Picking a group-size range for the activity, not the headcount

Open a group-size tool and the first instinct is to do the math: twenty-two people, want groups of four, that's five groups with two left over, so set min and max to four and hope the leftover sorts itself out. That's backwards. A size range isn't a division problem — it's a description of what the activity needs to work. A discussion where quiet people should still get airtime needs a different range than a hackathon team that needs enough hands to actually ship something. Get the range right for the activity and the headcount math takes care of itself; get it wrong and even a perfectly "fair" random split produces groups that don't function.

Start from the failure mode, not the average

Every activity has a size where it quietly breaks, and that's the number your range should protect against — not the size you expect on an average day. A book club discussion breaks past six or seven people, because past that point at least one person stops talking for the whole hour. A hackathon team breaks below three, because two people can't cover design, backend, and a demo in a weekend. A murder-mystery party breaks below four, because some roles need someone to accuse and someone to be accused. Decide what breaks your activity first, then set the min and max on the variable-size group splitter to keep every possible group on the safe side of that line.

This is different from picking a fixed team count and dividing. A fixed count optimizes for even groups; a range optimizes for groups that all work, even if that means one group of three and another of five out of the same twenty-two people. The splitter doesn't know or care what your activity is — it just keeps every group inside whatever window you give it — so the actual design work is picking that window correctly, once, before you paste in a single name.

Ranges for activities people actually run

A few concrete starting points, tuned to what tends to break each format:

Book club or discussion circle — 2 to 4. Small enough that nobody can hide for the full hour, large enough that a pair isn't forced to carry an entire conversation alone if one person runs quiet that week.

Hackathon or short project team — 3 to 5. Below three and a weekend project is missing a role; above five and half the team is idle while two people do the actual work. This is also the range where a fixed team divider starts to look tempting instead — more on that below.

Classroom activity stations — 2 to 3. Small enough that a rotation through four or five stations keeps every student doing something rather than watching, and small enough that you can visually confirm every station has someone the moment groups are announced.

Party or icebreaker mixer — 4 to 6. Big enough that a shy guest can coast for a minute without the group stalling, small enough that everyone still gets a turn talking within a reasonable few minutes.

Escape room or puzzle-room team — 4 to 6. Most commercial rooms are built and priced around this exact window, so let the venue's own capacity set your ceiling rather than guessing.

Why a wider range isn't automatically safer

It's tempting to set a generous range — say, two to six — on the theory that more flexibility can only help. In practice a wide range just means you've stopped deciding anything: a group of two and a group of six drawn from the same activity almost never function the same way, so you've traded a hard decision for an unpredictable one. A tight, deliberate range like three to five does more work for you than a loose one, because every group the tool produces is close enough to identical in character that you don't have to improvise instructions for the outliers.

When you actually want a fixed count instead

If your activity has a hard structural requirement — exactly four teams for a bracket, exactly three judges per panel — a size range is the wrong lever entirely. Ranges exist for activities that genuinely tolerate variation; once the activity needs an exact number of equally-sized groups, reach for the random team / group divider and set the team count directly instead of trying to force a range to land on it.

Reusing a range across sessions

Once you've settled on a range for a recurring activity — a weekly seminar, a standing hackathon night — it's worth writing it down rather than re-deriving it each time. But don't reuse the same random split; reroll it every session so the same three people don't keep landing together by coincidence. If avoiding repeat groupings matters more than the range itself, layer in a no-repeat group mixer that remembers past pairings, or — for splitting a video call specifically — a breakout room assigner built for that exact workflow.

Set the range once, then stop thinking about it

The whole point of choosing a deliberate range up front is that you only have to think about it once. Decide what breaks your activity, set the variable-size group splitter to protect against it, and every future run — this week, next week, with a slightly different headcount — comes out the right shape without you doing the math again.