"Split into groups of four" is a sentence every teacher says and every teacher dreads. Self-selection leaves students out; teacher-assigned groups invite favoritism accusations; and random draws — the honest middle ground — usually fail because basic generators ignore constraints and history. Here is the method that actually works, step by step.
01 Count first, decide group size second
A class of 27 wants groups of 4: that is six groups of 4 and one of 3 — accept the odd group, because forcing equal size creates an eighth group of 3 anyway. The generator handles the arithmetic: group sizes always differ by at most one, and it tells you the split before you commit.
02 Encode the constraints before you draw
Spend five minutes at term start writing down two lists: keep-apart pairs (combinations that derail) and keep-together pairs (combinations that must stay). Enter them once in the group generator. From then on, every weekly reshuffle respects them as hard rules — the tool refuses to draw an invalid split rather than quietly breaking your constraint.
03 Rotate weekly, remember history
Fresh groupings every week keep energy high, but pure randomness repeats pairings by chance. The history-aware engine penalizes last week's exact groupings, so "we were just together" becomes rare — and when it happens, it is provably chance, with a seed to show it.
04 Show the draw, keep the receipt
Project the draw live. The seed on screen means any student can verify the exact split at home. When the complaint comes — and with adolescents it will — the answer is the receipt, not a defense of your intentions. Verification ends the conversation; justification prolongs it.
Term start: paste your roster, set group count, enter keep-apart/keep-together pairs, draw once, save the share link in your planning doc. Every subsequent Monday: open the link, redraw, done.