Better groups start with a clear plan.

Random assignment is the easy part. A few choices before you generate teams can make the activity work better for everyone.

Start with the activity, then pick the size

For a quick partner discussion, pairs give each person room to speak. For a problem with several roles, a larger team can divide the work. There is no single ideal size: match the group to the time available, materials, and participation you want.

Count groups when space is fixed

If your workshop has six tables, choose “Number of groups” and enter six. With 25 participants, the result is one group of five and five groups of four. The extra person is allocated during the shuffled deal; you do not need a separate leftovers group.

Set a maximum when participation matters

If you want at most four people in a discussion, choose “Maximum group size.” With 18 participants, the tool creates five balanced groups: three groups of four and two of three. Maximum size is a ceiling, not a promise that every group will have that many people.

Handle names carefully

Enter one person per line. Empty lines are ignored; repeated names are not removed, because two people can share a name. “Sam” entered twice becomes “Sam (1)” and “Sam (2)” in the results. Numbering skips any label already used by another person, so a name entered as “Sam (1)” stays distinguishable. Initials or recognizable nicknames are often enough for an informal activity. Check your organization’s rules before entering a roster on a shared device.

Try a fresh mix for the next round

Keep the previous result visible, check “Try to reduce repeat pairings,” and generate again. The tool compares up to 40 candidate shuffles with that one previous result. It selects the candidate with the fewest repeated pairs, stopping early if it finds zero. This changes the selection toward fewer repeats; it is not a uniformly random choice among all possible team arrangements.

Some repeats are unavoidable. Six people split into two groups of three cannot all meet entirely new teammates in the next round. Use the reported repeated-pair count as a practical signal, not a guarantee of a completely new mix.

Give each group a way to begin

1. State the task and how long they have.
2. Make the expected output specific: one idea, a short demonstration, or a shared answer.
3. Choose how groups will report back. The one-person picker can select someone from the whole roster, but it does not pick a representative from each group.
4. Review assignments for practical needs before asking people to move.

Know when random assignment is the wrong fit

Random teams do not account for skill, prior experience, accessibility, or interpersonal circumstances. Use a considered assignment process when those factors matter. This tool is designed for low-stakes activities, not employment decisions, prize compliance, or other consequential selections.