· · 10-minute read
How to run a brainstorming session
A good brainstorming session needs one clear question, time to generate ideas without judging, and a way to choose what happens next. Use this agenda, the rules, and a workshop that fits the job.
A good brainstorming session does three things. It gives the group one clear question, time to produce many ideas without judging them, and a simple way to group those ideas and choose what happens next. A useful agenda has four parts: open, diverge, cluster, and converge.
The brainstorming session agenda below takes about 60 minutes. A smaller group with a narrow question can finish in 45 minutes. A broader question may need up to 90 minutes. The rules matter as much as the times. If people debate the first idea, the rest of the agenda never really starts.
Three workshops later in this post cover the jobs a brainstorm usually needs: a clear question, many rough ideas, or a decision and next steps.
Brainstorming session agenda
Use one question, written so a new person could understand it. "How might we reduce the steps a new customer takes before they send their first file?" is specific enough to work with. "Improve onboarding" is too broad to answer. Say what the group can decide in the room, and what is already fixed.
This 60-minute plan is a full brainstorm meeting agenda. You can shorten or lengthen the middle sections. Do not drop converge to make more time for ideas. A longer list with no owner is easy to save and hard to use.
- Open (5 minutes). Read the question aloud. Say what is in scope and what is not. Read the rules, including the point where judging is allowed.
- Diverge (25 minutes). Produce ideas with rapid ideation or a round-robin. Capture each idea. Do not discuss which ones are good yet.
- Cluster (15 minutes). Group similar ideas and give each group a short name. Merge duplicates. You are sorting. Choosing comes in the next step.
- Converge (15 minutes). Choose a small set to carry forward. Name an owner and the next action for each one. If the group cannot choose, name the question that still needs a decision.
For a 45-minute session, use 5 minutes to open, 15 to diverge, 10 to cluster, and 15 to converge. For 90 minutes, use 10 to open, 35 to diverge, 20 to cluster, and 25 to converge. The extra time belongs in diverge and cluster, not in a second round of debate about the first idea.
Brainstorming rules
Read these rules in the opening, and keep them visible. They are short on purpose. People can follow them while they are thinking.
- One idea at a time, or everyone writes in parallel. Pick one mode. If people are talking, one person speaks at a time. If people are writing, they write together and stay quiet. Doing both at once usually means the fastest talker sets the list.
- No judging while generating. Cost, feasibility, and taste wait until converge. A quick "that will never work" ends diverge early.
- Build on other people's ideas. Add to an idea, combine two, or take one in a new direction. Say which idea you are building on, so the link is visible.
- Go for quantity. A larger set gives you more to group and choose from. The first few ideas are often the obvious ones.
- Stay on the question. If an idea answers a different problem, park it in a separate list. You keep the idea without pulling the session off the question you opened with.
These rules apply to a meeting in a room and to a remote session. The medium changes how you capture ideas. It does not change when judging should start.
Rapid ideation
Rapid ideation means many ideas in a short, fixed time. People usually write at the same time, in silence, so they are not waiting for a turn and not polishing sentences for an audience. One or two minutes per idea is enough to move past the first thought. Rough wording is fine. The aim is a wider set of options before anyone argues.
Use it when a few people tend to speak first, or when the group is large enough that taking turns would fill the meeting. You still need a clear question. A vague prompt produces vague ideas, only faster.
In Clarvelo, Crazy Eights is the short workshop for this. Each person produces several distinct ideas on a timer, in about eight minutes, before the group debates them. It is useful when the question is already clear and you need more options. A finished answer can wait.
Round-robin brainstorming
Round-robin brainstorming gives each person one turn before anyone goes again. Go around the room, or around the list of names in a remote meeting. Each person offers one idea. Anyone can pass. When the round ends, start again until new ideas slow down.
This works in a small group, up to about eight people, because each turn stays short and every person is heard. It also makes it natural to build on the idea just said. In a larger group, a full round takes too long, and people near the end forget what they wanted to add. Silent rapid ideation fits better there.
A turn is one idea, then the next person. Do not turn the round into a discussion. Save discussion for cluster and converge.
A Lightning Decision Jam starts with this kind of turn-taking. Each person adds one point, then the group votes. The round is for collecting material.
When a brainstorming session fails
Most failed sessions have the same few causes. The fix is usually a change in process. Asking people to think harder does not help if the first idea is already being debated.
- Debate starts while ideas are still appearing. The first suggestion gets examined, and quieter ideas never get said. Keep diverge and converge separate. When you need many rough options before any debate, use Crazy Eights.
- One voice fills the time. The same person frames the problem, offers the ideas, and critiques them. Use a round-robin so each person adds one idea, or silent rapid ideation so ideas are written before anyone speaks.
- The meeting ends with notes and no next step. The group ran out of time after generating ideas. Brainstorming gives you options. Choosing among them needs its own time. A Lightning Decision Jam is built to leave with a direction and next steps.
- The prompt is still a pile of complaints. "Improve onboarding" means something different to each person, so the ideas cannot be compared. How Might We turns that pile into a question the team can answer.
Let an AI facilitator run the session
Someone has to hold the question, the timer, and the rules. If that person is also trying to contribute ideas, the rules slip, or their own ideas never get written down.
In Clarvelo, an AI facilitator runs the live workshop. It shows the prompt, collects what people write, and moves the group from one step to the next. Guests join with a link. They do not need a separate whiteboard account. The organizer can take part with everyone else.
People still bring the context and make the judgment. The AI facilitator keeps the process intact so the group can think together.
Pick the workshop that matches the job
Use one of these three when you want Clarvelo to run the steps. Each one is good at a different job.
How Might We
How Might We is for turning a messy pile into a clear question. Use it when the group has observations, complaints, or half-formed ideas, and not yet a question they can answer together. People rewrite that material as How Might We questions, group similar ones, and vote on the frame most likely to lead to useful work. It takes about 60 minutes and fits 1 to 8 people. It is a poor fit when the question is already sharp and the group only needs more options.
Crazy Eights
Crazy Eights is for getting many rough ideas, from everyone, before debate. Use it when the question is already clear. Each person generates several distinct ideas on a short timer, about eight minutes. The point is breadth. The ideas can be rough. Judging comes later, in a different step or a different session. It fits 1 to 8 people. It will not, on its own, choose a winner or assign owners.
Lightning Decision Jam
Lightning Decision Jam is for going from ideas to a decision and next steps in one session. Use it when discussion has been long and the group still has no closure. People collect points in turn, vote on what matters, reframe the chosen problem, and compare possible actions by impact and effort. It takes about 80 minutes and fits 2 to 10 people. You leave with a direction and a realistic next move, rather than another list to revisit later.
Choose How Might We when the question is muddy, Crazy Eights when you need more options fast, and a Lightning Decision Jam when the session has to end in a decision. You can use them in that order on different days. You do not need all three in one meeting.
Frequently asked questions
How long should a brainstorming session be?
Most teams can do useful work in 45 to 90 minutes. Sixty minutes is a practical default: a short open, time to generate, time to group ideas, and time to name a next step. Past 90 minutes, the meeting often slides into open discussion. Under 45 minutes can work for a small group with a narrow question, especially if you only need the diverge step, as in Crazy Eights.
How many people should join?
A group of about 3 to 8 people is easier to run. Everyone can contribute, and clustering stays manageable. Above about 10 people, silent writing works better than taking turns, because a full round-robin becomes slow. In Clarvelo, How Might We and Crazy Eights fit 1 to 8 people. A Lightning Decision Jam fits 2 to 10.
What is the difference between brainstorming and a decision workshop?
A brainstorming session produces and groups ideas. A decision workshop chooses among options and names what happens next. You can do both in one meeting if you protect time for each, using the converge block in the agenda above. A list of ideas does not become a decision by itself. When the group needs a direction and owners before they leave, use a Lightning Decision Jam.
Can an AI facilitator run a brainstorming session?
Yes. In Clarvelo, the AI facilitator runs the live workshop. It shows the prompt, keeps the time, and moves the group through the steps. Guests join with a link, and the organizer can contribute with everyone else. People still supply the ideas and the judgment. Use How Might We to frame the question, Crazy Eights to generate options, or a Lightning Decision Jam when the session should end in a decision and next steps.
What should you do with the ideas after the session?
Before you stop, name a small set to pursue, who owns each one, and the next action. That action might be a short test, a clearer How Might We question if the ideas showed the problem was still fuzzy, or a Lightning Decision Jam if the group still needs to choose. Send the grouped list to the people who will act on it, with the question those ideas were answering. A document with no owner is where most brainstorming sessions quietly end.