How to Facilitate a Brainstorm That Produces Real Ideas

To facilitate a brainstorm that produces real ideas, stop opening with 'right, shout them out'. The single change that does most of the work is separating generation from discussion: give everyone six to eight minutes of silence to write their own ideas privately, then spend the talking time building on a pile that already exists. Groups that generate out loud reliably produce fewer distinct ideas than the same number of people working alone and pooling afterwards, because only one person can speak at a time and everybody else spends the wait rehearsing rather than thinking.

The second change is controlling the order in which ideas enter the room. Whatever is said first sets the frame, and every contribution after it tends to be a variation on that frame rather than a genuinely different direction. Collect in writing, reveal the whole pile at once, and the anchoring problem largely disappears.

This guide covers the brief you write beforehand, a 60-minute run sheet for six to twenty people, how to run the silent phase so it does not feel like an exam, how to converge without flattening the odd ideas, the four ways brainstorms die and how to recover each one, and what has to happen in the 24 hours afterwards.

Why open brainstorms produce fewer ideas

The classic failure is structural, not cultural. In a verbal brainstorm only one channel is open, so with twelve people in the room each person gets under five minutes of airtime in an hour. While you wait for a gap you hold your idea in working memory, rehearse it, and stop generating new ones. That effect - production blocking - is the best-documented weakness of group brainstorming and it gets worse as the group gets bigger, not better.

The second drag is evaluation apprehension. Half-formed ideas are exactly the ones worth having, and they are the ones people will not say in front of their manager or a client. If your senior stakeholder is in the room and speaking, expect the junior half of the group to contribute almost nothing of substance.

The third is anchoring. Say 'what if we did a referral scheme' in minute two and you will collect nine referral schemes and nothing else. None of these problems are solved by asking people to be more creative or by putting a bowl of sweets on the table. They are solved by changing the format so that writing comes before talking.

Three formats, three different outputs

Pick the format from the job. Open verbal is fine for a five-minute warm-up on a low-stakes question. For anything you intend to fund or build, use silent generation, and use a nominal group structure when the group is politically loaded and you need the ranking to be defensible.

Write the brief before you book the room

A 60-minute run sheet for six to twenty people

This is the shape that survives contact with a real calendar. Every block has a hard end, announced before it starts and enforced with a visible timer. The most common facilitation error here is letting the silent phase run long because the room seems busy - do not, because the last two minutes of silence are where the obvious ideas run out and the interesting ones start, and if you extend it people relax instead of pushing.

If you only have 30 minutes, cut the build phase and the second vote, not the silent generation. Generation is the part that cannot be compressed without changing the output.

Silent generation: the eight minutes that decide the session

Set the expectation numerically. 'Write eight ideas' produces far more than 'write down your ideas', because a target makes people push past the first three obvious ones. Nobody is graded on hitting it, and you should say so, but the number does the work.

Keep the channel private while people write. If contributions appear on the shared screen as they are submitted, you have reinvented the anchoring problem in text form - people read what is already there and converge. In Session Flo you can collect an open-text activity with the display hidden and release the whole board in one go, which is the behaviour you want here.

Silence genuinely is uncomfortable for the first ninety seconds, and facilitators break it far too early. Say at the start: 'the first two minutes will feel awkward, that is normal, keep writing.' Then stay quiet yourself. Typing noise is a good sign; if the room is completely still at minute four, your prompt was too abstract and you should stop the clock and rewrite it with the group rather than let eight minutes drain away.

Converging without flattening the good ideas

Convergence is where most brainstorms lose the value they just created. The default failure is voting on the raw list: forty items, eight of which are near-duplicates of the same safe idea, so the safe idea wins on split-free votes while the one genuinely novel suggestion sits alone with two.

So cluster first, out loud, with the group correcting you. Read a card, ask 'is anything else the same thing?', merge, and give the cluster a headline in the words of whoever wrote the original. Merging is where ownership gets accidentally erased, and people stop contributing next time if their idea vanished into a group label they did not recognise.

Then vote on clusters, not cards, and vote privately. Three votes each is the right number for a shortlist of ten to fifteen clusters - enough to express a preference, too few to spread thinly. Reveal all votes at once rather than showing a running tally, otherwise the first few voters set the direction for everyone else.

Protect one outlier deliberately. Before you close, take the highest-variance idea - the one two people loved and everyone else ignored - and ask its author for sixty seconds. Roughly one session in four, that is the idea the group ends up funding once they hear the reasoning.

Four failure modes and the recovery

Remote and hybrid brainstorms need extra scaffolding

There is decent evidence that video calls narrow the visual field and reduce the number of creative ideas people generate compared with being in the same room, even though the same groups select just as well from the ideas they have. The practical reading is not 'never brainstorm remotely' - it is to protect the generation phase and let the selection phase happen on video.

So for a remote session, tell people to turn cameras off during the eight minutes of silent writing and look away from the screen. It feels strange the first time and it measurably helps. Bring cameras back on for the reveal, the build and the vote, where seeing each other is genuinely useful.

Hybrid is harder than fully remote because the room can converge in side conversation while the dial-ins wait. The fix is to put everyone on the same input channel: even people sitting in the same room submit ideas on their own phone or laptop, and nothing is written on the physical whiteboard that is not also on the shared board. If you keep a parallel analogue artefact, the remote half of the group is permanently a step behind.

The 24 hours after decide whether it mattered

Send the full raw list, not just the winners, within a day. People need to see that their unpicked idea was recorded rather than discarded, and the unpicked list is the first place to look when the chosen idea turns out to be harder than expected. A session recap that goes out while the session is still fresh gets read; one that goes out the following week does not.

Alongside the list, publish three lines: what was chosen, who owns it, and the date something visible happens. Then actually return to the list at the next team meeting, even for thirty seconds. The reason people arrive at brainstorms cynical is almost never the format - it is that the last four produced nothing they could point at.

Facilitator pre-flight checklist