How to run a brainstorming session that starts in silence

To run a brainstorming session, write one question at the top of the board and give everyone ten minutes to write ideas alone, in silence. Then read the notes out, build on them aloud, and vote with dots. Keep criticism out until the vote.

The silent stage is the part that the classic rules leave out. The experiments below point towards it, though none of them tested writing first and talking afterwards, so that order is our inference.

What the experiments counted

The rules come from Alex Osborn's book Applied Imagination. We haven't read it, so this account comes from a 1987 paper by Michael Diehl and Wolfgang Stroebe, which we have. It lists four rules: more ideas are better, wilder ideas are better, improve or combine ideas, and don't criticise. It also quotes Osborn's claim that "the average person can think up twice as many ideas when working with a group than when working alone".

That claim didn't hold when it was tested. The paper describes a 1958 study by Taylor, Berry and Block, with people working either in groups of four or alone. The researchers then pooled the solo ideas in fours and counted each idea once. Those pooled sets had "nearly twice as many different ideas as the real groups". Diehl and Stroebe list 22 experiments of this kind. In 18, the pooled solo workers did better. The other four, all with two-person groups, found no difference.

Their own experiments then looked for the cause. They tested free riding, fear of being judged and production blocking, and concluded that "production blocking accounted for most of the productivity loss". Blocking comes from "the rule that only one group member speaks at a time", so everyone else has to wait.

Two limits apply. These are laboratory counts of ideas, and the paper calls the results on quality "more equivocal". They also don't say a group is useless. In our view, a group earns its place after the writing, when people combine notes, ask what an idea means and agree who tries what. So our agenda keeps both stages. Diehl and Stroebe didn't test this sequence, and putting the writing first is our inference from their results.

A 45-minute agenda

The third stage is where IDEO.org's brainstorm rules help, especially "Defer judgement", "Build on the ideas of others" and "One conversation at a time". Its brainstorm method has people describe each idea as they stick it up. We've moved that description to after the writing, which is our change and not IDEO.org's.

One question, eight notes

The team below is invented, and so are its votes and next steps. It's a group of five building a made-up whiteboard app. This isn't Prompt Board's data, and nobody tested these ideas. The question and the notes come from our brainstorming template.

The question. "How might we get new users to their first board in under a minute?" It names an outcome you could time and leaves the answer open. Our guide to how might we statements has the checks for that.

The silent ten minutes. Everyone writes one idea per note. Sensible notes go in All ideas and odd ones go in Wild ideas, without discussion. The sample board shows eight notes to stay readable.

Reading out and building. Each author reads their notes, and the others may only add to them. Someone looks at "Import a photo of a whiteboard" and writes a tamer version, "Start from a photo of your notes". Someone else looks at "Invite a teammate at sign-up" and points out that it adds a step before the first board. That's criticism, so it waits for the vote. The note stays up.

The dots. Five people with two dots each make ten dots.

The top two move to Shortlist. Those dots record what five colleagues prefer. They say nothing yet about new users.

The next steps. One person mocks up the prompt box, and another mocks up the template picker. Both then time new users against the minute in the question. If neither gets there, the team comes back to the board and tries the notes with one dot.

When a session goes wrong

The notes dry up after three minutes. Leave the silence alone for a while. Then ask for the opposite of a note on the board, or for the idea that would get someone told off.

One voice takes over the reading out. Go round in a fixed order, and have the most senior person read last.

Too many notes to read. Group the duplicates first. If the groups need names, that's an affinity map and a job of its own.

Off-topic ideas keep arriving. Put them on an idea parking lot and carry on. IDEO.org's rule here is "Stay focused on the topic".

With a class

A class usually brainstorms to open up a discussion, so it can skip the vote. Our class brainstorm template puts one open question at the top, such as "What makes a good friend?". Students write for three minutes, group the notes and talk about the groups. The order stays the same as above, with writing before talking, so students who rarely speak still have a note on the board.

Frequently asked questions

How long should a brainstorming session be?

Our agenda takes 45 minutes, of which ten are silent writing. That's our own plan and no study set it. Shorten the build stage first if you have less time, and keep the writing.

How many people should join?

We don't have a tested number to give you. In the experiments Diehl and Stroebe list, the only groups that showed no loss had two people. Our example has five because five people can read everything out in 15 minutes.

Is it still brainstorming if nobody talks at first?

Yes, as far as Osborn's four rules go, as the paper lists them. None of them says that ideas must be spoken. Writing first only changes when the talking happens.

Run your own

Write the question before the meeting and send it round, then open the brainstorming template and clear the sample notes. Start the timer before anybody has a chance to speak. When you have a shortlist, put a name and a test against each note before people leave.