Team building
A new team's problem is not bonding. It is that nobody yet knows how anyone else works, decides or reacts when things go wrong. The thirteen activities below are chosen for that gap, and each comes with setup, timing, what it surfaces and the debrief question that turns a game into something the team can use on Monday.
Most team building writing is aimed at teams that already exist. A team in its first month has a different problem: no shared shorthand, no agreed standards, no evidence about what happens the first time two people disagree. Activities built for a five-year-old team land badly here, because they assume a familiarity nobody has.
Three things separate a session people quote in month six from an hour they resent.
New teams sit in the forming stage: agreeable, tentative, privately working out whether it is safe to object. Read that as alignment and you will be surprised in month three. See Tuckman's stages.
| Activity | Group size | Time | What it surfaces |
|---|---|---|---|
| Two truths and a lie | 6-20 | 15 min | Assumptions already forming about each other |
| Working history line | 4-12 | 30 min | Where people's standards come from |
| Skills market | Any | 20 min | Hidden capability, unnoticed overlap |
| Personal user manuals | Any | 30 min | How to work with each individual |
| Working agreements | 4-15 | 60-90 min | Conflicting defaults, before they cost anything |
| Expectations exchange | Any | 40 min | Unspoken needs between roles |
| Pre-mortem | 4-20 | 45 min | Doubts nobody will say aloud in week one |
| Marshmallow challenge | Teams of 4 | 30 min | Planning versus testing, quiet takeovers |
| Back-to-back drawing | Pairs | 20 min | How much of a brief survives one hop |
| Lost at sea | 5-12 | 45 min | Whether quiet expertise gets heard |
| Minefield | Pairs | 20 min | Giving instructions under pressure |
| Onboarding scavenger hunt | Teams of 3 | 30 min | Knowledge that lives in one head |
| Pair for a purpose | Any | 15 min weekly | Who knows whom well enough to ask |
Everyone writes two true statements about themselves and one plausible lie; the group votes. Six to twenty people, fifteen minutes. Most facilitators stop at the reveal, which is why it has a reputation as filler. Run the second half: ask why people guessed as they did.
Reveals: the assumptions people are already making on day three, usually built on job title, accent or how fast someone talks. Ask afterwards: what did we assume about someone that turned out to be wrong, and what were we assuming it from? The icebreaker generator has more low-risk openers.
Each person sketches a timeline of their working life and marks three points: a job that shaped how they work, a piece of work they are proud of, something they learned the hard way. Four to twelve people, thirty minutes including sharing. Paper, whiteboard or a slide each.
Reveals: where people's standards and reflexes come from, which explains most early friction. Ask afterwards: what in your history would we never guess from your job title? Keep the third marker optional; nobody should feel obliged to confess a failure to people they met last week.
Three columns each: what I am good at, what I want to get better at, what I will take off someone else's plate. Post them in one shared document, then spend ten minutes matching. Any size, twenty minutes.
Reveals: capability the org chart hides, plus duplication nobody has noticed. Ask afterwards: who have you just discovered you should be talking to every week?
These four are the highest-value sessions for a new team, and the ones most often skipped for something more entertaining. Run them in the first fortnight, while everything is negotiable.
Each person writes half a page: how I like to receive feedback, how I behave under pressure, the fastest way to get a decision out of me, what people misread about me. Any size, thirty minutes to write, then everyone reads everyone else's. Do not have people present them aloud in turn; that converts self-knowledge into performance art.
Reveals: mismatches you would otherwise discover mid-crisis. Ask afterwards: whose manual surprised you, and what will you do differently because of it? Add a shared standard for feedback, such as the SBI model, so the first criticism does not arrive as a personality clash.
The team writes its own rules: meeting norms, response times, how decisions get made, what counts as done, how disagreement gets raised. Four to fifteen people, sixty to ninety minutes, facilitated. Adopt only what people will actually keep: six real agreements beat twenty aspirational ones.
Reveals: how far apart people's unexamined defaults are on things nobody thought were debatable. Ask afterwards: which of these will be hardest for you personally to keep? Diary a review for eight weeks out, and read facilitation if you are running it yourself.
Everyone answers two questions in writing: what I need from you to do my job well, and what you can expect from me. Swap, read, negotiate the gaps out loud. Any size, forty minutes.
Reveals: demands people are making silently, and requests they are too new to voice. Ask afterwards: what did someone ask for that you had no idea they needed? Pairs well with early one-to-ones.
Tell the team it is a year from now and the project has failed badly. Everyone writes, silently and anonymously, why. Collect, cluster, then pick two or three risks worth acting on this month. Four to twenty people, forty-five minutes. The technique is generally credited to the psychologist Gary Klein.
Reveals: the doubts a new team is far too polite to raise in week one. The anonymity is not optional. Ask afterwards: which risk did more than one of us write down, and who owns it from today?
These are experiential in the proper sense: the team does something, then examines what it did. The doing is the cheap part. See experiential learning for why the second half sticks.
Teams of four get twenty sticks of spaghetti, a metre of tape, a metre of string and one marshmallow. Eighteen minutes to build the tallest free-standing structure with the marshmallow on top. Designed by Peter Skillman and popularised by Tom Wujec.
Reveals: whether the team plans or prototypes, and who takes over without being asked. The classic pattern is leaving the marshmallow until the last minute and finding out too late. Ask afterwards: when did we stop testing our assumption, and who noticed but did not say?
Pairs sit back to back. One holds a simple diagram and describes it; the other draws it without seeing it and without asking questions. Ten minutes, then swap with questions allowed.
Reveals: how much of a brief evaporates in one hop, and the difference questions make. Ask afterwards: what would have taken ten seconds to say and saved you five minutes? If anyone has a visual impairment, swap the diagram for a small block model built by touch. See active listening.
The group ranks fifteen salvaged items by survival value: alone, then as a team, then against an expert ranking. Five to twelve people, forty-five minutes. Any structured consensus exercise will do; the ranking itself is not the point.
Reveals: whether the team's answer beats its best individual, which happens only when quiet expertise gets airtime. Ask afterwards: whose individual score was best, did we listen to them, and if not, what stopped us?
Scatter objects across a floor. One person is blindfolded or simply closes their eyes; their partner talks them across from the edge without touching them. Twenty minutes, everyone in pairs.
Reveals: how instructions degrade under pressure, and how fast people drop a shared vocabulary. It excludes anyone with limited mobility or balance, so offer a seated version as standard: directing a partner's hand around a printed grid. Ask afterwards: what did your guide say that you could not act on?
Distributed teams form slowly because nothing happens by accident: you schedule what an office gives you for free. More in virtual team building activities.
Give small groups a list of things to find inside the team's real systems: where the roadmap lives, who approves spend, the last decision record, the document everyone quotes. Teams of three, thirty minutes, week one.
Reveals: how much of your operating knowledge exists only in one person's head. Ask afterwards: what could none of us find, and who is fixing that this week?
Rotate pairs weekly for the first two months: a fifteen-minute call with one small shared task rather than an open-ended chat. The task is what makes it work. A blank coffee call between two strangers is an ordeal for one of them.
Reveals: who is becoming reachable, and who is still being routed through the manager. Ask afterwards: who do you now know well enough to ask a stupid question?
Sequence matters more than selection. Spend the first fortnight on the unglamorous four in the middle of this list, and the challenges afterwards have something to be measured against. Whichever you run, block twice as long for the conversation as you think you need, then hold your nerve through the first silence. That is where the useful part starts.
Tour De Force runs this as live, experiential training for teams worldwide — online and in person. Talk to us, or play Gamified learning appsThe Weekly Challenge to see the method in ten minutes.
Questions
For a team in its first month, the highest-value activities make working style explicit: personal user manuals, a working agreements session, an expectations exchange and a pre-mortem. Lighter openers like two truths and a lie or a skills market are useful for getting names and context out of the way, but they do not substitute for agreeing how the team will operate.
Through small, repeated, low-cost disclosures rather than one dramatic exercise. Trust in new teams grows fastest from shared work with visible follow-through: someone commits to something small and delivers it. Structured sessions help by making expectations explicit early, so the first disagreement is about a standard rather than about someone's character. Avoid anything demanding personal exposure in week one.
State the purpose of the team in one sentence and ask each person to say it back in their own words. Run one-to-ones with everyone. Make decision rights explicit: who decides, who is consulted, who is informed. Then give the team a small, real, shared task with a deadline. Shared work builds familiarity faster than any introduction round.
Long enough to debrief properly. A useful rule is to spend at least as much time discussing the activity as doing it, which makes ninety minutes a sensible block for one substantial exercise plus conversation. A working agreements session needs sixty to ninety minutes on its own. Full days work only if they are structured around several debriefs, not several games.
Ones with a task attached. An onboarding scavenger hunt through the team's real systems, rotating pairs with a small shared job, personal user manuals in a shared document, and a written pre-mortem all work well remotely. Open-ended virtual socials do not, because the informal side conversation that makes an in-person social bearable cannot happen on one audio channel.
Usually for one of three reasons: there was no debrief, so nothing transferred back to the work; the purpose was never stated, so it read as forced fun; or the real problem was structural. If people are unclear about roles, priorities or who decides, an activity will not fix it and will be resented for pretending to. Answer those questions first.
Keep reading
Why a brand-new team is polite first and difficult second, and what to do in each stage.
GlossaryThe condition that decides whether anyone in a new team says what they actually think.
Free toolLow-risk opening questions for a team that has just met, filtered by time and group size.
Get in touch
Book a 30-minute discovery call and we'll shape an experiential programme around your goals.
Book a discovery call →