How to run brainstorming sessions that actually generate good ideas

Executive overview

Brainstorming sessions fail because humans hate the unknown and rush toward the first plausible answer — not the best one. Cognitive closure drives groups to settle on suboptimal solutions and stop searching.

The fix is not better facilitation inside the room. It is redesigning the conditions before, during, and after the session to maximise the volume of ideas — because quantity drives quality, and most teams underestimate the required volume by orders of magnitude.

Better brainstorming is a supply problem: get more raw material first, then refine.

Why brainstorms produce weak ideas

• Cognitive closure: unresolved questions are distressing; groups converge on the first plausible idea to escape discomfort
• The Einstein effect: people fixate on pattern-matching solutions and stop searching, even when told a better answer exists
• Once fixated, people are blind to better options — not just reluctant, but literally unable to see them
• Groups amplify individual biases: one person's premature relief becomes collective relief
• Leaders focus on solution output rather than idea input — "looking through the wrong end of the telescope"

Ideas as connections, not inventions

• No idea emerges from nothing; ideas are unexpected connections between things already known
• Think of each person's knowledge as a bag of Lego bricks — brainstorming is about combining bricks across bags
• More bricks = more possible connections = more innovative output
• Diverse teams matter because each person's "obvious" is built from different experience — what's trivial to one person is a breakthrough to another
• The range anxiety / mid-air refuelling example: hearing two unrelated concepts lets the brain do the connecting automatically

Quantity drives quality

• Research by Dean Keith Simonton: the single greatest predictor of idea quality is idea quantity
• Stanford data: ~2,000 ideas yield ~100 prototypes → ~5 products → ~1 commercial success
• Most teams guess 20-30 ideas is "a lot" — off by two orders of magnitude
• Idea flow = number of ideas generated per unit of time; optimising this is the primary lever
• MrBeast example: 30M-view videos aren't 30x more effort — they start with a dramatically better idea, then 2-3x the execution effort
• Reapportion time toward ideation before polishing; most teams do the inverse

The creative cliff illusion

• Most people expect creativity to decline sharply over time in a session — a "cliff"
• In reality, creativity doesn't degrade nearly as much as expected
• Expectation is self-fulfilling: teams that expect their best ideas early stop early; teams that expect improvement keep generating
• A session can produce a ramp, not a cliff — if the facilitator sets that expectation explicitly

The innovation sandwich

Structure a session across three stages, not just 60 minutes in a room:

Before

• Share the problem in advance — don't wait for people to gather before revealing it
• Ask participants to arrive with 10-20 ideas, not one
• Frame multiple "how might we" questions; different framings surface different solutions
• A problem reframed is half solved (slow elevator = boredom problem, not speed problem)

During

• Open with a warm-up that demonstrates the "yes, and" mindset vs. the default "why it won't work" mindset — participants feel the difference immediately
• Work in groups of max six; intimacy, trust, and speed require small groups
• Tell participants not to aim for creative — aim for obvious; what's obvious to each person is new Lego to everyone else
• Facilitator signals that the best ideas haven't arrived yet — counters the cliff illusion

After

• Resist the premature declaration of victory at the end of 60 minutes
• Close with: "What do these ideas make you think of?" — not "which idea is best?"
• Schedule a follow-up session; assign the problem to people's subconscious in the interval
• Make resource and accountability decisions only after the second session

Idea flow vs. idea pond

• An idea has no value until it moves — from ideation to experimentation
• Reduce friction between idea and first scrappy test; don't wait until you're proud of the prototype
• Reid Hoffman: if you're proud of your first prototype, you waited too long
• Iterate: scrappy fail → scrappy fail → scrappy refined fail → partial success → breakthrough
• Leading indicators (volume of experiments) predict lagging indicators (breakthrough quality); optimise leading indicators first