Team building
Most icebreakers fail for one reason: they ask people to perform. The ones that work are short, safe to answer badly, and finish with one question that connects what just happened to the meeting you are about to have. Here are seventeen that hold up, with how to run them and what to ask afterwards.
An icebreaker is the first two minutes of the meeting, not a warm-up for it, and it tells people what kind of room this is. If the opener is safe to answer badly, the next hour usually is too. If it feels like a test, people go quiet and stay quiet.
If you cannot say in one sentence why you are running it, don't run it.
| Activity | Group size | Time | What it surfaces |
|---|---|---|---|
| One-word check-in | Any | 2 min | The gap between the room's mood and your agenda |
| Weather report | Any | 2 min | Pressure people won't state directly |
| Peak, pit, pivot | Up to 8 | 5 min | Workload and blockers, before status |
| The object on your desk | Any | 3 min | Every voice out loud early |
| One to ten | Any | 3 min | Hidden disagreement about readiness |
| Two truths and a lie | 4-15 | 10 min | How confidently we judge on thin evidence |
| Five things in common | 8-40 | 8 min | Whether a group self-organises or waits |
| Ask me about | 6-40 | 6 min | Expertise no job title advertises |
| Speed pairs | 12-40 | 10 min | Everyone speaks without addressing a room |
| Where you woke up | Distributed | 5 min | Local context invisible on a call |
| Back-to-back drawing | Pairs | 12 min | The cost of one-way communication |
| Marshmallow challenge | 4 per team | 25 min | Whether a team tests early or plans in theory |
| Lost at sea | 4-6 per team | 30 min | Whether the group beats its best individual |
| Silent line-up | 8-30 | 5 min | How a group invents a system under constraint |
| Chat waterfall | Any, remote | 3 min | Independent answers, nobody copying |
| Scavenger hunt | Any, remote | 5 min | Energy, cameras on, bodies moving |
| One image for your week | Up to 10 | 10 min | How badly written updates get misread |
One word for how you are arriving. No sentence, no justification. Round the room with eight people, use the chat with thirty. It shows the gap between the room's state and your agenda: if six people type "swamped", your ambitious plan is not landing today.
Debrief: "Is there anything in those words we should deal with before we start?"
Same shape, better cover. Describe your week as weather: sunny, patchy fog, storm arriving Thursday. The metaphor lets people signal strain without explaining themselves, which is why reserved groups take to it faster than a direct question.
Debrief: "Whose forecast should the rest of us pay attention to this week?"
One high from last week, one low, one thing that has changed. Thirty seconds each, capped at eight people or it swallows the meeting. Set the boundary first: work only, unless someone volunteers otherwise.
Debrief: "Is there a pit here that someone else in this room could remove?" That turns a round of updates into a reallocation of help.
Everyone holds up one thing within arm's reach and says why it is there. Nothing profound comes of it. Its job is to get every voice into the room in the first two minutes, which changes the odds they speak again later.
Ask one question with a numerical answer: how ready are we for Friday, one to ten? Everyone gives a number, then six words if they want. The spread matters more than the average. A room averaging seven on answers of three and ten agrees about nothing.
Debrief: ask the lowest number to explain first, then the highest. If the leader answers first, everyone clusters within a point of them, so the leader goes last.
Three statements about yourself, one false, and the group guesses. Ten minutes for up to fifteen people. It has a flaw: it rewards confident improvisers and quietly pressures everyone else to produce something interesting about their private life. Bound it instead. "Two truths and a lie about your working life" is safer and more useful.
Debrief: "What made you believe the lie?" The content is not the trivia, it is how confidently we read people on almost no evidence.
Groups of four, five minutes, five things everyone shares. Rule out the obvious: works here, owns a laptop, is a mammal.
Debrief: "How did you find them? Did you take turns, or did one person interview the rest?" Whatever pattern the group fell into in five minutes is roughly the one it will use on real work.
Everyone writes one line: "Ask me about ___." Something they know or have done, not something they own on a plan. Post them in the chat or on a wall. In cross-functional groups this surfaces expertise no job title advertises.
Debrief: "Whose line surprised you? Go and ask them this week." Without that instruction it evaporates by Thursday.
Pairs, one question, three minutes, then rotate. Two or three rounds. Best for twelve to forty people who barely know each other: everyone talks, nobody performs to a room.
Where it goes wrong is hybrid. If half the group is in a room and half is on a call, run every round in breakout rooms so the remote half are participants rather than spectators.
For distributed teams. Each person names their city and one thing that is true there today: weather, a public holiday, an election, roadworks. It puts context that is invisible on a video call onto the table.
Debrief: "Does anything there change how we plan this week?" Often it does, and the alternative was finding out on Thursday.
These are short experiential exercises rather than icebreakers. Use them to open an offsite or a kick-off, not a Tuesday stand-up. Each produces a behaviour you can point back to later.
Pairs sit back to back. One holds a simple line drawing and describes it; the other draws it, and in round one may not ask a question. Round two, questions allowed. Compare both attempts. Twelve minutes. It shows the cost of one-way communication better than any slide: the round-two drawings are visibly better and nobody needs telling why.
Debrief: "What was the first question you wanted to ask in round one, and where does that same silence happen here?" Pairs well with active listening work.
Teams of four. Twenty sticks of spaghetti, a metre of tape, a metre of string, one marshmallow, eighteen minutes, tallest freestanding structure with the marshmallow on top. It reveals whether a team tests early or plans in the abstract and finds out with three minutes left that the marshmallow is heavier than it looks. It also shows who picks up the pen.
Debrief: "When did the marshmallow first go on? What would have happened if you had tried that at minute two?" Skip the debrief and you have run a craft activity.
Individuals privately rank fifteen salvaged items by survival value, then the group must agree one ranking. Thirty minutes, teams of four to six. Score both against the published expert ranking. The group usually beats its own best individual, unless one confident voice dominated. Either result is worth the half hour.
Debrief: "Whose suggestion got dropped, and was it dropped for being wrong or for being quiet?" Ban voting before you start: the learning is in the disagreement, and a vote is how groups skip it.
The group arranges itself in order of something without speaking: years here, distance travelled today, number of languages spoken. Five minutes. Gestures allowed, words not.
Debrief: "Who started the system everyone else followed, and how did the rest of you know to follow it?"
Use this instead of the human knot, which achieves something similar but requires holding colleagues' hands and contorting. That excludes people with injuries, mobility limits, or a reasonable objection to being touched at work.
Short openers only. For anything longer, see virtual team building activities.
Pose one question. Everyone types an answer and does not press enter. Count down from three, everyone sends at once. Nobody has read anyone else's answer, so you get thirty independent responses instead of one answer and twenty-nine agreements with it.
Debrief: read out two answers that disagree and ask both people to say more.
Name a category and give people sixty seconds to fetch something matching it: an object that has survived three house moves, something the colour of your worst meeting. Two rounds maximum. It gets people out of their chairs, which is most of the point on a long call.
Avoid categories that expose living circumstances. "Show us your view" lands badly on anyone working at a kitchen table.
Everyone drops one image into the chat with no caption. Photo, meme, screenshot, chart. The group guesses what it means before the owner explains. Ten minutes for eight people.
Debrief: "Whose image did the group read completely differently from what they meant?" That gap is the one that swallows written updates every week.
Every activity here ends the same way: one question about what just happened, two answers, one line drawn to the work. That is the whole method, and it is what separates facilitation from entertainment. If you remember one question, use this one: what did that just show us about how we work together?
Which opener fits depends on where the team is. A new group needs low stakes and structure. A team mid-disagreement needs something that surfaces it safely rather than papering over it, which is easier to judge once you know the stage the team is in and how much psychological safety it has to spend.
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
A one-word check-in or a one-to-ten readiness question. Both take under two minutes, get every voice into the room early, and tell you whether your agenda is realistic today. Anything longer than a tenth of the meeting will feel like an imposition.
For a routine meeting, two to five minutes. For a kick-off, offsite or training day where the team itself is the subject, fifteen to thirty minutes is reasonable because the activity is doing real work. The test is whether you can explain in one sentence why you are running it.
Anything that avoids a queue of individual answers. Speed pairs, a chat waterfall, five things in common in small groups, or an "ask me about" wall all scale past thirty people because everyone participates at the same time rather than in turn.
Chat waterfall works best on large calls because everyone answers simultaneously and nobody anchors on the first response. For smaller calls, try one image for your week, a sixty-second scavenger hunt, or a where-you-woke-up round for distributed teams.
Anything touching family, money, health, religion, politics, relationships or personal history. Also avoid "most embarrassing moment" and "greatest fear" style questions. They look playful, but somebody in the room always has a genuine answer they cannot give, and the polite deflection costs them more than it costs you.
People hate icebreakers that have no point, no opt-out and no debrief, which describes most of them. An opener that takes two minutes, lets people pass, and ends with one question linking it to the meeting rarely gets the eye-roll. The complaint is nearly always about execution, not the idea.
Keep reading
Longer formats for remote and hybrid teams, with the same debrief-first approach.
GlossaryThe skill that turns an activity into learning people remember.
Free ยท 10 minutesA short team scenario you can run at the start of any meeting.
Get in touch
Book a 30-minute discovery call and we'll shape an experiential programme around your goals.
Book a discovery call →