Snowball (1-2-4-All)

A structured brainstorm in which the same question is worked through four times, each time in a larger group. Everyone first writes their own ideas in silence, then shares in a pair and together picks the strongest, then in a four merges two selections into five to ten key ideas, and in plenary all ideas are clustered and every cluster is given a name. Optionally the group closes with a priority round. Internationally the structure is known as 1-2-4-All.

With the default settings one question takes about twenty minutes: 2 minutes alone, 3 in pairs, 4 in fours, 5 for clustering and 3 for the optional prioritising, plus a couple of minutes for regrouping and sticky notes. In a class of thirty that grows to around thirty-five, because clustering scales with the number of fours. Whoever prepares the activity sets the time per step; if you enter several questions, the time counts per question.

Eight to roughly thirty-two. Below eight you end up with one or two fours and step 3 loses its function — take the placemat instead. Above thirty-two the plenary clustering becomes too slow: split it into two clustering rounds, or have the fours hang up their sheet rather than read it out.

Think-pair-share stops after the pair: think, share, done. Snowball adds the two steps that make the difference — the four, where a real choice has to be made for the first time, and the shared clustering, which turns loose ideas into an ordered overview. If you only want to get everyone thinking for a minute, Think, Write & Share is quicker; if you want to end with a usable harvest, Snowball is the right form.

Paper or a notebook for step 1, sticky notes and markers from step 2 onwards, and a large sheet or a piece of the board to cluster on. For prioritising: three sticker dots per participant, or simply three marker strokes. In LudensFlow you also get the step cards with the instruction for each step, a timer per phase and the printable participant sheet — print one per participant and everybody goes home with their own ideas and the group's cluster map.

One question, phrased openly, and concrete enough that something worthwhile can be written down about it in two minutes. "What do you think about sustainability?" produces air; "what could we do differently in this school next month?" produces ideas. Avoid two questions in one sentence — the class will split anyway and clustering becomes impossible. If you are stuck, ask Microsoft Copilot: "Write 3 open, provocative central questions about [topic] for a brainstorm with students aged [age]. One question per line, no numbering." Every line you keep becomes its own round in the preparation.

You guard the structure, not the content. Formulate one open question beforehand and keep it visible on screen. Decide how the groups are formed — voluntarily, at random, or deliberately mixed on a criterion you name neutrally — so regrouping does not turn into a negotiation. During step 1 you walk around and say nothing. At steps 2 and 3 you only repeat the ceiling and the one-idea-per-note rule. While clustering, ask "where does this one go?" and do not decide yourself; always leave the name of a cluster to the group. If a cluster you did not expect wins the priority round, do not smooth it over — that is usually the most valuable moment of the lesson.

Yes. Put each question on its own line in the preparation: every line becomes a separate round with all the steps, and the total is recalculated for you. Two questions in one lesson work well when they light up two sides of the same topic — for example "what helps our class?" and "what gets in the way?". More than two rarely fits into one period.

Yes, with one adjustment: give step 1 explicitly as a writing task with microphones off, otherwise the silence evaporates. Pairs and fours go into breakout rooms, and you cluster together by dragging items on a shared whiteboard. Online, allow a little more time for switching between groups than you would in a classroom.

The structure is described among the Liberating Structures of Henri Lipmanowicz and Keith McCandless, who named it 1-2-4-All. It builds on think-pair-share, formulated by Frank Lyman in 1981, and on nominal-group research that showed as early as the 1970s that people who write separately first produce more ideas together than a group that starts brainstorming out loud.