Methods & Tools Software Development Magazine

Software Development Magazine - Project Management, Programming, Software Testing

Scrum Expert - Articles, tools, videos, news and other resources on Agile, Scrum and Kanban

Self-Selecting Teams Part 1 - Why You Should Try Self-Selection

Sandy Mamoli, Nomad8,, @smamol, David Mole, @molio

Fast growth has a way of forcing organisational change on a business, but it also presents opportunities to try new ways of working. When one of New Zealand's biggest eCommerce providers hit a new level of growth, we saw an opportunity to drive productivity by reorganising its technology department into small, stable, Agile teams. And we decided the best way to go about it was using Self-Selection: in other words, to trust the people who work in the department - the engineers, testers, BAs, designers, and Uxers - to come up with the best solution.

1. What is self-selection and why should you care?

Self-selection is a facilitated process of letting people self-organise into small, cross-functional teams. We think it is the fastest and most efficient way to form stable teams, based on the belief that people are at their happiest and most productive if they can choose what they work on and who they work with.

To avoid confusion, we are not referring here to self-organising teams. Self-organising teams are groups of motivated individuals who work together toward a shared goal and have the ability and authority to take decisions and readily adapt to changing demands. We like self-organising teams, but that's not what this article is about.

This article is about self-selection, which is the process you use to set up self-organising teams in the first place. Self-selection happens at an organisational rather than at a team level and is a way to get everyone into teams. Another term for a self-selected team is a self-designed team.

We have done it and so can you

In this article - the first of a two-part serie - we will tell the story of how we used self-selection to decide on the structure and composition of 22 new Agile teams, a process that involved more than 150 people. We will not only share a case study but also a repeatable process for facilitating self-selection at scale - and maybe even convince you that self-selection is not only valid but highly rewarding and in all likelihood a successful approach for many organisations.

Who are we?

We are consultants who have spent several years doing transformational work with one of New Zealand's biggest eCommerce providers. If you aren't a New Zealander chances are that you've never heard of the company we are talking about. If you are a 'Kiwi', you most definitely will have: three quarters of the population have a member account on its transactional site.

The site is a popular place for Kiwis to buy, sell and trade everything from cars and antiques to clothes, crafts, property and farm gear. It is a Kiwi success story, having grown over the past 15 years to a unique position where it commands two-thirds of New Zealand's internet traffic with 1.5 billion page serves a month. And it is a fast-growing business: a year ago there were 110 employees in the technology department, now there are 190. In all, there are 400 staff and no sign of the growth slowing down.

What problem did we need to solve?

We reached a point about two years ago where the company was increasing its staff by roughly one person a week but adding new people no longer meant we were necessarily getting any more done; if anything delivery was slowing down. Somewhere along the way a web of dependencies had evolved where every person and project was reliant on someone else and there were a large number of handovers between professional groups. Projects were constantly being left on hold because there was no one available to work on them; everyone was busy somewhere else.

We wanted to avoid these delays of waiting for people to be freed from other projects, and we wanted to minimise handovers with their associated loss of tacit knowledge and create small units where the whole is greater than the sum of its parts. We decided we needed to pull people out of a complex matrix and get them into fixed, stable teams where we could make sure that one person would work on only one team, and one team would work on only one project at any time.

Happier, more productive teams

Anyone who has ever been on a high-performing team will know what it feels like when a team begins to truly gel, when everyone is committed to and enthusiastic about a shared goal and when people know each other well enough to support and hold each other accountable for great performance. These high-performing teams exist not only in software development but also in sports and in any area where a group of people need to manage their interdependencies while working towards a shared, compelling goal [1].

These teams have also been more productive. Recent research by Rally Software showed an almost a 2:1 difference in throughput between software development teams that were 95% or more dedicated compared with teams that were 50% or less dedicated. The 2014 white paper "The Impact of Agile Quantified" [2], based on the analysis of the process and performance data of nearly 10,000 teams, indicated that stable Agile teams result in up to 60% higher productivity.

One reason for greater productivity in stable teams is that they don't have to go through the team-building stages of forming, storming, norming, performing - as defined by Bruce Tuckman in 1965 [3] - over and over again. People working in small, stable teams are also happier and more content. Certainly our own internal surveys show that job satisfaction has increased since we started working in fixed teams.

The need for speed

When we first started implementing stable teams it was on a small scale, introducing one team at a time in a controlled manner. But as the company grew ever faster we realised we were going to have to be able to scale up, and quickly. People by now were starting to want to be part of this new way of working. We'd also seen measurable benefits and knew we wanted to extend the 'stable' approach right across the technology department. So the only question now was how to make it happen in a fast and efficient way.

The art of team design

We started looking into the design of teams, which some research suggests is the most important factor in team performance. Studies [4] conducted by J. Richard Hackman, for example, found 60% of the variation in team effectiveness is attributable to the design of the team, 30% to the way the team is launched, and 10% to leader coaching once the team is under way.

We believe that designing a team doesn't necessarily mean picking the best people, but rather deciding on the best combination of people based on their interdependent skills, preferences and personalities. We looked at two methods of designing teams:

  • Managerial selection: Managers decide by executive decree
  • Self-selection: People decide for themselves which team they want to work in

Managerial selection breaks when organisations grow

Managerial selection is the traditional way of deciding who should be in which team. Managers design teams based on their knowledge of people's skills, personalities and who they think would get along with whom. In a small company this often works well - a good manager is aware of relationships between people and knows the skills, personalities and preferences of each of them. Often they come up with team compositions that can be mostly right and it is a quick way to get team selection done.

Where this model breaks down is when a company grows. Managers might still know their direct reports' skills and personalities, but it becomes increasingly difficult to understand the intricacies of relationships between people as the number of relationships increases exponentially. Managerial selection made sense in its historical context of industrial factories where workers' tasks were relatively simple and repetitive, but makes much less sense in the complex and collaborative workflows of a tech department today.

The same is true of the carrot and stick approach to motivating staff, which emerged from Frederick Winslow Taylor's [5] theory of management in the 1900s which said that people charged with repetitive and boring tasks were best incentivised by monetary rewards. US author Daniel Pink turned the tables on that idea in his 2009 book Drive: The Surprising Truth About What Motivates Us [6], pointing out that today's work tasks are mainly creative and complex, and citing research that shows the best motivators in such an environment are autonomy, mastery and purpose:

  • Autonomy [7] provides employees with autonomy over some or all of the four main aspects of work: when they do it, how they do it, who they do it with and what they do.
  • Mastery allows employees to become better at a subject or task that matters to them and allows for continuous learning.
  • Purpose gives people an opportunity to fulfil their natural desire to contribute to a cause greater and more enduring than themselves.

In the light of Pink's book, knowing that highly motivated people perform best, and considering our own observations of managerial selection breaking down as a company grows, we started to look more closely at self-selection as a team design tool. What better place to start offering autonomy than by letting people decide for themselves which team they wanted to work in?

Self-Selection has a good (and interesting) track record

Self-selection is not a new or unproven idea. Leo McKinstry [8] described one of the earliest and most successful self-selections at large scale in his 2009 book "Lancaster - The Second World War's Greatest Bomber" [9] about the RAF's Lancaster bomber crews in the early 1940s. During World War II new flight crews had to be formed after short training periods and the creative solution to forming these teams quickly and efficiently was to have them self-select into teams. The result was one of the most effective, well put-together teams [10] in the history of the war.

Fast-forward to 2004 and Atlassian [11], an Australian IT solutions company, created the Ship-it day concept [12]. A 24-hour hackathon, Ship-it Day became highly successful and was cited by Daniel Pink in Drive. Ship-it Days give people 24 hours to work on whatever they want - as long as it is not part of their regular jobs - and the aim is to complete something within a 24 hour period. The idea was originally named "Fedex Day" after Fedex's 1980's slogan "When it absolutely, positively has to get there overnight".

We have had many Ship-it Days over the years at our Kiwi eCommerce provider and it is always been a joy to see an entire organisation self-organise into small teams and work away on projects of their own choosing. During our last Ship-it Day we had roughly 80 people in 15 teams working on 15 projects that all benefited the company in one way or another. We saw Ship-it Day as a study in what happens when we give a group of people complete freedom to work on what they think is important, with whomever they like, and using any approach they think will get the job done.

Here are some of our observations about what happens when people self-select:

  1. People naturally form small, cross-functional teams. Teams are between three and six people and team composition is based on skill rather than role. There is no one person per skill and t-shaped people [13] who are good at collaborating are in high demand.
  2. No one chooses to work on more than one team or project. Time and again organisations fall into the trap of optimising "resources" rather than focusing on outcomes. People often believe that multi-tasking, having people work across several projects, and focusing on resource utilisation is the key to success, when in reality it is not. It is interesting to note that when people are determined to ship, no one thinks it is best to do more than one thing at a time and nobody believes they are more valuable as specialists across teams than as generalising specialists within one team.
  3. People communicate face-to-face. There are barely any discussions about process or how to communicate. People just talk and co-ordinate and collaborate as needed. Things are much faster that way.
  4. A shared, clear goal makes everything so much easier. When people buy into the goal and know which problem they are solving and why, things become a lot easier. It is easy to make decisions and reach consensus when people understand and support the objectives and constraints around a project or product. Selecting which project to work on worked wonders for ensuring that the team had a shared and compelling goal.
  5. People are highly motivated, enjoy the experience and get lots of work done. Some of the projects people built, such as a "Is someone in the shower?" app, a virtual receptionist, or the room booking app "Get A Room", were simply incredible and are still in use.

The solution to our problem

It was partly our Ship-it Day successes that inspired us to try self-selection on a bigger scale and gave us confidence that people could in fact work like this every day, and not just in 24-hour hackathons. We realised that the most likely way to solve our challenge of getting people into well-designed teams quickly was to take the problem to the people involved. To leverage their knowledge, motivation and enthusiasm to come up with a better and more widely accepted solution than managerial selection could have given us.

2. How to Conceive and Carry Out a Self-Selection Event

Having decided that self-selection was the most efficient way to establish stable, happier and more productive teams the next question was: how do we go about it at scale? Should we follow the Lancaster Bombers' lead and just get everyone into a giant hall and tell them to get on with it? Or was there something more workable for us than that?

We tried to research the concept and ran the usual Google searches but it appeared that either no one had carried out a self-selection exercise at this scale or if they had they hadn't published the process or results. So we had to come up with our own process from scratch - one that's since been revised, tested and improved and which we are now sharing with you here. We have also shared these steps with a number of other organisations that have since run successful self-selection processes.

Before Self-Selecting: The Preparation Checklist

Preparation is incredibly important when you're taking on a self-selection exercise at scale. We have seen self-selection events fail spectacularly when preparation wasn't done well. In fact, we suggest erring on the side of over-preparation, not least because it will make you feel more confident and relaxed during the event. A relaxed facilitator is more likely to achieve a good outcome.

Readiness Check

The first step in preparation is to ask the question: "Are you in a strong enough position to do this?" Do you have an environment where this could work? Do you have the kind of people who are flexible and open enough to try a process like this? Are there any festering problems that could derail the process? Are there things you need to establish before you start on self-selection?

You may not be able to pull off self-selection if other things are too problematic. For example, if you don't already have the concept of fixed teams you may struggle to get people to choose a new home. So you might decide that you need to solve one or two problems before you start on self-selection. You can see how we went about that by reading/watching about our experience with Portfolio Kanban. [14] You might also consider running a Ship-it Day ahead of time, which will allow you to observe behaviour and see whether people naturally self-select and how well they work in teams.

Run a trial

In addition to (or instead of) a readiness check you can run a trial self-selection event to gauge the process and manage your risk. A trial can give you a lot of information at little cost and with little or no downside. We opted to carry out a trial self-selection event at our satellite office, which gave us a more controlled environment and fewer people to test our process on.

We ran the event with 20 people and started the day with just a carrier bag of sticky notes and a lot of good intentions. By the end of the day those 20 people had formed into the three teams that we had aimed to end up with and we knew we had a process that worked. Of all the things we learned that day, one of the most powerful was that our worst fears were unfounded: there were no fights, no crying in the corner, and no empty teams at the end of the day. We don't think we could have carried out our full-scale selection process without first having done this successful trial run.

Another way to trial the process could be to run a practice self-selection event where you can test the concept and refine it without any real risk. We have known companies to do this and it can work well. But be aware that people behave very differently when they are not making 'real' choices - it is like playing poker for no money: everyone goes all in because they have nothing to lose.

Define the Teams to Select

Ahead of the self-selection event, it is important to clearly define the teams that are required. This could simply be based on your current structure or gaps, or it could be a more complex proposition and even require a company-wide prioritisation to get the right teams established. Each team should have a name, a clear mission for what it will do and a product owner established in advance. When people choose what team they want to work in, they will want to understand what they are likely to be working on, the problem they are trying to solve, and who will provide guidance.

This is a potentially lengthy step but the work can be delegated to the product owners. The better they can explain their team, its mission and how they themselves work, the better chance they have of attracting a great team on self-selection day and in turn being successful. You don't want people walking out of the self-selection event feeling like they are not sure what's happening or thinking 'this is not what I signed up for'.

Logistics: where, when and who

Establishing early where and when your self-selection event will take place can help your communication effort, build trust that it will actually happen, and give confidence to people that everything is planned and under control.

The following details should be defined as early as possible:

  • What time/date will your event take place? (It is probably a good idea to have a backup date in case of problems or illness.)
  • Where will it take place? (Note, you will need a big, open collaborative area with lots of wall space.)
  • Who will be invited?

Communication: early, often and honest

Communicating well might be the single most important thing you can do to pull off a successful self-selection event. We have seen a pattern emerge when people hear about self-selection for the first time. The first reaction tends to be positive but it can move quite quickly to fear and resistance. Fear of something new and different, fear of what might happen, fear of being stuck with someone you don't get on with or of being stuck in a team that you can't change your mind about later.

People will throw a lot of questions and what-if scenarios at you and you need to be prepared to answer them honestly. Getting the communication strategy right could be the difference between your self-selection event going badly (or not taking place at all) or being a roaring success.

Based on our experience, we recommend the following approach to communication:

  • Talk to as many people as possible, from start to finish
  • Actively listen to their concerns
  • Be patient with people as they work through their fears
  • Record people's fears and what-if scenarios
  • Manage risks actively
  • Paint a very honest picture about the worst-case scenario, which is never as bad as people think
  • Talk to people individually and present in groups
  • Show real examples from your Ship-it Days, trials and from other companies
  • Ask people whether it is they, or their manager, that knows more about where they should be placed

Establish the Rules and Constraints

We recommend keeping rules and constraints to a minimum. The more there are, the harder it will be to solve the problem and the more likely that people could feel they are not self-selecting at all but simply moving into - or worse, being manipulated into - pre-chosen allocations.

We had just three rules during our self-selection events. Teams had to be:

  • Capable of delivering end-to-end
  • Made up of 3-7 people
  • Co-located

Prepare your materials and stationery

This needs to be an interactive and visual event, where everyone can see what's going on and participate throughout the process. You're going to want to start the event with some of the details already on the wall: a visualisation of the status quo, a team diagram for each team you're aiming to create, checklists for the skillsets required, a visualisation of the rules and so on. On the day, people will indicate which team they want to join by sticking their photo on that team's diagram.

So you're going to need a lot of stationery and materials including:

  • Photographs of the people participating
  • Skills checklists
  • FAQs
  • Team diagrams for the wall
  • The rules written up to display around the room

The day before our self-selection event we had what can only be described as an arts and crafts day. Cutting out photographs of everyone involved, preparing colour-coded skills checklists and preparing large team diagrams pre-populated with any information that we already knew such as team names and/or the product owner photo.

Little details are important here. You need to leave enough room on each team diagram to add the required number of people's photographs, and be careful where you put the product owner's photo in the diagram so you don't inadvertently imply an unwelcome hierarchy.

It is also important to decide whether your self-selection event will be starting from scratch or whether you will start your problem-solving process from the status quo. The advantage of starting from scratch is that it can be simpler and easier to throw out current constraints and start from a blank sheet. It can reduce the complexity of the problem and allow people to consider options they may not have otherwise realised would be possible. On the other hand, starting from the status quo makes the process very real, and it helps those people who want to stay exactly where they are (and there will probably be some) to do so and not feel like they are being bumped.

3. On the Day: Running the Self-Selection Event

Set up the Room

We recommend getting there early so you have plenty of time to set up the room, and have all your stationery, photographs and materials neatly laid out. The last thing you want is for a team to solve a problem during the day but not be able to reflect it on the finished product because you're short of blu-tac or sticky notes.

Since self-selection is a physical event, the emphasis when setting up the room should be on space and collaboration. It may be a good idea to get the group to help set up the room by moving chairs and so on, because then they will start the day as you hope they'll carry on: by moving around and talking to each other.

Product Owner Presentations

People need to know what they will be signing up for, they also need the opportunity to ask questions, so the day should start with the product owners standing up to pitch their team to the group. They should explain the purpose of the team and the type of things that they will work on. If they know their current projects those should be explained too, but projects can change and be aware that this could be the wrong level of detail. The type of product the team will support and the type of work (front end vs. back end, for example) might be more appropriate.

Explain the Rules

Talk people through the rules for the event, and have the rules displayed prominently around the room.


Start your first self-selection round. This is where people walk around talking to product owners and thinking about which team they want to join. When they choose a team, they blu-tac their photo into that team diagram. Be strict on your timing. We recommend a 10-minute timebox because that's enough time for people to have the conversations they need to have and also for them to overcome any nerves about moving or selecting a team. When the time is up, be very strict about stopping for your first 'checkpoint' (you may need a whistle!)


It is vital to check in and publicise each team's current status after each round. To do this we use a checkpoint. At the end of each timebox, everyone stops and you use your checklists and other visual indicators to show how many full teams you have at this point. If this is your first checkpoint then don't expect to have solved the problem straightaway. The hope at these checkpoints is that one group will say they are missing two developers, for example, and another group will demonstrate an over-supply of developers, thereby ensuring those two groups can talk during the next round.

One by one a self-appointed spokesman from each team should announce:

  • Whether the team is full
  • What gaps they have
  • Any problems or blocks they have encountered

Rinse and Repeat

The 10-minute timeboxes, each followed by a checkpoint, should then be repeated indefinitely until all the problems are solved, or the same problems are being repeated and people appear to be stuck. If the problem is solved then, congratulations, you can send everyone home! If the problem isn't solved it is time to change gears and tweak the format.

Tackle the outstanding problems

You may want to send people who are part of new, fully formed teams home at this point and you could change the room format to bring people closer together. If you still have problems that aren't being solved by more rounds of self-selection then you may have to dive into them head first. The detail will vary according to the specific problem being solved and the people involved. For us, the process involved tackling one obvious problem or bottleneck at a time and engaging the whole group to solve the problem.

For example we had a shortage of designers and no matter which way we cut it, we just couldn't make fully formed squads. One solution would have been to thinly spread the people we had across multiple teams, but that would have left everyone short so as a group we decided to populate as many full teams as we could and then use empty cards to represent the people we needed to hire after the event. This was great information to have: prior to the event we didn't know who we needed to hire or which teams they would join.

Go Home!

Don't stay too long, this is a surprisingly exhausting process and if you haven't solved the problem so far then it might be time to let people go away, give it more thought and come back to tackle it again the next day.

4. After the Day: Making this Real

At the end of a self-selection event you will have a lot of paper and hopefully lots of self-selected fully skilled teams. But so far this is just a lot of diagrams and you need to go about making this real. If possible, you should meet with each new team the very next day. It is vital that you build on the momentum you've created and don't let people go back to their day jobs. We found that the Lean Coffee [16] meeting format worked really well for talking to each team, allowing them to voice any concerns, and importantly to start talking about how and when to set the team in motion.

Quite often other work will need to be finished first so creating a schedule is important. It is then a case of keeping momentum going as your new teams work through their scheduled backlog and then turn their energies to new projects.

In Part 2 published in the next issue of Methods & Tools, we will talk about how to settle in your new teams, permission vs. forgiveness, and tips for getting the best out of your self-selection event.


  4. - sthash.65FkEH9y.dpuf
  13. Video:
  14. Paper:

The secont part of article has been published in the Spring 2015 issue of Methods & Tools

Agile and Scrum Resources

Scrum Expert

Agile Videos and Tutorials

Click here to view the complete list of archived articles

This article was originally published in the Winter 2014 issue of Methods & Tools

Methods & Tools
is supported by

Software Testing

The Scrum Expert