How to make balanced teams

"Balanced" means different things to different people. A PE teacher wants teams that can actually compete, a workshop host wants each table to mix departments, and a parent at a birthday party just wants nobody picked last. A random split handles some of this well and some of it not at all. This guide shows which is which, with the numbers. It also covers a tiered method that gets you close to skill-balanced teams using the random team generator.

Two kinds of balance

Size balance means teams are the same size, or as close as possible. That's arithmetic, and a generator can guarantee it.

Skill balance means teams are roughly equal in ability, experience or whatever matters for the activity. No random tool knows who is good at football. The ExesTools team generator doesn't rank people or keep anyone apart by rule. It shuffles the whole list and deals names into teams like cards. Skill balance comes from how you set up the list, not from the tool.

Size balance with odd numbers

The generator keeps team sizes within one person of each other. You choose either Number of teams (2 to 50) or People per team (1 to 100), and it works out the rest. It takes up to 200 names. When the count doesn't divide evenly, extras go to the earlier teams in deal order (round-robin after the shuffle), so 11 names across 3 teams gives 4, 4, 3.

With Pairs and an odd roster, one pair becomes a trio so nobody is left out. If you need strict pairs, say for a doubles match, switch on Leave one person out. The extra person sits out the round. To stop the same person sitting out every time, pick them with the random name picker and keep a note.

How often pure random goes lopsided

Random teams avoid the public ranking of captain picks, but chance can still stack a team. Here are the odds for 24 players in 4 teams of 6:

• Two particular friends end up together 5 times in 23, about 22%. Once your friend is placed, 5 of the 23 remaining places are on their team.
• Your four strongest players all land on different teams only about 12% of the time: (18 × 12 × 6) ÷ (23 × 22 × 21).
• Three or more of those four end up on one team about 14% of the time. I checked this with a 200,000-deal simulation.

So in roughly one deal in seven, one team gets most of your strongest players. That isn't a bug. It's what random means, and it's why the tiered method below exists.

When a split looks lopsided, Reshuffle gives a completely new deal from the same list and settings. Agree on the rule before the first deal: either you accept it, or you allow one reshuffle. Reshuffling until you get the teams you wanted is captain picking with extra steps.

Tiered random teams: the method that works

Use this when skill balance matters, such as a sports lesson or a quiz night with a few obvious experts. You run the generator once per tier.

The rule that makes it work: each tier must hold a whole multiple of the number of teams. For 4 teams, use tiers of 4, 8 or 12. Then every team gets exactly the same number from each tier.

A worked example, with 24 students and 4 teams:

1. Sort privately into two tiers, on paper rather than out loud. Tier A has the 8 most experienced players. Tier B has the other 16.
2. Run tier A. Paste the 8 names, set Number of teams to 4, type Optional team names Red, Blue, Green and Yellow, and press Generate teams. Each team gets exactly 2 tier-A players. Those team names stay filled in with the rest of your settings.
3. Run tier B with the same team names still showing. Each team gets 4.
4. Combine. Red is Red from both runs, and so on. Every team has 6 players: 2 experienced and 4 others.

Within each tier the placement is still random, so nobody can say you built the teams. For comparison, a single random deal of all 24 gives each team exactly 2 of the top 8 only about 7% of the time.

Press Copy teams after each run and paste both results into one document. That makes combining quick.

If the numbers don't divide, put the remainder in the last tier. With 26 students and 4 teams, use tiers of 8 and 18. Tier B then deals 5, 5, 4, 4: the extras go to the earlier teams in deal order (round-robin after the shuffle), so with team names Red, Blue, Green and Yellow, Red and Blue get 5 and Green and Yellow get 4. Two teams finish with 7 and two with 6. If you run a third tier, give its extras to the other teams so the final sizes even out.

Keeping certain people apart

The generator can't keep two people apart by rule. Three practical options:

• Put them in a tier the same size as the number of teams. With 4 teams, a tier of 4 sends one person to each team, so those two can't end up together. A tier of 8 doesn't guarantee it.
• One agreed reshuffle. If the first deal puts them together, reshuffle. Say beforehand that you'll do this.
• Swap by hand, openly. Move one person and swap someone of a similar tier back.

For gift exchanges, where keeping people apart is the whole point, the Secret Santa generator has proper exclusions.

Captains, if you must

Captain picks are quick, but the same people get picked last, in public, every time. If a group insists on captains, make it fairer:

• Pick the captains at random, not by volunteering.
• Use a snake draft: A, B, B, A, A, B. The captain who picks second in one round picks first in the next, which offsets the first-pick advantage.
• Captains pick privately on paper, and the teacher reads out the teams. Nobody hears their name called last.

Classroom groups

Paste your class list once. The roster, settings and optional team names stay in this browser, so they're there next lesson. The last generated split is not saved — press Generate teams again (or Reshuffle) when you need the cards back, or Copy teams / Print if you want a record.

• Groups of 3 suit short tasks where everyone needs to talk. Groups of 4 suit projects with distinct roles, such as researcher, writer, presenter and checker.
• Print makes a paper handout, and Copy share link sends the same names and teams to a co-teacher.
• Fix typos first. "Maria" and "Mariah" can split one student into two lines. If the same name appears twice, the tool shows a notice so you can add an initial.
• Reshuffle doesn't give new partners. The tool has no memory of earlier rounds. Over a term, keep a note of past groups and swap by hand if a pair repeats.
• One name at a time? Use the classroom spinner instead of making teams of one.

Events and breakout rooms

For breakout rooms, set People per team to the room size and paste the cards into the meeting chat. For a quiz night, use theme team names and send the share link so phones in the room can open the same groups. For a quiz with obvious experts, use the tiered method with tier A as the experts.

What I checked for this guide

The table matches the rule and examples on the live team generator page (checked 2026-10-13). The 22% and 12% figures are exact fractions. The 14% and 7% figures come from simulating 200,000 random deals of 24 players into 4 teams of 6.

Frequently asked questions

How do I make random teams with an odd number of people?

Choose the number of teams, and the generator keeps sizes within one person of each other. For example, 11 people in 3 teams gives 4, 4 and 3.

Can the team generator balance teams by skill?

Not by itself. It deals names at random. To mix skill levels, sort people into tiers whose sizes are multiples of the number of teams, run the generator once per tier with the same team names, and combine.

What happens to the extra person in pairs?

They join one pair to make a group of three. Switch on Leave one person out if you need strict pairs.

Will reshuffling give everyone new partners?

Each reshuffle is a fresh random deal, but the tool doesn't remember earlier rounds, so the same people can land together again.

How many people can I split at once?

Up to 200 names, into up to 50 teams or up to 100 people per team.