Random group generator
Press Draw to see the result
Paste a list of names and pick what to do with it across the top of the panel. Groups splits the list by size or by number of groups, dealing one name at a time so eleven people into groups of four come out four, four and three, never four, four, two and one. Pairs puts people two at a time. Seating fills rows of a fixed width. Bracket draws a single-elimination first round and hands out the byes. Tasks deals a second list — jobs, chores, shifts — out over the names. Every one of them takes the same optional seed, so any draw can be reproduced.
How to make random groups
For classroom work, fixing the group size is almost always the right mode: the activity is designed for groups of three or four, and the class size changes week to week. Fixing the number of groups suits a room with a set number of tables instead. Random grouping is also doing something specific pedagogically. It breaks up the clusters that form when students pick their own, which is usually the whole reason for grouping them.
The first thing that follows from dealing the names out one at a time instead of slicing the list is that the requested size is a ceiling and not a target: the leftovers are spread across the existing groups and never form a short one at the end, so eleven into groups of four is 4/4/3 and never 4/4/2/1.
The four modes beside it are genuinely different deals rather than the same one relabelled, which is worth knowing because the obvious substitutions each get one detail wrong. Pairs takes people two at a time and folds an odd last person into the pair before, so eleven people give four pairs and one three — where groups of two would leave the eleventh on their own. Seating fills rows of the width you set and keeps the empty seats together at the end of the last row, so the plan reads like a room instead of like a list; groups would scatter the gaps. Bracket pads the field up to the next power of two and hands out the byes, one per match, so eleven entrants make a sixteen-slot draw with five byes over four rounds. The byes go to whoever the shuffle put first, because this is a draw and not a seeding: if the byes should go to the top seeds, assign them yourself. Tasks takes two lists, the people and the jobs, and deals the jobs over the people; six chores among four flatmates comes out 2/2/1/1 and never 3/1/1/1.
What the deal cannot do is remember anything. It has no idea who was paired with whom last week, so avoiding a repeat means redrawing or fixing it by hand, and a rotation that genuinely never repeats needs a round-robin schedule, which is a different kind of thing. It has no opinion either about who should sit at the front or who works badly with whom. Draw again if an arrangement looks wrong; nothing here will argue with you.
What people use it for
- Splitting a class into project groups of four
- Pairing people up for partner work without leaving anyone unpartnered
- Laying out a seating plan in rows of a fixed width
- Filling tables when a room seats a fixed number per table
- Drawing a knockout first round with the byes handed out
- Dealing a chore rota out over a household
- Assigning shifts or presentation slots to a team
- Making a draw the losing side can check afterwards from the seed
Questions
Groups, split by size. The activity dictates the group size, and the class size changes week to week.
They are spread across the existing groups instead of being left in an undersized one, so sizes differ by at most one.
Yes: the Pairs mode. An odd number folds the last person into the pair before, so you get a three rather than somebody on their own. Setting the group size to two under Groups does the opposite and leaves them alone.
Yes, as often as you like. Each draw is independent of the last.
No. Every possible arrangement is equally likely.
Yes. Type a seed before drawing: the same seed and the same list always give the same groups, which is how a draw gets checked.
Not automatically. A shuffle has no memory. For a rotation that never repeats, you want a round-robin schedule rather than a draw.
Not automatically. Redraw, or move one of them afterwards.
Yes: the Seating mode lays the names out in rows of the width you choose and keeps the empty seats together at the end of the last row. For tables rather than rows, use Groups with one group per table.
Yes: the Tasks mode. It takes two lists — the people and the jobs — and deals the jobs out over the people, one share each.
The counts differ by at most one. Six jobs among four people is 2/2/1/1, never 3/1/1/1.
Yes: the Bracket mode pads the field up to the next power of two, hands out the byes one per match, and prints how many rounds the draw takes.
A free pass to the second round, given when the field is not a power of two. Eleven entrants in a sixteen-slot bracket means five byes. They go to whoever the shuffle put first, since this is a draw and not a seeding.
Not until they are played, since later rounds depend on who wins. Draw the first round, play it, then pair the winners.
No. Copy the result before you close the tab; nothing is stored on the site.