long does it take to build psychological safety?
Longer than a workshop. Rituals like recurring feedback rounds often need several weeks before people speak openly. Repetition matters more than intensity.
What can leaders do to build psychological safety?
Frame work as a learning problem, show your own fallibility, ask genuinely open questions, and acknowledge people who speak up immediately and visibly. Edmondson derived these four behaviours from more than 20 years of field research.
In summary
High performing teams are not built from talent density. They are built from how the work gets done. Psychological safety is the mechanism through which learning, and therefore performance, becomes possible.
And the strongest lever sits closer than most leaders assume. As Lea puts it in the episode: culture follows the behaviour of the highest-status person in the room. What you treat as normal becomes normal.
Which means you do not have to fix your team. You can start with yourself.
A question for this week: in which moment over the past seven days could you have created safety, and did not?
At Deep Work Club we work with leadership teams on exactly this, without lectures and without scripts. See how we work with teams, or read on about why leaders avoid difficult conversations and radical candor as a feedback model.
The full episode: Deep Talk episode 3, High Performing Teams, with Lea Mersch and Till Westermann. This is a German-language podcast. Listen on Spotify.
Sources
Edmondson, A. C. (1999). Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly, 44(2), 350–383.
Edmondson, A. C., Bohmer, R. M., & Pisano, G. P. (2001). Disrupted Routines: Team Learning and New Technology Implementation in Hospitals. Administrative Science Quarterly, 46(4), 685–716.
Edmondson, A. C. (2018). The Fearless Organization. Wiley.
Google Re:Work (2016). Understand Team Effectiveness. rework.withgoogle.com
Kozlowski, S. W. J., & Ilgen, D. R. (2006). Enhancing the Effectiveness of Work Groups and Teams. Psychological Science in the Public Interest, 7(3), 77–124.
Schein, E. H., & Bennis, W. G. (1965). Personal and Organizational Change Through Group Methods. Wiley.
Deci, E. L., & Ryan, R. M. (1985/2000). Self-Determination Theory.
Gallup (2025). Q2 2025 U.S. Employee Survey, n = 17,660.
Arnsten, A. F. T. (1998). The Biology of Feeling Frazzled. Science, 280(5370), 1711–1712.
Brown, B. (2010). The Gifts of Imperfection. Hazelden.
Deep Talk Podcast, episode 3: High Performing Teams (27 April 2026), Deep Work Club.