Why Most Brainstorms Fail, and the 20-Minute Format That Works

You know the scene. A whiteboard, a marker, someone cheerful saying "no bad ideas!", and forty minutes later a list of things everyone had already thought of, written in the handwriting of whoever talks fastest. Brainstorms fail for predictable reasons, and once you can see the reasons, the fix is almost mechanical.

Three reasons the usual format underperforms

Anchoring. The first idea spoken sets the frame for every idea after it. If it is about pricing, the next ten are about pricing. Production blocking. Only one person can talk at a time, so while you wait your turn you are either rehearsing your idea or forgetting it. Evaluation anxiety. Nobody believes "no bad ideas" and everyone edits before they speak, which removes exactly the strange ideas you came for.

Research on group creativity has pointed at these three for decades. Groups that write first and talk second reliably produce more ideas, and more unusual ones, than groups that talk from the start. Our members have been running a format built on that for about a year. It takes twenty minutes.

The format

• Minute 0 to 2: one sentence of framing. The host states the question in the form "How might we..." and gives one constraint. Example: "How might we get new members to post in their first week, without sending more emails?" No discussion.
• Minute 2 to 8: silent writing. Everyone writes as many ideas as they can, one per line or sticky note, alone and in silence. Quantity is the only target. Set a visible timer.
• Minute 8 to 12: silent reading. Ideas go on the wall or into a shared document. Everyone reads all of them without comment. Duplicates get stacked.
• Minute 12 to 16: build, do not judge. Each person picks someone else's idea and writes one way to make it bigger, cheaper or weirder. Still in writing.
• Minute 16 to 20: three dots. Everyone gets three votes. The host reads out the top five and writes down who is doing what next. That is the end.

Why the silence matters

Writing in parallel removes production blocking: ten people produce ten ideas a minute instead of one. Writing before reading removes anchoring: nobody's first idea can steer anyone else's. And writing removes the performance. A member in one of our product groups put it well: "I have ideas in the shower, not in meetings. This format is the shower with other people in it, which sounds worse than it is."

Doing it online

The format works in a group thread with almost no changes. The host posts the framing. Members reply with their lists within a day instead of six minutes. The host then starts a second thread for builds, and a poll for votes. It is slower, but the thinking is often better because people get to sleep on it.

What to do with the top five

Most of them will not survive the week, and that is fine. The point of a brainstorm is not to pick the winner in the room. It is to leave with options that nobody would have reached alone, plus a name next to each one. The 48-Hour Idea Test exists for the next step. Run your top idea through it before anyone opens a design tool.