How to Split People Into Fair Teams in Seconds (Without the Schoolyard Draft)
By the Super Simple Digital Tools Team · Updated June 2026
The classic way to pick teams, two captains taking turns choosing names, is slow, public, and quietly cruel to whoever gets picked last. A team randomizer fixes the social problem by making the split impersonal: a computer deals the names, so there is no popularity contest and no one to blame. The trade-off is that randomness optimizes for fairness of process, not for evenly matched sides, so the trick is knowing when raw randomness is exactly what you want and when to add a light touch of structure on top.
Start by deciding what you are constraining. If you have a fixed number of stations, courts, or breakout rooms, set the number of teams and let the tool figure out how many people land in each. If instead the activity demands specific group sizes, say pairs for an interview exercise or fours for a card game, set the team size and let the number of groups fall out of your headcount. Getting this choice right first saves you from regenerating repeatedly because the groups came out the wrong shape.
When the roster will not divide cleanly, the tool spreads the leftovers so teams never differ by more than one member. Twenty-three people into four groups becomes three teams of six and one of five. This even-as-possible distribution matters more than it sounds: a single oversized group can mean one extra voice in a discussion or one extra player on a field, and keeping the gap to one person keeps things feeling fair without manual fiddling.
The fairness of the draw itself comes from the Fisher-Yates shuffle, which guarantees every ordering is equally probable and runs in a single fast pass through the list. That mathematical neutrality is the whole point: because nobody, including you, can predict or steer the outcome, the result is defensible. If a parent, student, or colleague questions a split, you can honestly say the assignment was pure chance with no thumb on the scale, which is far easier than defending hand-picked groups.
Random does not mean perfectly matched, though. For a casual icebreaker that is fine, but for a competitive game one team may happen to collect all the strongest players. The fix is simple and keeps most of the fairness: generate the teams, then run a short trade window where each team swaps one player, or sort players into rough skill bands first and randomize within each band before dealing them out. You get the impartiality of a random draw and the balance of a thoughtful one.
- Paste names straight from a spreadsheet column or attendance list, one per line, instead of typing them, to avoid typos and missed names.
- Regenerate a few times if the first split clusters friends or all the strong players together, the shuffle is independent each run, so a new draw costs nothing.
- For repeat sessions, deliberately re-randomize each time so students or staff work with different people instead of the same clique every week.
- Need skill balance? Split your list into tiers (strong, average, new), randomize each tier on its own, then assign one name from each tier to every team.