01
Scope
- Define the question, the audience, and what is out of scope.
← toolsDigitalCoa.ch
Methodology
STORM
Sequential Thinking for Organized Research Method.
STORM breaks research into stages so the AI does not guess from a single giant prompt. It is built for layered thinking, contradictions, and structured synthesis.
STORM workflow
Stage 1
Scope
Stage 2
Perspectives
Stage 3
Sources
Stage 4
Contradictions
Stage 5
Outline
Stage 6
Research
Stage 7
Synthesis
Stage 8
Review
Purpose
- Break research into manageable stages.
- Force perspective before synthesis.
- Turn messy research into an organized workflow.
When to use
- Deep research and consulting tasks.
- Market analysis, policy analysis, and strategic reviews.
- Cases where multiple viewpoints need to be compared.
When not to use
- Simple factual questions.
- Translation or basic rewriting.
- Short brainstorming tasks.
Decision tree
Use STORM when the work is too complex for one prompt and needs staged reasoning.
- If the topic is broad, scope first.
- If there are conflicting perspectives, map them.
- If synthesis is too early, keep researching stage by stage.
Process
02
Perspectives
- Generate viewpoints before asking for answers.
03
Sources
- Ask which sources should answer the question and classify them by strength.
04
Contradictions
- Find where experts disagree and what remains uncertain.
05
Outline
- Build the structure before writing the full report.
06
Research
- Work section by section instead of all at once.
07
Synthesis
- Combine facts, insights, risks, limitations, and recommendations.
08
Review
- Run the result through PERCH before making a decision.
AI prompts
Prompt 1
Follow STORM exactly: scope the problem, generate perspectives, identify sources, map contradictions, outline the research, and only then synthesize.
Prompt 2
Never write the final synthesis until you have explored contradictions and missing evidence.
Examples
Example
Researching whether GBS should adopt AI across admissions becomes a staged process rather than a single broad prompt.
Common mistakes
- Asking one giant question and hoping the model guesses well.
- Jumping to synthesis before contradictions are mapped.
- Skipping the review stage.
Downloadable checklist
- Question scoped.
- Perspectives mapped.
- Sources identified.
- Contradictions mapped.
- Outline built.
- Review completed with PERCH.