Structured Brainstorming and Ideation

Solo brainstorming and its three blockers

Solo brainstorming is hard for three reinforcing reasons: you spin your wheels, you stay trapped in your usual thought patterns, and you self-censor before you’ve even articulated an idea. ChatGPT disarms all three because it doesn’t judge, it doesn’t get tired, and it suggests angles you wouldn’t have considered. The goal isn’t to land on “the right idea” on the first try, but to generate enough ideas that at least one is excellent. ChatGPT’s documentation actually lists brainstorming as a use case for conversational mode, reserving ChatGPT Work for multi-step tasks or heavier deliverables. Your job shifts from finding ideas to filtering them.

Four techniques that work

The simplest approach plays on raw volume. You describe the objective, context, and constraints, then explicitly ask the model not to filter anything. Asking for twenty ideas instead of five pushes you past the obvious: the first suggestions often repeat what you already had in mind. A marketing team that stops at five ideas will likely end up with the same old contest, newsletter, and webinar.

I need ideas for [objective].
Context: [context]
Constraints: [budget, timeline, resources]

Generate 20 ideas, from the most obvious to the most original.
Do not self-censor -- I want volume, not perfection.

The second technique varies the point of view rather than the quantity. ChatGPT reasons successively as different personas, and each brings its own reflexes: the engineer thinks tooling, the CFO thinks cost basis, the UX designer thinks user friction. The competitor is often the most fertile persona because they target your blind spots.

How would each of the following solve this problem:
- A Silicon Valley engineer
- A conservative CFO
- A user-obsessed UX designer
- An entrepreneur starting with zero budget
- A competitor trying to outpace us

For each persona, propose 3 distinct approaches.

The third technique flips the question entirely. Asking how to fail frees people up far more than asking how to succeed, and every identified cause of failure mechanically becomes a lever for improvement when reversed.

Instead of looking for ways to increase sales,
tell me the 10 best ways to drive our customers away.
Then, invert each point into an opportunity for improvement.

The fourth is the SCAMPER method, a creativity classic that suits an AI model well: where a group tires of sweeping through all seven angles, ChatGPT tackles them one by one without skipping any out of boredom.

Apply the SCAMPER framework to [product/service/process]:
- Substitute: what can we replace?
- Combine: what can we merge?
- Adapt: what can we borrow from another domain?
- Modify: what can we change in size, shape, or color?
- Put to other uses: what other uses are possible?
- Eliminate: what can we remove?
- Rearrange: what happens if we reverse the order?

For each letter, propose 3 concrete ideas.

Whichever technique you choose, how you frame the problem determines the quality of the harvest. “Give me ideas” doesn’t compare to “Give me 20 ideas to reduce our churn rate from 15% to 10% in 3 months”: the second version includes a target, a starting point, and a deadline, which creates implicit relevance criteria.

From raw idea to actionable idea

Generating ideas is only half the work. The most common mistake is to evaluate on the fly, discarding each proposal as soon as it seems unrealistic: this kills the production phase very quickly. Strictly separate the two phases, then ask for an explicit evaluation based on criteria you define.

Here are the 20 generated ideas:
[list of ideas]

Evaluate each idea on 3 criteria (score 1 to 5):
1. Potential impact
2. Ease of implementation
3. Alignment with our objectives

Rank them by total score and recommend the top 3.
For each selected idea, propose a 5-step action plan.

The action plan isn’t decorative. An idea that stays an idea is just a wish: breaking it down into five steps makes it defensible to leadership and executable the following week. Also keep in mind that a generated idea isn’t a fact: verify any figures, market examples, and references ChatGPT provides before including them in a presentation.

Recurring ideation

You don’t have to reserve this practice for big moments. You can spend ten minutes on Monday morning tackling the week’s challenges, kick off each project with an ideation session on possible approaches, or turn your monthly retrospective into a process-improvement workshop. Store your prompts and evaluated idea lists in a project: the effect becomes cumulative, since you build up both proven prompts and a reservoir of ideas to draw from.

Your turn

One real challenge, four techniques. Take a real, current professional challenge, not a textbook case. Frame it in one clear sentence, choose one of the four techniques, generate at least fifteen ideas, filter them using an impact/effort matrix, then develop the top three into action plans. Brainstorming works like a muscle: your prompts get sharper with every session.

Official sources: Prompting (ChatGPT documentation) · Projects in ChatGPT (OpenAI Help Center)

Key takeaways

• ChatGPT is a brainstorming partner that doesn’t judge and doesn’t get tired
• Four techniques: volume, personas, inverted constraints, SCAMPER
• Ask for more ideas than you need to move past the obvious
• Separate the generation phase from the evaluation phase
• Build brainstorming into your weekly routine for a cumulative effect