Imagine that we are setting up a world basketball tournament. You are given first pick to draft players for your team and you end up with the line up in the picture below –
This line up includes some of the all time greats — Lebron James, Dwayne Wade, Carmello Anthony, Tim Duncan, Amar’e Stoudemire and Allen Iverson to name a few. The best of the active NBA players in 2004. How do you think this team would match up against others? This was the 2004 USA Olympic Men’s Basketball team. They ended up placing third (out of 12) in the competition after conceding losses to Puerto Rico, Argentina and Lithuania. It was a spectacular example of a team performing below its talent level.
There is a general assumption that if you want to get the best performance out of a team you have to hire the best talent available on to that team. This is a remnant of the old-school management thinking that concentrates on resource efficiency. The general idea is that our people are individual resources and we get the best from them if these are the best resources we can find. Most modern management methods employ teams to achieve goals for the organization. Are the concepts of individual resource management compatible with the employment of teams? How much does cultivating individual talent matter on teams?
While we have evolved our methods to be more team oriented, our hiring, firing and promoting practices have not kept pace with that evolution. Organizations and managers believe that when it comes to forming teams, having A+ players is the only option. Is this true? Is their a direct relationship between the amount of top talent on teams and the performance of teams? We don’t have much data on software development teams, but there is a great amount of data and research available for sports teams. Most of this post draws inferences from data and conclusions made in the paper — “The too-much-talent effect: Team interdependence determines when more talent is too much versus not enough”. It is a great read and this post is going to do very little justice to the contents of it.
There are two sections of the paper that attract attention. One is a study of the performance of teams in the NBA and another about performance of teams in the MLB. Both the studies use existing and available rating systems to rate players in the league. Next, they figure out what team performance looks when plotted against the percentage of top talent on the team. Lets take a look at these one at a time.
The NBA players are rated using the statistics of Estimated Wins Added (EWA). EWA is a measure of how many wins the player adds to his team’s season as compared to a replacement player. The teams were rated using a simple end of year win percentages for a 10 year period. The players were rated for this same 10 yer period as well. The resulting graph looks like the one below —
There is a distinct upwards trend as the percentage of top talent goes from 0 to 50. After that though, the trend flattens out and then becomes negative. In fact, the data shows that having 80% of your team be a top talent is as effective as having 20%. This makes very little sense when looked at through the lenses of conventional thinking. What is the reason for better rated (and probably more expensive) athletes combining to perform worse than less talented ones? The very next graph in the study provides some hints that help us answer the question.
The following graph plots the percentage of top talent on the team to Intrateam Coordination. Intrateam Coordination is calculated using a combination of assists per game, field-goal percentage (since better coordination leads to better chances) and defensive rebounds. These factors are combined into a single measure that reflects how well the players on the team collaborate and coordinate to win games.
The two graphs are remarkably similar. It seems that while acquiring more talent helps a team gets more coordinated, after a certain point, more talent starts to hurt coordination on the team. There is another study in the paper that attempts to explain this effect. Without going into too many details, the study concludes that the competition within teammates to be “the player” on the team hurts coordination and consequently the team’s performance. As teammates jostle against each other to create more individual “Sportscenter” moments, the team loses out.
The same paper has a similar study on Baseball teams and players in the MLB. Top talent in this case is rated using the WAR (wins above replacement) rating, which is very similar to the EWA used for basketball players. Those players that were in the top 33% according to WAR ratings were considered to be top talent. Team performance was measured that same way as in the basketball study as well. The win-loss percentage over the last 10 seasons was the measure of how well the team has performed. The resulting plot is as follows —
Notice a difference? Teams with greater amount of talent almost always perform better than teams with less talent. The results are similar to those for NBA teams for the initial part of the graph but as opposed to the NBA graph the line keeps going up as level of talent increases. The reason for this is that as opposed to basketball, plays in baseball require much less intra-team coordination. Most plays rely on individual brilliance rather than team coordination. This allows individually brilliant players to strive for “Sportscenter” moments without affecting the performance of the overall team.
Baseball And Basketball Are Different, So What?
What does this study and its results have to do with Agile teams? There are multiple implications of what we have learned from sports teams. These implications should have a bearing on how we make decisions around hiring, firing, promoting and demoting people within the organization. I have noticed these parallels in teams I have been a part of and also across the broader organization.
When hiring folks onto the team, given that we already have good talent on the team, collaboration and team fit should be more important. In that case, do not hire the most talented developer you come across, instead hire the one that fits the team the best.
In general, teams in an Agile process are looked at as highly collaborative. The first question we need to answer is — Is this true for the team we are dealing with? Are we building a basketball team or a baseball team? An easy way to discover this might be mapping out the value stream for the team. Start from the initial request all the way to the goal. Does meeting the goals of this team require a lot of “passes” or hand-offs or does it require team members to hit “home runs” and they can achieve these goals without interacting with other team members.
If we determine that this is a basketball-type team, we need to reflect this in the make up of the team. When hiring folks onto the team, given that we already have good talent on the team, collaboration and team fit should be more important. In that case, do not hire the most talented developer you come across, instead hire the one that fits the team the best. For a basketball team, based on which side of the curve we are, we can make the decision of hiring top talent vs collaborative players. Ideally of course, we hire top talent that can collaborate well. Similar to what Miami Heat was able to do with Wade, Lebron and Bosh. On the other side of things, if you are looking to fire folks, you should not just look at your “lowest” talent individuals. Look for the ones that decrease collaboration and remove them from the team. You might actually end up getting much better results as the percentage of top talent will move in the appropriate direction.
If we determine that each team member can work independently and there are minimal hand-offs, we might take a different approach. When building a Basketball-style team top talent results in top results. Every opening you have, get a top talent in. If you are looking to fire, figure out who your lowest performers are. These decisions are much simpler for baseball teams.
Most product development teams are Basketball teams. There is a lot of collaboration and coordination required in order to produce valuable software. We need to understand when making the decisions of moving people on or off the team, how it effects the talent-coordination balance. More top talent on the team does not help these teams, better coordination does.
At an organization level, each team acts as an individual. In essence, a development organization is a team of teams. We again have the same choice to make. Do we need our department to work as a basketball team or a baseball team? In my opinion, contrary to the team level, the development organizations teams should be baseball teams. Each of them, independently capable of achieving great things. For this to be successful, we have to be able to create the baseball environment. We have to create an environment that allows each team to be able to accomplish its goals independently.
If there is a high level of interdependence between teams, our organization is a basketball team. In this case, as we continue to improve every individual team, the overall performance of the organization will start to decline. Since the teams are optimized for and focused on achieving their own goals, requests from other teams are considered interruptions. These interruptions cause angst as they get in the way of the team’s own goals. The angst is only reciprocated on the requesting team’s side as their success is being hindered by another team. The more coordination that is required and the more inter-dependencies that exist between these teams, the slower the overall progress will be.
…as we continue to improve every individual team, the overall performance of the organization will start to decline. Since the teams are optimized for and focused on achieving their own goals, requests from other teams are considered interruptions. These interruptions cause angst as they get in the way of the team’s own goals.
If 80% of our teams are great teams, we need to create a baseball system, where they can achieve great things independent of each other. Make each team capable of achieving their goals without too many hand-offs. Teams that are individually great should not have to rely on other teams to get the job done. They should be able to hit “home runs” without having to rely on other teams. This requires us to create the right environment and promote the right capabilities. The team should be able to do everything from eliciting requirements from customers to getting feedback on delivered code. With no other teams impeding them, the team can actually deliver on its talent. Consequently, by running a baseball style system, the entire org wins more often.
A Team At Every Level
Most Organizations are hierarchical. Managers lead teams that report to directors, who report to VPs and so on. At each level, these ranking members are a part of the management team that corresponds to that level. Here again, we need to make the decision of do we want this level to act as a baseball or a basketball style team. Maybe, the managers are a baseball style team as they are supposed to make their individual teams as successful as possible. Also, the Directors could be a basketball style team as they have to work together to collectively ensure that the org is heading in the right direction. The question is how much collaboration do we expect at each level.
This needs to be factored into our decisions regarding promotions. Promoting a superstar manager to a director position could hinge upon his/her ability to collaborate with the rest of the director team. A highly talented manager that does not work well with others would end up hurting the director team a lot more than helping it. Obviously, a dysfunctional team at the director level could spell disaster for the organization as a whole. It all starts with defining how much collaboration we need at each level of our organization.
I would love to believe that there will be basketball teams all the way up and down the organization. It seems that organizations are not necessarily structured to promote that. Some departments and certain levels of the hierarchy seem to thrive better in the individual glory of baseball. Whatever the case maybe, building and evolving teams is not a one size fits all. We have to understand the kind of work and the level of collaboration we want from this team. We can, accordingly adjust the team to be more of a baseball or a basketball team.
We live in a fluid world, and our expectations from these teams are going to be fluid as well. At some point, we might expect a baseball-type team to start collaborating more or a basketball team to hit individual home runs. When this does happen, we need to make the appropriate adjustments. This might mean changing out a few team members. Promoting or demoting some folks. Investing in training to improve expertise. Investing in creating a collaborative team structure. We need to pick and choose the strategies that would move the team in the direction we want them to move into.
Rewarding Basketball And Baseball Teams
In this post, we have talked about creating the environment for teams to thrive in. One of the critical elements of the culture and the environment teams work in is the bonus and reward program. What are the incentives that we are putting in front of our people. What kinds of behaviours are being rewarded and what type of behaviour is not.
If we expect our teams to work in a collaborative manner, do we reward the teams that show the highest degree of collaboration? Or do we reward individuals that knock it out of the park? If we have awards and bonuses for individuals who are “MVP”s or “All-Star”s or “Employee Of The Year”, we should not be surprised when people chase individual glory rather than team goals. When folks working on basketball-type teams are handed out awards for individual excellence, do not expect them to take the time out assist other team members. There will be more long range three point shots taken than teammates setting up easy lay ups.
The rewards for baseball teams should be more individual oriented, but for basketball teams, they should be team oriented. The team should learn to sink or swim as one. Either they all get rewarded or none of them does. They get rewarded for winning championships, not for one individual scoring the most points while the team does not even make the playoffs. We need to reward the behaviour we want to promote. We cannot promote collaboration through individual rewards. We can only harm it and the team as a result. Optimizing one individual’s performance in a collaborative team is often at the cost of the team’s overall performance.
All the talent and all the collaboration in the world will not get us anywhere if the team does not have a defined direction. A basketball team has clear goals — Score as many points as possible and keep the other team from scoring points. Every team, at every level of the organization needs to know the goal it is working towards. Otherwise, we can spend a lot of money on acquiring top talent and invest a lot of time developing collaboration in order to achieve very little.
Determine the type of team you need. Build it so that it sits high on the talent-collaboration curve. Make sure the goal is clear. After that, sit back and watch them win championships.
Reproduced with kind permission of the author. Original post found at Basketball, Baseball and Agile Teams.