Have you heard about Google’s Project Aristotle? It was a research project that aimed to find out what combination of personality types, skill sets, and backgrounds made up the most effective teams at Google. Are teams who hang out outside of work more effective? Do you group introverts with other introverts? Should teams share a preference for managerial style? Stuff like that.
(Bear with me—I promise this will tie back to retrospectives.)
So what did the researchers find out?
“We looked at 180 teams from all over the company,” Dubey said. “We had lots of data, but there was nothing showing that a mix of specific personality types or skills or backgrounds made any difference. The ‘who’ part of the equation didn’t seem to matter.”
—Charles Duhigg, What Google Learned From Its Quest to Build the Perfect Team
And later in the article:
Most confounding of all, two teams might have nearly identical makeups, with overlapping memberships, but radically different levels of effectiveness. “At Google, we’re good at finding patterns,” Dubey said. “There weren’t strong patterns here.”
Did you catch that? Let that sink in for a minute.
Researchers at Google, who have access to more teams and more data than perhaps at any other company in history, and who are experts at finding patterns could not find any strong patterns in what combination of individuals make up an effective team. “The ‘who’ part of the equation didn’t seem to matter.”
However, the researchers did find one attribute that effective teams shared: group norms.
In particular, they found two group norms that mattered most:
- Conversational turn-taking—everyone talked in roughly the same proportion (not necessarily at any given moment, but, say, over the course of a week)
- Psychological safety—no team member felt they would be judged for speaking up or blamed for taking risks
So… this is fascinating and all, but…
What does this have to do with retrospectives?
I’m not going to go into detail about what a retrospective (retro) is. For purposes of a working definition, I think the Agile Manifesto sums it up most succinctly:
At regular intervals, the team reflects on how to become more effective, then tunes and adjusts its behavior accordingly.
If Project Aristotle shows that individual capability is eclipsed by overall team effectiveness, and team effectiveness is enabled by group norms, then overall effectiveness is largely determined by group norms. And I know of no faster way to change group norms and to empower teams—not individuals on a team, but teams themselves—than to hold effective retros.
What follows are my personal observations on what makes a good retro. Before I joined Rackspace not-quite-2-years-ago, I had never participated in a retro. Since then I’ve participated in a delightful variety of retro styles. I’ve been in retros run by a full-time facilitator with Scrum certification (which were awesome, thanks Bretny Khamphavong!), retros run by the team manager, and retros run by individual team members. I’ve seen a spectrum of retros ranging from highly structured retros with planned activities to completely informal retros that consist of, “who wants to talk about what?”
Out of all them, I think the most effective retros can be boiled down to 4 key elements:
1. Everyone Talks
I think this is the most easily overlooked of the 4 retro elements I’m going to talk about.
Unless your team is a group of extroverts who are OK with interrupting each other, your retros almost certainly represent only a subset of your team’s members (and are only as smart as a subset of your team’s IQ) if you are not actively soliciting the viewpoints of all members on the team.
Even if your team has no natural introverts, it’s just as easy to overlook the viewpoints of new hires or junior members on the team. The “weak signals” they send can be an incredibly valuable source of action items to experiment with.
Job #1 of all the retro formats and activities out there is to get everyone on the team talking.1
When I am tasked with facilitating retros, I tend to stick with the simplest activity I know works:
- Sticky notes and pens are handed out to all participants at the beginning of the retro
- Participants are asked to write down ~N things since the last retro that they felt a) went well, or b) could use improvement, or c) make you go hmmm
- Everyone takes a turn sharing the sticky notes they wrote down
- Any burgeoning discussions are asked to be tabled until everyone has had a chance to share what they wrote
Note: while I like to be inclusive about who is welcome to attend a retro, the “everyone talks” rule puts an upper bound on how many people can productively participate in a retro.
It is difficult to take a team whose members do not practice good conversational turn taking, and change the team’s behavior so that every member—naturally, during the week—speaks in roughly the same proportion. What is easily within our power though, is to carve out time at regular intervals to explicitly solicit the viewpoint of every member on the team. And frankly, I suspect that gets us most of the way there.
2. No-Blame Discussion
Once everyone has had a chance to talk, discussion can begin. Anything team-related, especially ideas for improvement, is on-topic, but discussion typically starts with topics identified during the initial retro activity.
On the teams I’ve been on, an informal approach has worked well. People start the discussion by talking about what they think matters, the topic of discussion changes naturally, and all the important topics tend to get discussed.
If no one on your team wants to start the discussion, or if you have the opposite problem—people want to discuss particular topics in such depth that not all the important topics get discussed in the time allotted—then having someone act as a discussion facilitator can help. If you did the sticky note activity I described in the previous section, the facilitator can begin the discussion by grouping sticky notes by common themes. The themes with a lot of sticky notes are the topics to discuss first, and if there’s time left over people can discuss anything else.
The single most important rule to practice at retros is this: when reflecting on the past, no blame can be assigned to team members for past actions. Any decision or action (or inaction) is fair-game to question about how it might be handled better in the future, but it must be done in full-faith2 that the person or people in question acted in what they thought were the best interests of the team. This rule is so important that it has been called the Retrospective Prime Directive:3
Regardless of what we discover, we understand and truly believe that everyone did the best job they could, given what they knew at the time, their skills and abilities, the resources available, and the situation at hand
—Norm Kerth, Project Retrospectives: A Handbook for Team Review
What this rule does is create a psychological safe space where team members can talk openly about mistakes so the team can actually learn from them.
Even more generally than avoiding blame, if you want to create a full sense of psychological safety, your retro needs to be a place where people can question the status quo without being perceived as negative, and people can suggest new ideas without feeling stupid or intrusive.
I have been fortunate to work exclusively on teams that provided me with a sense of psychological safety, so I have no credible advice on how to get there if your team lacks it. However, what I can talk about is a time I felt vulnerable at a retro.
At the beginning of a critical new project I argued for the need to do a “re-write” story that was supposed to take one developer maybe a week. It got the go-ahead. Not quite three weeks later, myself and two other developers had finished the story. We got it done, but it was hard work and it had jeopardized the project overall. Needless to say, it was brought up at the next retro and we talked about it a lot. Decisions were questioned. Missed opportunities were identified. But not once did I feel personally called out. No one took my mistake and doubted my earnestness to help or my competence in general. Going further, we brainstormed ways that every member on the team (even the ones not directly involved) could help get us out of such situations in the future.
I can only imagine how that would have went on a team that didn’t value psychological safety. I probably would have stopped arguing for my ideas for a while (which I like to think help more often than they hurt), not just at retros but all the time. I also think I would have been soured on the idea of retros, attending them with a sense of meeting resignation instead of enthusiasm for improvement.
As a last thought on the importance of psychological safety, consider this quote from Google’s page about Project Aristotle:
[…] the safer team members feel with one another, the more likely they are to admit mistakes, to partner, and to take on new roles. And it affects pretty much every important dimension we look at for employees. Individuals on teams with higher psychological safety are less likely to leave Google, they’re more likely to harness the power of diverse ideas from their teammates, they bring in more revenue, and they’re rated as effective twice as often by executives.
So far we’ve seen how the “everyone talks” rule promotes conversational turn-taking, and the “no-blame discussion” rule creates a sense of psychological safety. These first two elements follow directly from Google’s Project Aristotle research findings. The next two elements are all about taking the advances the team has made during the retro meeting and figuring out how to carry them over to the 98% of time that the team spends outside of retros.
3. Agree On Action Items
At retros without a facilitator, it’s easy to have really good discussion all the way up to the end of the allotted time, and then everyone just leaves. There’s value in doing only that, but over time it’s easy to lose faith that talking at retros is making any difference. Even when it is making a difference, perhaps in subtle ways, if actions taken are not made explicit and brought up later, it’s still easy to lose faith.
Before wrapping up the retro, I recommend that the facilitator identify all the items discussed that could potentially be turned into action items. It is up to the team to decide which handful of the action items, if any, to agree to tackle before the next retro.4
Retro Tip: sometimes the team has seen clear mistakes and the resulting action items are both pressing and obvious. That’s great, do them. Other times everything is going fairly well and there’s nothing that needs to change. Use those opportunities to experiment. Try something new you heard about that maybe won’t be any better but shouldn’t hurt. If you do that enough times your team will eventually be in a much better place.
4. Make Action Items Visible
After the “everyone talks” rule, I think the “make action items visible” rule is the second most overlooked element of an effective retro.
On a previous team, we used to dutifully take notes about everything discussed in the retro: all the problems, all the proposed solutions, all the shoutouts, and, yes, all the action items. The notes were recorded in the proper place—the “retrospectives” section of the agile project management tool we use—and they were shared with everyone on the team. Did anyone ever look at the agreed on action items? Occasionally, usually the day before the next retro. Did they get done? Sometimes, typically when someone wanted to do it anyway before we talked about it during the retro.
There’s no quicker way to make sure something doesn’t get done than to assign it to a group with no clear individual responsibility and to leave a reminder for what needs to be done in a place that no one sees during their day-to-day routine.
On our team we solved the problem by adding the retro action items to a special section of the Kanban board that we work from everyday. It doesn’t really matter where they get recorded—maybe it’s a whiteboard, maybe you write a Slack bot that posts them every morning—the important thing is that every member of the team can’t help but glance at the retro action items at least once a day.
Once you’ve made the action items visible, the only thing left to do is make a point to talk at the next retro about what you learned from them.5
Retro Tip: depending on how you record action items / make them visible, the default state might be to “rollover” action items from one retro to the next even when they weren’t discussed. Resist this. Declare action item “bankruptcy” after every retro. If the team decides to commit to the same action item again, that’s fine, but make it an explicit decision.
Anyone can propose a time for the next retro. If you think it’s been a while since the last retro, make a point to schedule a new one. Too often I’ve seen everyone wait for some designated person, maybe the manager, to schedule a retro, and that person can get distracted because they’re busy. Besides, maybe you’re the only person on the team who knows there’s important things to discuss.
Go read up on Project Aristotle. If that doesn’t convince you that retros matter, then I don’t know what will.
I guarantee you that your team will improve over time if you regularly meet for retros that have these 4 elements:
- Everyone talks
- No-blame discussion
- Agree on action items
- Make action items visible
In my view, those are the minimum 4 elements of an effective retro.
In reality, they’re a starting point. Try them out. Figure out as a team what works and what doesn’t. If your team meets regularly to reflect on how things are going and has earnest discussions about how to improve the team as a whole, your team will get better over time. Maybe even by a lot, if Project Aristotle applies to your team too.
Feedback? Thoughts? Your experiences? You can share them on this Reddit thread.
- That’s just like my opinion, man. I recognize that retro activities are not limited to getting people to talk—they’re useful for getting people out of routine modes of thinking, for helping people understand where other team members are coming from, and a bunch of other purposes.
- Even if you truly believe that a team member did not act in the best interest of the team, and it undoubtedly happens (sometimes for good reasons even), the retro is not the time to discuss it. There are other venues to discuss such things, and when you bring it up at the retro you undermine any psychological safety that team members feel to openly discuss how things could be done better and even to try new things outside of the retro.
- I don’t know why people first started calling it the “Prime Directive,” but I hope it fares better than in Star Trek where every episode that references the Prime Directive finds some reason to break it later in the episode.
- The important thing is that it’s the team’s decision about what to tackle and the team’s responsibility whether it even gets done. I’ve never seen it done, thankfully, but if a manager or some outside force is dictating retro action items and holding the team accountable for them, then the entire point has been missed. The point is to develop a team’s sense of self-efficacy and to prompt action that the team has already agreed is worth doing.
- Again, the idea is to reflect on changes the team has accomplished and to maybe be aware of what hasn’t been done yet. If action items are routinely not accomplished, then that’s just another thing you can talk about at retros. Maybe the team is taking on too many items, or maybe they’re committing to things out of their control, or maybe the action items aren’t visible enough—that’s how the team I was on arrived at putting the action items on our Kanban board.