Key takeaways
- →A guess who activity works because the guessing forces people to think about each colleague individually, which small talk never does.
- →Collect the facts before the session, not during it - live collection produces thin answers and a five-minute silence.
- →One fact per round, 45 to 60 seconds to guess, and no more than six rounds in a single session.
- →Prompt the fact, do not leave it open: 'a skill you have that is not on your CV' beats 'tell us something interesting'.
- →Never include someone who did not submit, and never make the reveal a demand to explain the story unless the person offers.
- →For teams over 25, run it in fixed sub-groups so guessers actually know the people they are guessing between.
A guess who activity is simple: each person privately submits one fact about themselves, the facts are shown one at a time without names, and the group guesses who each one belongs to. It takes eight to twelve minutes for a team of a dozen, and it is one of the few remote formats that produces genuine recall of colleagues rather than a pleasant five minutes everyone forgets by lunch.
The reason it works is mechanical rather than magical. To guess, you have to run through the team in your head, one person at a time, and ask what you actually know about each of them. Most people discover mid-round that the honest answer for three or four colleagues is 'almost nothing'. That noticing is the point, and no amount of round-robin small talk produces it.
This guide covers where to run it, how to collect the facts, prompts that produce guessable answers, the timings, the safety boundaries that keep it from going wrong, and what to do in a distributed team where half the room has never met the other half.
What it is for, and what it is not for
Guess Who is a connection activity. It builds the shared background knowledge that lets a distributed team talk to each other like people rather than like Slack handles. It is not a check-in, it is not an energiser, and it is not a substitute for a team having anything to do together.
That distinction matters when you are choosing where to put it. A check-in asks how someone is right now and belongs at the start of a working session. Guess Who asks who someone is, and it belongs where you have ten minutes to spend on the team itself: a monthly team meeting, the first hour of an offsite, week two of onboarding a cohort, or the session after a reorg has thrown three groups together.
It also has a shelf life. Run it monthly with the same twelve people and by the third round the facts get thinner and the guessing gets accurate for boring reasons. Quarterly is about right for a stable team. For a team that keeps gaining new people, run it whenever the joiners outnumber the people they have met.
The run, in order
Three days before: collect
Send the prompt with a deadline. One fact per person, anonymous to everyone but you.
The day before: sanity-check
Read every submission. Cut anything identifying, anything unkind, and anything that needs a paragraph to explain.
Minute 0: set the rules
Thirty seconds. One fact at a time, everyone guesses, the person reveals themselves, no interrogation.
Minutes 1-9: six rounds
Show the fact, 45-60 seconds of guessing, reveal, ten seconds of reaction, next.
Minute 10: land it
Name one thing you learned about someone, then move straight into the agenda. Do not debrief it to death.
Collect the facts before the session
This is the single biggest determinant of whether the activity is good. Live collection - 'everyone type a fun fact now' - gives people ninety seconds under time pressure, and ninety seconds produces 'I have a dog' from nine people out of twelve. Nobody can guess between nine dog owners, and the round dies.
Collect asynchronously, two or three days ahead, with a clear deadline and a stated prompt. People think about it in the shower, they come back with something they are slightly pleased with, and your guessing rounds have texture. It also lets you screen the submissions, which you should always do.
Expect a submission rate of about seventy per cent on the first ask and near-total after one reminder. Whoever has not submitted by the deadline is simply not in the game, and you say nothing about it in the session. The rule that keeps this activity safe is that participation is opt-in and invisible: nobody should be able to tell from the session who declined.
Prompts that produce guessable facts
'Tell us something interesting about yourself' is the worst possible prompt. It is a blank page, it triggers the fear of being boring, and it produces either nothing or something so obscure that guessing is random. Constrain it instead. A narrow prompt is easier to answer and produces facts that sit in the sweet spot: surprising, but connected to something people might half-know about the person.
Rotate the prompt each time you run it. The prompt is what keeps the activity fresh long after the novelty of the format has gone.
Timing and group size
Top Tips
- Six to fifteen people is the ideal range. Everyone plausibly knows everyone, and six rounds covers a meaningful share of the team.
- Give 45 seconds per guess for a team of ten, 60 for a team of twenty. Longer than that and the guessing turns into speculation nobody believes.
- Six rounds maximum in one session. Round seven is where the laughter thins out and people start checking Slack, and the activity is remembered as whatever the last round felt like.
- For a team of 25 to 60, split into fixed groups of ten to twelve for the guessing and reveal in the main room. Guessing between sixty strangers is not a game, it is a lottery.
- Do not run all the facts in one go if you have more than fifteen. Two facts a week at the start of the standing meeting builds far more recall than one long session.
- Budget ten seconds of dead air after each reveal. That pause is where someone says 'wait, you did what?' and the actual connection happens.
Scoring, or not
You can run this with points - a mark for each correct guess, a leaderboard at the end - and for some teams the competition sharpens attention. For most, it changes the activity into a quiz and quietly shifts the reward from curiosity to being right, which is not what you came for.
A middle option works better than either extreme: track guesses so people see the split on screen, name who guessed correctly, but keep no running total across rounds. You get the fun of the reveal without turning the newest person on the team into the person who came last at knowing their colleagues.
If you do keep score, score the submitter rather than the guessers. A point for the fact that fooled the most people rewards an interesting submission and takes the pressure off people who genuinely have not met everyone yet.
What it gives a distributed team
Individual attention
Guessing forces each person to consider every colleague one at a time - the opposite of the group video call, where most people register as a name in a grid.
A safe first contribution
New starters submit the same as everyone else and are guessed about like everyone else, which is a much easier first appearance than 'tell us about yourself'.
Follow-up material
The best facts turn into two or three real conversations in the following week, which is the actual mechanism by which a distributed team gets closer.
A reason to be on camera
The reveal is a moment worth watching, so people turn cameras on for it without being asked to.
The safety boundaries
This activity asks people to reveal something personal to colleagues, and that is only safe under conditions the facilitator has to create deliberately. Psychological safety is not produced by the game; the game either sits on top of it or exposes its absence.
Three boundaries do most of the work. Facts are voluntary and non-participation is invisible. You screen every submission before it goes on screen. And the reveal is the end of the obligation - if someone says 'that was me' and nothing else, the round is over and you move on cheerfully.
Screen out anything that identifies someone before the guess (a fact about a person's specific role or a project only they worked on), anything that touches health, family difficulty, money or immigration status, and anything that reads as a jab at someone else. If a submission is borderline, message the person privately and ask if they would rather swap it. Nobody has ever been offended by that message.
Failure modes and how to recover
Nobody has any idea and the guessing goes silent
Normal in a team that has just merged or grown fast. Do not let the silence run. Offer a narrowing clue you prepared in advance - 'this person is not based in the UK' - and let the guesses restart. Two clues per round is fine; the guessing is the mechanism, not the accuracy.
Everyone guesses the same person instantly
The fact was identifying, which is a screening miss rather than a player error. Give the round to the room, laugh at it, and move on quickly. Keep a spare fact of your own in reserve to fill the gap.
One person is guessed for every single fact
Usually the most visible person in the team, and it gets uncomfortable by round four. Break it by removing their fact from the running order and telling the room plainly that the next three facts are not theirs.
A reveal lands awkwardly
Somebody shares more than the room expected, or a joke does not carry across cultures. Do not analyse it. Say one warm, neutral sentence, thank them, and go straight into the next round at the same pace. Speed is the recovery. Afterwards, message the person privately.
Half the team is in another time zone and half are not on the call
Run the guessing asynchronously over a working day: one fact posted per hour, guesses in thread, reveal at the end. It loses the live reaction and gains everybody. Do not run a live version that systematically excludes the same region every time.
Setting it up in practice
You need three things: a way to collect facts privately, a way to show one fact at a time without the submitter's name, and a way for people to guess visibly. A survey plus a slide deck does the job if you do the copy-paste work yourself. In Session Flo it is a single activity - the anonymous collection feeds the rounds directly, and guesses come in from phones while the fact is on the shared screen.
Whatever you use, do one dry run of the reveal mechanic before the session. The failure that embarrasses facilitators is not a bad fact, it is accidentally showing a name column on screen. Check that once, in advance, with real submissions in the system.
And put the facts in the recap afterwards, with names attached now that they are public. A distributed team that can scroll back through two quarters of these has something no all-hands slide has ever given it: a reason to remember each other.
Submit one yourself
Facilitators who play are guessed about too, and it changes the temperature of the room within one round.
Keep a reserve fact
Have two spare facts ready for rounds that collapse in ten seconds. It saves the pacing every time.
Never explain the joke
If a fact needs a paragraph of context to be funny, cut it in screening. Guessable facts are short.
Stop while it is still good
End one round earlier than you want to. That is what people remember, and it is why they turn up next time.
Frequently asked questions
How long does a Guess Who activity take?
Eight to twelve minutes live for six rounds with a team of ten to fifteen, plus about five minutes of your own time the day before to screen the submissions. Collection happens asynchronously and costs each participant roughly a minute. If you only have five minutes, run three rounds rather than speeding all six up.
What if people submit boring facts?
That is a prompt problem, not a people problem. Replace the open 'fun fact' request with a narrow question such as 'what was your first paid job?' or 'a skill you have that is not on your CV'. Constraints make it easier to answer well, and they stop everyone converging on pets and marathons.
Does it work for a team where most people have never met?
Yes, but change what success looks like. With strangers, the guessing will be near-random, so lean on the reveal rather than the accuracy: prepare a narrowing clue per round and give each person ten seconds after their reveal to add one line. In that setting it functions as a structured introduction with a game attached, which is far easier to sit through than a round of introductions.
Should the facts be anonymous during collection?
They should be anonymous to the group and visible to you. You need to know who submitted what in order to screen sensibly, to avoid revealing someone twice in a row, and to run the reveal. Tell people that up front. 'Only I will see who wrote what before the reveal' is a promise people accept, as long as it is true.
How is this different from Two Truths and a Lie?
Two Truths puts one person on the spot at a time and asks the group to judge their honesty. Guess Who spreads exposure across everyone, requires no performance, and asks the group to attribute rather than to detect. For distributed teams and for people who dislike being centre of attention, Guess Who is the gentler and usually the better option.
How often should you run it with the same team?
Quarterly for a stable team, or whenever new joiners make up a meaningful share of the group. Monthly is too often: the surprising facts run out and the activity degrades into a memory test. If you want something for the weekly meeting, use a rotating one-question check-in instead.
Keep reading
- how to run a virtual icebreaker that is not cringe — Broader principles for remote openers that people do not dread.
- the difference between icebreakers and check-ins — Choosing which one your meeting actually needs.
- onboarding new starters with interactive sessions — Where Guess Who fits into a joiner's first fortnight.
- running sessions across time zones — Adapting the format when the team cannot all be live at once.
- Session Flo plans for recurring team sessions — What it costs to run this monthly with a distributed team.