# Brainstorming: Generate Ideas and Choose the Next Step

Brainstorming guide: workshop protocol, individual divergence, convergence, transparent criteria, and next-step testing.

- Canonical URL: https://www.harmate.com/en/blog/brainstorming-generate-ideas-and-choose-the-next-step
- Author: Harmate Team
- Published: 2026-09-24
- Updated: 2026-09-24T00:43:05.944663+00:00
- Language: en

## Content

# Brainstorming: Generate Ideas and Choose the Next Step

Brainstorming is an idea-generation method, not a complete decision procedure. To reduce the first-speaker effect, start with a short round of individual brainwriting: everyone writes before discussion, then the group shares the proposals. Brainstorming becomes misleading when an idea list is confused with a solution, a vote replaces analysis, or the fastest speakers set the frame.

A sound protocol separates two moments: generate without immediate evaluation, then examine and choose with announced criteria. [ASQ describes this separation](https://asq.org/quality-resources/brainstorming) and recommends keeping ideas in contributors’ own words. Research by [Korde and Paulus](https://www.sciencedirect.com/science/article/pii/S0022103116302591) examines alternating individual and group ideation; their experiments suggest that alternation can support idea generation, but this does not guarantee decision quality or a successful workshop. A study of selection also warns that producing many ideas is not enough to select a better solution ([Rietzschel, Nijstad, and Stroebe](https://www.sciencedirect.com/science/article/pii/S0022103105000600)).

## Define the question before looking for ideas

“Find ideas” is not enough. Write the decision that will follow and the constraint that cannot be ignored:

- which situation are we trying to improve;
- for whom and in what context;
- what decision will be possible at the end;
- which constraints are non-negotiable;
- what sign will show that a path deserves a test.

A frame that is too broad produces slogans. A frame that is too closed leaves no room for divergence. A useful formulation describes a problem without prescribing the solution: “How might we help a distributed team get useful feedback after training without adding a mandatory meeting?”

[Open-ended questions](/en/blog/open-ended-questions-get-actionable-data-without-bias) can precede the workshop to collect situations and friction. They are not a pre-session vote. They provide contextual material the group can understand, complete, or challenge. If you use them, preserve the original wording and the collection context: one response is not a frequency measure.

Before the session, decide what will be kept. An idea may be attributed, anonymized, or associated only with a group, but participants should know the rule. Explain who will see the notes, how long they will be kept, and which decision they inform. This preparation helps people share a sensitive possibility without assuming that it will immediately become a public commitment.

## Brainwriting before discussion

Brainwriting does not guarantee that every idea will be kept. It gives each person time to generate options before discussion, hierarchy, or the first participant’s vocabulary frames the rest.

## A two-movement protocol

### 1. Prepare divergence

Announce the rules:

- suspend criticism during generation;
- record ideas in the contributors’ words where possible;
- ask for clarification without evaluating;
- allow silent writing before speaking;
- keep minority and incomplete ideas.

Start with a few individual minutes. Each participant writes several responses, including an impractical idea, a minimal variation, and an option that changes the frame. This prevents the first exchange from imposing one interpretation. ASQ describes a related “sticky storm” technique in which people prepare notes before sharing; a visible record reduces lost ideas and the effects of dominance.

Then share in a round. The facilitator records each proposal separately and asks “What do you mean by…?” rather than “Is that realistic?” Participants can build on an idea, but the initial wording and contributor remain visible. Clarification is not fusion: “send a reminder” and “ask for an example of use” may later be combined, but they should not become one proposal before the group understands each one.

Alternation need not be mechanical. With a new group, allow more silence and written contributions. With an already aligned team, add a perspective-changing prompt: “Which absent person would see the risk?” or “What could work without another meeting?” For a sensitive issue, collect options anonymously first and do not ask people to defend a proposal publicly.

### 2. Organize convergence

After generation, announce the rule change. Criticism is now possible, but it concerns options and conditions, never people’s worth.

Group ideas by mechanism or expected outcome, not only by shared words. Mark groupings that are debatable. Distinguish four operations that are often mixed together:

1. **Clarify**: make a proposal understandable without changing its meaning.
2. **Group**: bring together ideas that target the same mechanism.
3. **Combine**: create a new option from two ideas while keeping the originals.
4. **Defer or reject**: put an idea aside with a reason that can be checked.

Then define three or four useful decision criteria:

| Criterion | Question |
|---|---|
| Value for the situation | What specific problem could this improve? |
| Immediate feasibility | Which dependency or resource is missing? |
| Risk | Who might be excluded or experience an unwanted effect? |
| Learning | What reversible test could reduce uncertainty? |

A criterion should guide examination, not produce a magic score. If you use a matrix, keep the reasons, disagreements, and values that cannot be compared cleanly. An average rating does not make a serious risk disappear. The group can use ordinal judgments (“low”, “investigate”, “promising under conditions”) and state what would change its mind. A final vote may allocate attention, but it does not by itself settle safety, compliance, or equity constraints.

### When collection comes before the workshop

A group rarely arrives without a history. A short collection can surface situations that people would not volunteer in the room. Ask an open question tied to the decision: “Describe the last moment when the journey made you lose time.” Do not then turn the answers into an automatic ranking of ideas. Let the group read excerpts, identify contradictions, and say what it recognizes, disputes, or still needs to learn.

Closed questions can later measure an option that has already been defined or check the reach of a friction. They compress reality for comparison; they do not replace open exploration. A frequency of “difficult” responses does not yet say which obstacle should be fixed. Brainstorming opens hypotheses; a decision requires linking each hypothesis to evidence and a next test.

## Workshop example: improving feedback after training

Invented example: a team wants to help remote learners apply training. The question is: “How can we learn about real use within two weeks without sending a long survey or requiring a meeting?”

### Individual divergence

For six minutes, everyone writes without discussion. One person proposes a reminder with one question. Another imagines an optional office hour. A third wants to ask for an example of use or non-use. A fourth suggests removing the follow-up and asking managers to remind people about the training. A fifth proposes contacting only people who did not open the resource, although the team does not yet know whether opening means use. That idea is kept together with its unstated assumption.

The facilitator groups ideas only after sharing. They distinguish channel, question, and responsibility. They do not turn “participants do not answer” into a general fact: the team currently has only a few reports. They also record an absence: people who did not attend the workshop have not given their view.

### Sharing and choosing a test

The ideas form three families: a short follow-up, available support, and collection through managers. One idea combines a follow-up with an example of use. A real disagreement remains: some people want to understand obstacles, while others only want to know whether the training was useful. The group records this as a framing decision, not a wording detail.

The team examines the families through four criteria. The short follow-up is feasible and low-friction, but it may miss people who have not tried the skill. The office hour could provide richer context, but requires availability and may reach only the most motivated. A manager reminder is quick, but may make responses less candid and depends on a role that is not always available. The opening signal is easy to measure, but does not prove application.

The decision is not “the highest-scoring idea.” The group chooses a two-step test: send one open question to everyone (“What were you able to try, or what stopped you from trying?”), then invite people to clarify an obstacle if they wish. It defines what it will observe in advance: types of obstacles, people it did not reach, requests for help, and concrete examples of use. It keeps the manager reminder as an option to examine if the voluntary channel misses some profiles. Criteria, absences, original wording, and the reason for the choice stay in the record.

This example is invented. It shows that divergence is not about producing many options and convergence is not about quickly removing uncomfortable ideas. The useful output is a bounded test, not a promise of transformation.

## Protect the material produced

A minimal record keeps:

- the question and decision;
- original contributions and their author or anonymity rule;
- production stage and clarifications;
- proposed groupings and contested connections;
- selection criteria, limits, and disagreement;
- chosen idea, test, owner, and expected observation;
- ideas deferred and the reason for waiting;
- people absent, the source of prior information, and points still to check.

Add a post-session register: test date, observed event, decision after observation, and hypothesis that remains open. Otherwise the group may repeat the same brainstorming using reconstructed memories. A frequent idea is not automatically the best. A minority idea may reveal an important case or risk that the dominant group did not encounter.

If contributions are collected remotely, distinguish collected material from interpretation. Harmate can retain responses and help compare themes, but a theme is still a reading: keep verbatims, divergent cases, and context. Do not infer a cause, general preference, or priority simply because a grouping contains many responses.

## Brainstorming, brainwriting, and structured discussion

Brainstorming can be oral, written, or mixed. Brainwriting emphasizes written generation before speaking; structured discussion adds clarification and argument. These formats can be combined, but they do not create the same participation conditions. Work by [Paulus and Yang](https://www.sciencedirect.com/science/article/pii/S0749597800928887) highlights the attention people give to shared ideas and the value of reflection after exchange. That is a design clue, not a universal recipe.

[World Café](/en/blog/world-café-method-90-minute-agenda-and-15-facilitation-questions) organizes rotating exchanges around tables or groups; it does not replace individual generation when the first conversation might steer all later ones. [Concertation](/en/blog/stakeholder-consultation-deciding-without-false-consensus) addresses another need: making positions and conditions for a collective decision explicit. Choose the format for the decision, not for the most familiar name.

## Where Harmate can help

Harmate can collect contributions, import answers, preserve wording, and help compare themes before a report. Open responses often provide richer qualitative material for exploring causes, exceptions, or unexpected wording; closed questions help when categories are already stable and comparisons are needed. Neither format alone guarantees a valid or representative decision.

Harmate does not provide a native brainstorming workshop, judge the best idea, or replace facilitation and arbitration. It can help keep the link between a theme, the responses supporting it, and contradictions to revisit. The decision remains human and should state what was actually checked.

A workshop is finished when the next step is clear: an idea to test, information to seek, or a decision not to continue. If you cannot explain why a path was selected or deferred, convergence probably replaced analysis. Brainstorming is not meant to manufacture immediate consensus; it opens the field enough to choose the next verification with visible criteria.