Presentation: What Google Learned about Creating Effective Teams



1:40pm - 2:30pm


Key Takeaways

  • Understand some of the practical research Google has done around building effective teams.
  • Learn small changes and techniques that can make your teams better and more psychologically safe.
  • Hear personal narratives from stories at Google and other tech companies about building effective teams.


Over the years, researchers have conducted numerous studies on team dynamics. At Google, our People Analytics sought to identify those attributes that set effective teams apart from ineffective ones. The driver was that while so much of work at Google is done on teams, Googlers are rated individually; there was a gap. We felt that if we could distinguish the good from the bad, we could embark on a program to improve.

We made some surprising discoveries; many of our "common sense" initial assumptions were unfounded. We also learned that team dynamics can be surprisingly elusive to implement, even when one knows what the success criteria are. This talk addresses not only the research, but the insights of a manager who worked with his own team and others, to instill the findings and principles in the real world, along with key insights from the Pilot program at Google.


QCon: Can you tell me a bit about your role today?

Matt: My first job out of school was a police officer. I did that for eight years, and I really loved it. I thought that was going to be my entire career. However, an injury prevented that. At 30, I had to embark on a different path. To make a long story short, I ended up in technology in 2001, and I have been in some type of IT leadership role since then. My experience in tech has mostly been running large websites like, leading small teams in infrastructure, mail processing, mail support, back-ends, and now SRE at Google. 

We describe ourselves as software engineers that happen to do operational facing work. 50% or more of my team’s time is spent coding. It is just that we don’t write features; we are improving the services that we support. We are the ones that carry the pager. We are the ones that get called when things break. That’s our role at Google. We are not ops. We are improving the reliability of the services that we are on the hook for, and we do that largely through automation and software.

QCon: Your talk at QCon San Francisco is on what Google learned about creating effective teams. You use a championship model, not an all star team model. Can you elaborate on that?

Matt: What we originally thought (and what we found in the research as well) was that if you hire the right people, they will do the right things. This line of thinking predates my time here and comes from Larry’s early days at Google. He had said that you don’t even need project managers, program managers, or people managers if you hire the right people to begin with. We quickly realized that wasn’t correct and also found that there is no guarantee that a team will perform well even if you hire really brilliant people for every position. This is counter intuitive, as you might think that the best way to hire is to choose the best people for every open position (then let them do their thing), but that doesn’t always guarantee success. 

In fact, as far as effectiveness goes at Google, we found that how the team treats each other is much more important than who is on the team. That was a very surprising aspect of the research because, again, it was counter intuitive to what we had thought. We were thinking of it like a drafting a basketball team. You pick the right point guard, the right shooting guard, small forward and so on, and it’s all going to work. But it doesn’t, and there are key reasons why it doesn’t. That’s what Julia Rozovsky’s research has proven.

QCon: It reminds me of Vince Lombardi saying something about “You want the best 11 on the field, not the best 11 players on the field.”

Matt: Exactly. A good analogy for the Bay Area is to look at the San Francisco Giants in 2010, 2012, and 2014. If you look at each position, they were not the strongest team. Yet despite that, they played together, everyone performed their role to perfection, and they were able to win three championships in a really short period of time. That’s not by accident. 

It’s different when you just pick the best player and put him in his position. I guarantee you that a team that does will not be a championship team. It takes a team where everyone knows their role and plays well together. 

QCon: How are you going to put this talk together in order to help people learn to build great teams with balance?

Matt: I will start off with the data and share the latest findings that we have. Then I will share my story. It’s my experience with the teams that I have been on (the pitfalls and the discoveries I’ve made during my career). I will share the things that I’ve found really make a difference with teams at Google and other places where the teams have had a really great synergy.  I will also discuss what not to do. I made a lot of mistakes and some bad assumptions. I can share lessons learned from those experiences. 

The practical part of the talk focuses on the small things you can do as a leader to instill trust, what we call “psychological safety”, in the study. This is a force multiplier if you are able to get people who are willing to be vulnerable, make mistakes in front of each other, and take risks. These aren’t product risks but rather personal risks. They know they can openly express their ideas and not be shot down by someone who says, “this is never going to work.”

If people are really open (and there is diversity of thought), the team will have a great foundation of trust that you can then build on. Without that solid foundation it doesn’t matter how brilliant your people are, you will never really hit your stride as a team. There will always be people who won’t speak up or share their ideas. As a leader, your role is to make sure everyone is showing up with their full self, expressing their ideas, and performing at their best. 

That’s the art of leadership: getting people to do more than they normally would because they are a part of something bigger than themselves. That’s the secret sauce and the number one way to do that is to instill psychological safety. There are lots of small things that you can do to foster an environment where people feel safe and comfortable being their whole selves. It doesn’t have to be sitting around a campfire and playing the guitar. It’s a slight change in the way you run meetings. A small change in the way you interact with your people in front of others, the way people react when ideas are exchanged. All these things are small changes but can have a really big effect and payoff in the long run. 

QCon: Can you give an example of how you might take that first step with a team?

Matt: I had a team before where we had a very strong tech lead. He was a brilliant coder (an awesome engineer), but was very difficult to get along with. He would run over everyone else's ideas and being the more senior engineer, other people on the team would naturally defer to him. I worked with him to try and get him to be more inclusive and open to other people’s ideas. As brilliant as he was, he didn’t know everything and it was important for the team as a whole that diverse solutions were floated and not just his ideas. I was never really successful in getting him to change. However, I was able to get the other people on the team to feel comfortable enough to share their ideas and not be afraid of reprisal or his strong personality. I socialized a lot with the rest of the team and showed people that we really valued their ideas. In meetings, I would make sure to jump in and say “Hang on. Let’s listen to what Mike or Steven has to say here and let’s hear it all the way out.” It wasn’t optimal but it was definitely better than it had been in the past.

When he left, the team had built up this level of trust and really took off from there. That’s not to say that you need to find a bad apple and get them to leave. But through some small changes the seed was planted, and it had a profound effect on the team. When the junior engineers eventually became tech leads themselves they were very inclusive; they gave everyone the opportunity to share their piece, and no idea was too crazy to be on the table. Small changes like that are also an example to other teams within the company and can have a much more far reaching influence than just your immediate team. 

QCon: What are the five dynamics of an effective team?

Matt: The research has found that psychological safety is at the bottom of the pyramid, so it is the first foundational dynamic of an effective team. It’s critical that people can show up as their whole selves at work, that they treat each other with respect, and that they are able to try things that are not perfect. 

Teams that have a performance environment eventually run into problems, because the team is made up of high performers who all want to get results. Seems to be pretty intuitive, right? Well, if your team environment is only about that, people aren’t going to want to make mistakes because they always want to be performing at their best. They don’t want to look incompetent or make any mistakes. It’s impossible to learn in environments like this. What you want is more of a classroom culture where the team is in learning mode. With learning environments, people feel comfortable asking questions, sharing ideas that seem crazy, and are open to trying things that don’t necessarily work out. Teams with learning environments are more effective because they have that strong culture of trust. 

The second one is dependability. Can you depend on your teammates to do what they say they will do, when they say they will do it? It’s a bit like the sports analogy from earlier. If you’re the point guard and are worried that every time you pass to a certain forward, they aren’t going to catch the ball, then you’re going to be a little hesitant to pass the ball to that person. You need to be able to depend on your teammates to do their job. 

The third is structure and clarity. This is interesting because with highly dependable teams you don’t need a lot of structure. More structure does not equal better results. You need enough structure and clarity that people know their roles and responsibilities but not so much that their creativity is stifled. We found in the research that adding more structure to highly dependable teams actually hurts more than it helps. On teams that are not dependable, though, adding more structure helps the team overall. People who are not dependable usually need more structure because they are not getting their deliverables done on time. 

You need to have more policies and procedures or more SLOs around their deliverables so that they can start delivering. As that happens, people start trusting more and then dependability goes up. The takeaway here is that if things are working, and you have a highly dependable team, you don’t need to add any more structure. You can just leave it.

The fourth one is meaning. Is the work meaningful to you? It’s not about whether the work is meaningful to the company, but is your work personally meaningful to you? I work a lot with our Safe Browsing team and their mission is to keep users safe. They perform mission critical tasks for Google, but on a personal level, they also all find meaning in their work. 

The last one is impact. If your team is built on a foundation of psychological safety, when it’s dependable, has just enough structure, and people find meaning in their work, then you can say that it’s a team with great impact for the company and the people. They accomplish what they say they will and they are performing at a very high level. This is an effective team and will be seen as such on several levels- from the individual team members themselves, people outside the team, and company executives. 

QCon: What do you want people to walk out of the session with?

Matt: That the how matters more than the who. Pay attention to how the team interacts and find a way to foster a learning environment where people are free to try crazy things and make mistakes without fear of reprisal. If you can get to that point, then you are going to have a team that has the first critical building block of a psychological safety net that will get it on track to being a very effective team. 

Psychological safety is not an easy thing to establish. Once you are on a team that has it, it is really hard to go to a team that doesn’t have it. You notice right away. People say to me, “Hey Matt, why don’t you run this other team?”, and I don’t know if I want to struggle again. You get to this point where the culture is so strong. Everyone is developing, everyone has each other’s back. They love coming to work, they are energized, and they trust each other on a very deep level. They know good things are going to happen when they show up for work. That’s where you want to be and making some small changes can help you get there. 

Speaker: Matt Sakaguchi

Site Reliability Manager @Google

Matt leads a Site Reliability Engineering (SRE) team of 10 in San Francisco. Matt's team is one of many in Corporate Engineering SRE whose mission is to enable all Google employees to be the most effective workforce in the world, by providing robust infrastructure, scalable services, and seamless access to all the tools they need to be productive. Previously, he was the Technical Lead (TL) for Postini-SRE for several years. Prior to Google, Matt was a Systems Administrator, Network Operations Center (NOC) Manager, and Site Operations Manager for and other Tech companies. Before joining the Tech Industry, Matt was a full time police officer for 8 years; duties included: patrol, detectives, Drug Abuse Resistance Education (DARE), Firearms Instructor, SWAT team, and Special Investigations team (evidence collection).

Find Matt Sakaguchi at



Monday Nov 7

Tuesday Nov 8

Wednesday Nov 9

Conference for Professional Software Developers