Brainstorming doesn't work. Here is what does

Classic group brainstorming, where people call out ideas while someone writes them on a flipchart, reliably produces fewer and safer ideas than the same people would produce working alone. The fix is not to abandon group idea generation but to change its order: think alone and in silence first, then build on each other's ideas, then discuss. Brainwriting, Crazy 8s and a few other structured methods do exactly that.

The belief, stated fairly

Brainstorming was popularised by Alex Osborn, an American advertising executive, most famously in his 1953 book Applied Imagination. His rules are still pinned to workshop walls: defer judgement, go for quantity, welcome wild ideas, build on others' ideas. The promise was that a group following these rules would produce far more ideas than its members working separately.

The belief has real roots. People do spark off one another. A good session feels energetic, and the team leaves feeling it owns the result. For many managers "let's brainstorm" simply means "let's think together", and that instinct is healthy. The problem is the specific format, not the wish to collaborate.

Why it fails in practice

Researchers started testing Osborn's claim in the late 1950s, and the results have been consistent. In 1987 Michael Diehl and Wolfgang Stroebe reviewed the existing studies and ran their own experiments. They found that "nominal groups" (the same number of people working alone, with their ideas pooled afterwards) beat interacting brainstorming groups on the number of ideas. Their experiments pointed to one main cause.

Production blocking. Only one person can speak at a time. While you wait your turn you forget your idea, edit it, or decide it no longer fits where the conversation has gone. The more people in the room, the worse this gets.

Two other forces make it worse in any real organisation:

• Evaluation apprehension. "Defer judgement" is a rule on the wall, not a feeling in the room. People still watch their manager's face. In hierarchical teams, junior staff often say nothing until the senior person has spoken, and then they agree.
• Social loafing and matching. In a group, some people coast. Others unconsciously match the pace of the slowest contributor. The session settles at a comfortable, shared level of effort.

Add a common fourth problem that the laboratory studies rarely capture: anchoring. The first idea spoken aloud sets the frame. If the first suggestion is "an app", the next ten ideas tend to be apps.

The result is a familiar picture. Forty minutes, a full flipchart, and a list dominated by the two or three most confident voices, with ideas that were mostly obvious before the session started.

What works instead

The common thread in every alternative below is simple: separate generating from discussing, and protect the first minutes of thinking from the group.

1. Brainwriting. Each person writes ideas silently, then passes the sheet on, and the next person builds on what they see. The best-known version is 6-3-5, published by Bernd Rohrbach in 1968: six people, three ideas each, five minutes per round. You keep the "building on each other" benefit Osborn wanted, without anyone waiting to speak. See the brainwriting guide.
2. Silent individual writing, then round-robin sharing. This is the core of the Nominal Group Technique developed by Andre Delbecq and Andrew Van de Ven. Ten minutes alone, then each person shares one idea at a time in turn, with no debate until every idea is on the wall.
3. Crazy 8s. Fold a sheet into eight panels, sketch eight variations in eight minutes. Popularised through the Google Ventures Design Sprint, it forces people past their first idea, which is usually everyone's first idea. See Crazy 8s.
4. Worst possible idea. Ask for terrible solutions on purpose, then flip them. It lowers evaluation apprehension because nobody can be embarrassed by a bad idea when bad ideas are the brief. See worst possible idea.
5. Prompted generation. Give people a lens rather than a blank page: SCAMPER, analogies from another industry, or a set of "how might we" questions. A blank page invites the obvious; a constraint invites the specific.

Then, and only then, cluster the ideas with affinity grouping and talk about them. Discussion is valuable. It is just in the wrong place in a classic brainstorm.

Do's and don'ts

Do

• Start every idea round with at least five minutes of silent, individual writing.
• Sharpen the question before you generate anything. A vague prompt produces vague ideas, whatever the format. Our piece on writing a problem statement covers this.
• Ask the most senior person to share last, or collect ideas anonymously.
• Keep generation and evaluation in separate blocks, ideally with a break between them.
• Count ideas from everyone, not just from the people who spoke.

Don't

• Don't open with "any ideas?" to the whole room. The first answer becomes the anchor.
• Don't let the facilitator write only the ideas they understand. Ask the author to write their own.
• Don't run a 40-minute open session. Short, timed rounds beat long open ones.
• Don't judge the session by how lively it felt. Judge it by how many different directions ended up on the wall.
• Don't vote straight after generating. Let ideas be grouped and clarified first, and choose your voting method carefully (see why dot voting lies to you).

Illustrative example

Illustrative. A regional transport authority in northern Europe wanted ideas to reduce missed medical appointments among older passengers who rely on its demand-response buses. Its first session was a classic brainstorm with twelve staff. The operations director spoke first ("better reminder texts"), and the hour produced roughly twenty ideas, most of them versions of reminders.

The second session used the same people and the same question, rewritten as "How might we make it easy for passengers to get to an appointment when their plans change on the day?" Each person spent seven minutes writing alone, then the group ran three rounds of brainwriting. The director shared last.

The wall now held ideas from twelve people instead of four. New directions appeared that nobody had voiced the first time: drivers who could rebook a passenger's return trip from the vehicle, a shared booking line with two local clinics, a printed weekly card for passengers who do not use phones. After affinity grouping, the team chose two to prototype. Nothing about the people changed. Only the order of thinking did.

Questions people ask

Is brainstorming really ineffective, or is that a myth too? The research is consistent that unstructured, spoken group brainstorming produces fewer ideas than people working alone and pooling their ideas. Group work still has value for building on ideas, clarifying them and creating shared ownership. Put the group work after individual thinking, not instead of it.

What is the difference between brainwriting and brainstorming? In brainstorming people speak ideas aloud one at a time. In brainwriting people write ideas silently and in parallel, then pass them on for others to build on. Everyone contributes at once, so nobody waits, and quieter people are heard as much as confident ones.

How do I run silent brainstorming online? Give everyone the same question in a shared space, set a timer for five to ten minutes, and ask people to add notes without reading others' at first. Then open the board, let people build on what they see for another round, and only then cluster and discuss.

Where to go next

Start with the brainwriting and Crazy 8s guides, and sharpen your question first with how might we. For more on why silence comes first, read silent first: why the best ideas come before the discussion. For choosing the right method for any situation, see which innovation tool should I use.