Most agency teams still treat brainstorming as a room, a whiteboard, and a confident facilitator saying, “No bad ideas.” The assumption is familiar: gather smart people, encourage free association, and breakthrough concepts will emerge.
The research is less flattering. Traditional verbal brainstorming often produces fewer ideas than the same number of people working independently and combining their output later. More recently, AI has complicated the picture further. It can help an individual produce a stronger idea while making the overall collection of ideas less diverse.
That doesn't mean brainstorming is broken. It means the format is doing more work than the enthusiasm. Better sessions separate idea generation from judgment, protect quieter contributors, control how people encounter one another's ideas, and introduce AI only when it won't narrow the opportunity space.
Table of Contents
- Why Most Brainstorming Sessions Fail
- What Decades of Brainstorming Research Shows
- The Evidence-Backed Workflow for Better Ideas
- Choosing the Right Brainstorming Format
- The AI Paradox in Creative Sessions
- Building Your Research-Informed Brainstorming Framework
- Templates and Examples for Agency Teams
Why Most Brainstorming Sessions Fail
The biggest mistake is treating participation as productivity. A room can be energetic, friendly, and full of sticky notes while still producing a narrow set of predictable concepts. The loudest participant establishes the direction, others adapt to it, and the team mistakes agreement for creative progress.
The foundational test came from Yale University in 1958, when Taylor, Berry, and Block compared interacting brainstorming groups with nominal groups. Nominal groups had the same number of people working alone before pooling their ideas. The nominal groups generated nearly twice as many ideas, later summarized as roughly a 2:1 advantage for individual-to-group idea production (the Yale brainstorming study and its historical context).
The three mechanisms that suppress output
Production blocking happens because only one person can usually speak at a time. While one strategist explains an idea, another participant must hold their thought, listen, or abandon it. In a fast-moving conversation, the team loses ideas before anyone records them.
Evaluation apprehension is the fear of looking foolish, off-brand, impractical, or insufficiently clever in front of colleagues. The familiar instruction to suspend judgment doesn't remove status differences. A junior designer may still hesitate before offering a strange concept to a creative director. This practical explanation of evaluation apprehension is useful for facilitators who want to understand why encouragement alone doesn't create psychological safety.
Social loafing spreads responsibility across the room. When everyone is responsible for generating ideas, each person can assume someone else will carry the creative load. Participants may remain engaged socially while contributing little original thinking.
Practical rule: If the first ideas shape the entire discussion, you aren't measuring the team's full creative range. You're measuring its response to the first confident voice.
The problem isn't collaboration itself. Research consistently points to a better sequence: generate privately, collect broadly, then discuss and develop. A group can add valuable stimulation during refinement, but asking everyone to invent, listen, judge, and build at the same time creates avoidable friction.
What Decades of Brainstorming Research Shows
Traditional group brainstorming remains common in agencies, yet the research challenges its central promise. The method was popularized by Alex F. Osborn in Applied Imagination, published in 1953. Later investigators tested whether interacting groups really generate more ideas than people working independently, rather than treating the practice as settled management wisdom. A historical review of brainstorming research traces that shift from prescription to evidence.
The consistent finding is uncomfortable for teams that equate collaboration with creativity: nominal groups usually produce more unique ideas, and often better ones, than live verbal groups. Production blocking, evaluation apprehension, and social loafing help explain the gap, particularly as groups grow larger. A review of group brainstorming research documents these recurring effects across the literature.
The format changes the result
Group size changes individual output. One field study found that groups of 5 to 7 participants generated more ideas per person than groups of 9 to 19. All groups produced more ideas during the first 5 minutes of a 15-minute session than later in the meeting, according to research on group size and interaction format. Longer sessions therefore do not guarantee broader exploration. Attention and novelty can fall quickly.
Electronic brainstorming introduces a different trade-off. Anonymous or asynchronous submissions reduce production blocking because participants do not wait for a speaking turn. The reported advantage is clearer in larger groups. Smaller teams can often gain similar benefits through written or nominal methods, without the setup and maintenance of additional software.
Quantity remains only one measure. Seeing other people's ideas can trigger combinations and improve originality, especially when the inputs are varied rather than repetitive. Structured debate can also outperform an unbounded “anything goes” atmosphere, because participants can challenge assumptions without making personal judgment the focus.
| Study / Researcher | Key Finding | Practical Implication |
|---|---|---|
| Taylor, Berry, and Block | Nominal groups generated nearly twice as many ideas as interacting groups in the 1958 Yale test. Read the historical study summary | Collect individual ideas before discussion. |
| Group-size research | Groups of 5 to 7 produced more ideas per person than groups of 9 to 19 in one field study. | Do not assume a larger room creates a larger opportunity set. |
| Nominal-group reviews | Published studies generally report more unique and higher-quality ideas from nominal groups than live brainstorming groups. | Separate generation from evaluation. |
| Brainwriting research | A 6-3-5-style hybrid approach outperformed nominal groups on idea count while matching them on quality. Explore the brainwriting findings | Use controlled idea exposure after parallel generation. |
For adjacent content and creative research, quso.ai's research findings offer a useful supplementary resource. The practical conclusion is compact: individual work creates breadth, structured interaction creates recombination, and deliberate criteria create better selection. Agencies can preserve collaboration by assigning each mode a different job instead of asking one conversation to generate, judge, and develop ideas simultaneously.
When a session repeatedly settles on its first plausible answer, examine the social dynamics through this explanation of groupthink in psychology. The problem is not intelligence. It is whether the process exposes independent thinking before consensus takes hold.
The Evidence-Backed Workflow for Better Ideas
A productive session starts before anyone enters the meeting. Write a problem frame that names the audience, the desired change, and the relevant constraint. “How might we make this campaign better?” invites familiar answers. “How might we make first-time buyers feel confident before they compare alternatives?” gives the team a useful boundary without dictating the solution.
A practical sequence
Frame the challenge. State what the team is solving, who benefits, and what must remain true. A tight frame prevents the conversation from drifting into unrelated possibilities.
Ideate in silence. Give each person a private period to write ideas without discussion. Use brainwriting, prompts, sketches, or the 6-3-5 method. The point isn't to force a particular technique. It's to preserve parallel generation.
Pool ideas anonymously. Add contributions to a shared board or collect them through cards. Remove names during the first review. Participants should encounter concepts before they encounter status, ownership, or perceived seniority.
Build in small groups. Ask pairs to combine, extend, or challenge selected ideas. Small-group work creates cognitive stimulation without forcing every participant to compete for the same speaking window.
Evaluate independently. Score ideas against pre-agreed criteria such as novelty, feasibility, audience value, and strategic fit. Only then discuss the pattern of scores and the reasons behind them.
The critical design principle is temporal separation. Generation and evaluation should not happen simultaneously. A participant who is judging an idea while inventing it will usually self-censor, while a group that critiques every suggestion in real time will narrow the field before it has explored it.
For a 90-minute session, allocate time to framing, silent generation, anonymous pooling, clustering, small-group development, and independent scoring. The exact allocation should depend on the complexity of the brief, but the order should remain stable. Remote teams can collect ideas asynchronously before the live meeting, then use the shared session for clustering and refinement. The brainstorming process steps can help facilitators turn that sequence into a repeatable operating habit.
Facilitator test: If participants can explain why an idea was rejected before they have finished generating, evaluation has arrived too early.
Choosing the Right Brainstorming Format
There isn't one universally superior brainstorming format. The right choice depends on whether the team needs broad exploration, deliberate recombination, or a defensible selection process. Agencies get into trouble when open discussion becomes the default for every brief, from a positioning challenge to a last-minute headline exercise.
Traditional verbal brainstorming works when the team needs quick social energy and live development, especially after participants have prepared individually. It performs poorly when used as the only ideation method because speaking order, status, and immediate judgment shape the output.
Brainwriting is a stronger default for divergent work. Participants write independently, then expose ideas in a controlled way so others can extend them. The 6-3-5-style research summarized earlier supports this hybrid logic, preserving parallel generation while enabling recombination. Teams looking for a plain-language explanation can review how brainwriting works.
Nominal group technique is useful when the team needs independent contributions followed by structured prioritization. It protects breadth and makes each person's input visible without requiring a long debate.
Electronic brainstorming fits distributed teams and larger participant pools. Anonymous input reduces the pressure of speaking publicly, but excessive anonymity can reduce accountability. A named facilitator still needs to clarify the question, remove duplicates, and make the next action explicit.
| Format | Best For | Idea Quantity | Idea Diversity | Team Size Sweet Spot | Key Risk |
|---|---|---|---|---|---|
| Traditional verbal | Live refinement and shared context | Variable | Can narrow early | Small teams | Dominant voices |
| Brainwriting | Broad exploration and recombination | Strong | Strong when prompts vary | Small to medium teams | Weak follow-through |
| Nominal group technique | Independent generation and prioritization | Strong | Strong | Small to medium teams | Mechanical discussion |
| Electronic brainstorming | Distributed or larger participation | Strong with good design | Depends on prompt and exposure | Larger or remote teams | Social loafing |
| Hybrid asynchronous | Preparation plus focused live work | Strong | Strongest when ideas are independently seeded | Distributed teams | Fragmented context |
The decision rule is straightforward. Use written or asynchronous methods when diversity matters most. Use live conversation after the team has a visible pool of independent ideas. Use anonymous tools when status differences are suppressing contributions, but retain enough ownership to support accountability during development.
The AI Paradox in Creative Sessions
AI creates a tempting shortcut. A blank prompt produces an immediate list, which can help a tired team start moving. But speed and volume don't prove that the team has explored a wider creative territory.
A 2025 Nature Human Behaviour study, summarized by Wharton, found that ChatGPT-assisted brainstorming often increased the creativity of individual ideas while making the overall set less diverse. Across five experiments, diversity declined consistently, with 37 of 45 statistical comparisons showing significant declines (Wharton's summary of the Nature Human Behaviour study).
Why stronger individual ideas can create a weaker portfolio
An AI system can produce a polished variation of a familiar association. That makes a single suggestion feel useful, but when several participants use similar prompts, their outputs can cluster around related concepts. Early suggestions also create an anchoring effect. Once the team sees a plausible direction, later ideas tend to respond to that direction instead of exploring unrelated frames.
This is the paradox agencies need to manage. Clients don't pay only for a competent idea. They pay for a distinctive interpretation of the brief, a point of view that competitors are less likely to reach. AI can support synthesis, but early exposure may compress the range that makes synthesis valuable.
Use AI after human exploration, not before it. Ask participants to generate independently first, then use a model to combine, stress-test, invert, or organize the human pool. Keep AI outputs labeled and separate from original contributions so the facilitator can see whether the tool is expanding the set or merely polishing its center of gravity.
For teams producing campaign or video concepts, efficient AI-powered video script tips can be useful during later development, when the creative direction already exists and the task is shaping execution.
The satisfaction question matters too. A 2025 SSRN study found that students using generative AI earned comparable rubric scores to those brainstorming without it, yet overestimated AI's positive effect on their ideation ability and felt less satisfied with the session (the SSRN study on AI, ideation, and satisfaction). Good output isn't enough if participants feel less ownership of the thinking.
Building Your Research-Informed Brainstorming Framework
A useful framework preserves the research principles while leaving room for the realities of agency work. Briefs change, clients interrupt, and teams vary in confidence. The structure should be fixed where it protects independent thinking, flexible where it supports the task.
Five phases that hold up under pressure
Begin with pre-session individual work. Send the brief early and ask participants to develop initial thoughts privately. This establishes a baseline before group dynamics influence the work and gives quieter contributors something concrete to bring into the room.
Frame the challenge visibly. Put the audience, desired change, constraints, and decision criteria on one page. A facilitator should revise ambiguous wording before the session starts, not attempt to repair a vague prompt while participants are already generating.
Run guided ideation. Use silent writing, varied prompts, and controlled exposure. The facilitator's job is to create movement without rewarding speed, volume of speech, or conformity. Rotate prompt types so the team explores audience, behavior, tension, analogy, constraint, and reversal.
Select against criteria. Ask participants to score ideas independently before discussion. A rubric won't eliminate judgment, but it makes judgment inspectable. It also prevents the most charismatic presenter from turning a weak concept into the apparent group favorite.
Commit to next steps. End with named owners, a decision date, and the specific evidence needed to develop or reject each promising concept. Let ideas rest before final commitment when the project allows. A short incubation period can give participants distance from the excitement and disappointment of the session.
Design principle: Change the prompts and timing when necessary. Don't remove the independent generation phase just because the team is busy.
Group composition deserves attention beyond job title. Include people who understand the audience, the channel, the commercial goal, the operational constraint, and the cultural context. Cognitive diversity only helps when the process allows different perspectives to appear before the group converges.
Facilitators should model curiosity without performing certainty. They can ask, “What assumption does this idea challenge?” or “What would make this useful to the audience?” They shouldn't reward the first polished answer, invite premature voting, or use seniority to settle disagreement.
Templates and Examples for Agency Teams
The best test of brainstorming research is whether a team can use it on a deadline. These templates keep the core safeguards intact while changing the amount of live discussion.
A 60-minute hybrid brainwriting session
Use this for a campaign angle, rebrand territory, or messaging problem.
- Before the session: Send a one-page brief and ask participants to arrive with private notes.
- First 10 minutes: Everyone writes individually, using three different prompts.
- Next 10 minutes: Collect ideas anonymously and remove duplicates without judging them.
- Next 20 minutes: Small pairs combine and extend the strongest clusters.
- Next 10 minutes: Participants score concepts independently for novelty, feasibility, audience value, and fit.
- Final 10 minutes: Discuss the highest-potential concepts and assign development owners.
This format protects against production blocking and evaluation apprehension while retaining live cognitive stimulation.
A 90-minute asynchronous-first sprint
Use this for distributed teams or a brief that needs input from account, strategy, creative, and client-facing roles.
- Preparation: Post the problem frame and separate prompt threads in the shared workspace.
- Individual phase: Participants contribute independently before reading the full pool.
- Synthesis phase: A facilitator clusters ideas and identifies gaps.
- Live meeting: The team examines clusters, combines promising directions, and records unresolved questions.
- Close: Participants score the refined concepts and agree on the next deliverable.
The facilitator should keep original ideas visible beside merged concepts. Otherwise, useful minority directions can disappear during synthesis.
A 45-minute constraint session
For tight deadlines, name one hard constraint, such as a particular audience behavior or channel limitation. Spend the opening period generating privately, then use the remaining time to select and sharpen concepts against that constraint. The restriction should focus thinking, not dictate the answer.
| Template | Duration | Ideal Team Size | Core Research Principle | Best For |
|---|---|---|---|---|
| Hybrid brainwriting | 60 minutes | Small to medium | Parallel generation before collaboration | Campaign and brand concepts |
| Asynchronous-first sprint | 90 minutes | Distributed cross-functional team | Independent input with controlled synthesis | Complex briefs |
| Constraint session | 45 minutes | Small working group | Focused divergence followed by criteria-based selection | Tight deadlines |
Before any session, confirm the decision the team must make, prepare varied prompts, choose the collection method, and define the scoring rubric. During the session, record every idea, protect silent time, and interrupt discussion that turns into advocacy too early. Afterward, preserve the raw pool, document why concepts moved forward, and assign the next action.
Bulby is one option for teams that want a guided environment for anonymous idea submission, creative prompts, automatic collection, and AI-assisted summaries. It can support the structured path from challenge definition through idea development without requiring the facilitator to manage every contribution manually.
Use your next session to test the sequence, not your team's patience. Give people private generation time, collect ideas anonymously, delay evaluation, and introduce AI only after the human opportunity space is visible. If you want a guided platform for running that kind of collaborative process, visit Bulby and explore how it can help your agency turn raw contributions into actionable creative directions.