Corporate life · Fair comparisons
Does Brainstorming Work? Groups Against the Same People Working Alone
The short version: brainstorming in a group produces fewer ideas than the same number of people working alone and then combining their lists. In a classic 1987 study, four people working separately came up with about twice as many ideas as four people brainstorming together, and later reviews confirmed the loss, larger in bigger groups. The main reason is simple: only one person can talk at a time, so ideas are forgotten or dropped while waiting. A workshop that counts its ideas without that comparison measures the group against nothing.
Kevin's department spent a morning in a workshop room with beanbags, sticky notes and a facilitator. Rule one: no idea is a bad idea. By lunch the wall had 40 ideas on it. The facilitator called it a record.
Kevin's boss was delighted: "Imagine how few we'd have had working alone." Kevin had imagined exactly that. He also remembered that he'd had three ideas during the session and said one, because Darren was talking about a loyalty app for 20 minutes.
The claim
Brainstorming was popularised in the 1950s by an advertising executive who wrote that people produce twice as many ideas in a group as alone. The rules are familiar: say everything, judge nothing, build on each other. It feels productive, because the wall fills up fast.
But "40 ideas" is a number with nothing to compare it to. The fair question isn't "did the group produce ideas?" but "did the group produce more than the same people would have produced on their own?"
The fair comparison
Psychologists tested this by building a control group of the right kind. Take four people and put them in a room together to brainstorm. Take four other people, put each in a separate room with the same task and the same time, then add up their ideas, counting duplicates once. The second set is called a nominal group: a group only on paper.
| Four people... | Ideas produced |
|---|---|
| Brainstorming together | baseline |
| Working alone, lists combined | about twice as many |
That was the result of a 1987 study, and a review of the research a few years later found the same loss across many experiments, bigger for bigger groups. The quality of ideas didn't make up for it: the alone-then-combined lists contained at least as many good ideas.
What slows the group down
- One speaker at a time. While you wait for your turn, you forget the idea, or decide it's no longer relevant. The researchers called this production blocking, and found it was the main cause.
- Worry about judgement. "No idea is a bad idea" doesn't stop people noticing who's in the room, including the boss.
- Coasting. In a group, nobody can tell who contributed what, so some people contribute less.
Statistics for noobs: compared with what?
A result only means something next to the right comparison. "40 ideas" needs "and the same people alone would have had...". "Sales rose 10% after training" needs "and the untrained team's sales rose...". The comparison has to match everything except the thing you're testing: same people, same time, same task.
Where it costs you money
"Our investment committee picked these stocks together" sounds safer than one person's picks. Groups can also converge on the loudest view. A fair test compares the committee's picks with independent picks made separately and combined, and with a simple index.
The moral
Kevin suggested trying it the other way: ten minutes writing ideas alone, then pooling them, then discussing. The next session produced 71 ideas and lasted half as long. The facilitator called it a record and asked to run it again next month, with beanbags. Darren's loyalty app was on 4 of the 8 individual lists.
What this can't tell you
- Mostly lab studies. Students on short tasks; long-running work teams may behave differently.
- Ideas aren't decisions. Groups can be useful for choosing between ideas and getting people behind them. The finding is about generating ideas.
- Electronic brainstorming differs. When people type ideas at the same time, the one-speaker problem shrinks, and large groups can do better.
The workshop, the beanbags and Darren are an illustration. The findings come from a 1987 experiment in the Journal of Personality and Social Psychology and a 1991 meta-analysis in Basic and Applied Social Psychology. Educational content, not business or financial advice.