Skip to content

Question Burst

Stuck thinking is usually stuck on the wrong question. A question burst generates many questions about a problem in a short, constrained burst (questions only, no answers), to break attachment to the current framing, then ranks them and picks the single most catalytic one. Because a model can generate questions endlessly, the value here is not the generation, it is the ranking and selection: this skill produces a ranked set ending in one chosen next question, never a bulk dump. The output is that ranked question set.

  • Stuck, or over-attached to a single framing of the problem.
  • At the start of exploring an ambiguous problem, before committing to an answer.
  • When a better question would unlock more than another answer.
  • To produce a bulk list of questions with no ranking or selection (low signal; the main failure mode for an AI).
  • When the issue needs answers and convergence, not more questions.
  • When the catalytic question is already known.

When asked to run a question burst, follow these steps:

  1. State the problem in one line.
  2. Burst. Generate roughly 12 to 20 questions about it. Questions only, no answers, no preamble. Mix angles: why, how, what-if, who, what-would-change-if. Keep it brief.
  3. Rank. Order the questions by how much answering them would change the approach, not by how easy they are.
  4. Select. Choose the single most catalytic “next question” and give a one-line reason it would shift the problem.
  5. Emit the ranked question set per references/TEMPLATE.md.

Use the template in references/TEMPLATE.md. The deliverable is the ranked questions plus the one chosen next question, not a flat list and not answers.

Before finalizing, verify:

  • The burst was questions only, no answers.
  • The questions are ranked by catalytic potential, not ease.
  • Exactly one “next question” is selected with a reason.
  • The output curates, it does not just dump a long list.
  • The output is the ranked question set artifact.

Tier P. The method is Hal Gregersen’s question burst (MIT Sloan): generate many questions under a strict questions-only rule, then find the catalytic ones. MIT Sloan reports participant benefits (broader view, recognizing one’s own role); there is no controlled decision-outcome evidence, and for AI the generation half has little value, so this skill is built around curation. Evidence is transferred from human workshops, not AI-validated. Full grading: evidence/dossier.md.

See references/EXAMPLE.md for a completed ranked question set.

A full worked run (the shared Northwind scenario)

A completed run of think-question-burst, on the shared Northwind scenario. This is the quality bar a generated set should meet.

Northwind is a B2B SaaS weighing a self-serve free-tier launch and feeling stuck on “build it or not.”


  • We are stuck debating whether to build a free tier to hit the Q3 growth target.
  • Why do we believe a free tier drives growth here specifically?
  • What is our current trial-to-paid conversion, and why is it falling?
  • Who are the users we actually want more of?
  • What would have to be true for a free tier to pay for itself?
  • What is the cheapest experiment that would tell us if free works?
  • Who loses if we launch a free tier?
  • What problem did the free tier idea originally come from?
  • Is “growth” the goal, or is it revenue, or qualified pipeline?
  • What happens to Sales the day we launch free?
  • How would a competitor without a free tier hit this target?
  • What would we do if we could not build anything new for Q3?
  • Are we solving a packaging problem or a funnel problem?
  • What is the most irreversible part of this decision?
  • Who has hit this target before, and how?
RankQuestionWhy it would change the approach
1Are we solving a packaging problem or a funnel problem?If it is the funnel, the entire free-tier premise is wrong and the fix is cheaper.
2What is the cheapest experiment that would tell us if free works?Converts a one-way-door build into a reversible test.
3Is the goal growth, revenue, or qualified pipeline?The real objective changes which option even qualifies.
  • “Are we solving a packaging problem or a funnel problem?” - it is upstream of everything else: answering it either kills or confirms the free-tier hypothesis before any build, and it is answerable from data we already have.

Note: the value is the ranking. A model can list these 14 questions instantly; the work was deciding that question 1 reframes the whole debate, and feeding it into a problem-restatement or evidence sort next.

What the research does and does not show, with graded sources

Single source of truth for the question-burst skill. The SKILL.md, sidecar, and evals derive from this.

Skillthinking-framework-skills.question-burst (installable name think-question-burst)
Familydivergent-ideation
Evidence tierP (practitioner; MIT Sloan reports participant benefits)
ConfidenceModerate that questioning shifts framing; for AI the value is curation, not generation
Statusdraft (authored 2026-05-31 from the discovery corpus)

1. The mechanism (what actually does the work)

Section titled “1. The mechanism (what actually does the work)”

Stuck thinking is often stuck on the wrong question. A question burst generates many questions about a problem in a short, constrained burst - questions only, no answers, no preamble - to break attachment to the current framing, then ranks them for which would most change the approach and picks the single most catalytic one to pursue. The discipline (questions only, a quota, a time box) suppresses the reflex to answer prematurely.

Critical adaptation for AI: a model can produce hundreds of questions instantly, so raw generation is worthless here. The value is entirely in the ranking and selection - identifying the few questions that would actually shift the problem. This skill therefore requires a ranked output and one chosen next question, not a bulk dump.

  • Hal Gregersen (MIT Sloan), “Better Brainstorming” / the question-burst method: generate at least ~15-20 questions in a few minutes under a strict questions-only rule, then study them for the catalytic ones.

No trademark. Named descriptively.

3. What the evidence shows, and what it does NOT show

Section titled “3. What the evidence shows, and what it does NOT show”

Supported (practitioner): MIT Sloan reports that participants who run a question burst commonly leave with a better emotional state, a broader view of the problem, or the recognition that they are themselves part of the issue. The questions-only constraint is the active ingredient.

NOT shown: no controlled evidence that it improves decision outcomes. And for AI specifically, the generation half has near-zero value (the well-known critique: LLMs generate questions trivially; the challenge is curation). Grade P, and design the skill around the curation, not the volume.

Evidence is from human workshop contexts, not AI-augmented use. Transferred, not AI-validated. The honest AI value is narrow but real: forcing a ranked, selected output (not a bulk list) turns cheap question generation into a genuine reframing aid.

Works best when: stuck, over-attached to one framing, or at the very start of exploring an ambiguous problem; when a better question is needed before any answer.

Fails or misleads when (poor-fit / anti-patterns):

  • Used to dump a bulk list of questions with no ranking or selection (the central AI failure mode; low signal-to-noise).
  • Answering instead of questioning during the burst.
  • When the issue needs answers and convergence, not more questions.
  • When the catalytic question is already known.

A ranked question set: the raw burst (kept brief), then the questions ranked by how much they would change the approach, and the single chosen “next question” to pursue with a one-line reason.

  1. Gregersen, H. (MIT Sloan), “Better Brainstorming” (HBR) and the Question Burst method.

Verification status: Gregersen/MIT Sloan attribution is well-attested. Treat participant benefits as practitioner-reported, not a measured decision-quality effect.

Thinking Framework Skills v0.3.0 · 38 frameworks