Move through a structured six-step creative process from problem to solution to action
Contents
Decision tool
Inversion
Ask what would guarantee failure and design your solution to avoid those conditi
Decision tool
SCAMPER
Systematically generate variations by Substituting, Combining, Adapting, Modifyi
Decision tool
Zwicky Box
Generate novel solutions by combining dimensions of a problem in unusual ways
I send a newsletter every week — free, no spam, unsubscribe anytime.
| Dimension | Best fit |
|---|---|
| Problem type | Open-ended challenges where the problem definition itself is uncertain or contested. The model earns its keep when different stakeholders would describe the problem differently — a signal that the real problem hasn't been articulated yet. Strategic pivots, product-market fit questions, go-to-market redesigns, organisational friction that everyone feels but nobody can name precisely. |
| Team composition | Cross-functional groups of 4–8 people with diverse perspectives on the problem. The divergent phases need cognitive diversity to work — if everyone in the room has the same mental model, you'll generate 30 variations of the same idea. Include at least one person who is close to the customer and one who is far from the problem domain. |
| Stakes and complexity | High enough to justify 4–8 hours of structured thinking. The model is too heavy for tactical decisions ("which vendor for office supplies?") and too structured for pure exploration ("what should our 10-year vision be?"). Its sweet spot is consequential decisions with a bounded solution space — problems where the answer exists but isn't obvious. |
| Organisational culture | Teams willing to tolerate ambiguity for the first half of the process. If the room demands solutions in the first 20 minutes, the model will collapse under political pressure. It requires a facilitator — or a cultural norm — that protects the problem-definition phase from premature convergence. |
| Failure pattern | What goes wrong | What to use instead |
|---|---|---|
| Rushing to Step 4 | The team treats Steps 1–3 as bureaucratic preamble and races to "Generate Answers." The catalytic questions are vague or identical to the original problem statement. The solutions generated are the same ones the team would have produced without any framework at all. The model's entire value proposition — reframing the problem before solving it — is gutted. | Enforce a time minimum on Steps 1–3 (at least 40% of total session time). Use a facilitator whose sole job is to block premature solution talk. |
| Solo application without external input | One person runs all six steps alone. The divergent phases produce narrow output because a single mind has a single set of assumptions. The "third-third" ideas never materialise because there's no cognitive friction to push past the obvious. The model degrades into a structured journaling exercise. | At minimum, bring in 2–3 other perspectives for Steps 3 and 4. Even informal input from people outside the problem domain can break the pattern. |
| Crisis or time-critical decisions | The model requires 4–8 hours of deliberate work. In a genuine crisis — server down, PR disaster unfolding, cash running out in weeks — that time doesn't exist. Attempting the full six steps under extreme time pressure produces shallow work at every step, which is worse than a fast heuristic applied well. |
| Time horizon | Decisions with implementation timelines of weeks to months. The model produces actionable plans, not philosophical insights. If you need a decision in the next hour, use something faster. If you're exploring a decade-long strategic direction, the model's action-planning step will feel premature. |
| Prior attempts | Especially powerful when the team has already tried the obvious solution and it didn't work. That failure is evidence that the problem was misdiagnosed — exactly the condition the first three steps are designed to address. The model's value increases in direct proportion to how "stuck" the team feels. |
| OODA Loop for fast-cycle decisions. Pre-Mortem if you have 60 minutes. Reversible vs. Irreversible Decisions to triage what actually needs deep analysis. |
| Well-defined technical problems | If the problem is already precisely defined and the solution space is constrained by physics, regulation, or engineering requirements, the first three steps add overhead without insight. You don't need to reframe "the API is returning 500 errors" — you need to debug it. | 5 Whys or Ishikawa Diagram for root cause analysis of known problems. Issue Trees for decomposing well-defined challenges. |
| Convergence failure | The team excels at divergence — generating dozens of problem framings, success criteria, and ideas — but cannot converge. Every idea seems equally promising. Nobody wants to kill anyone else's contribution. The session ends with 40 ideas and no decision. Productive thinking becomes productive procrastination. | Use a Decision Matrix to force-rank ideas against criteria. Apply the ITCH criteria ruthlessly. Assign a single decision-maker for each convergence point. |
| Facilitation vacuum | Without a skilled facilitator, the diverge-converge rhythm collapses. Dominant voices hijack the brainstorm. The group evaluates ideas during generation (killing novelty) or generates ideas during evaluation (preventing closure). The six steps blur into an unstructured conversation with a fancy name. | Designate a facilitator who does not contribute content. Use silent writing for all divergent phases. Time-box each step explicitly. |
Reframing applied the Leverage mental model
Reframing applied the 5 Whys mental model
Reframing applied the Momentum mental model
Reframing applied the OODA Loop mental model
Reframing applied the Quality mental model
Reframing applied the Environment mental model