Turn a Broad Strategic Question into a Falsifiable Research Plan

Convert a vague strategic question into competing hypotheses, disconfirming tests, minimum-viable data sources, and pre-committed decision thresholds.

Prompt

Prompt
Act as a principal research methodologist and strategy advisor. Your job is to turn a broad strategic question into a falsifiable research plan that prevents confirmation bias and wasted analysis.

Before writing the plan, ask me 4 short numbered questions in one message and wait for my reply:
1. What is the broad strategic question or decision you face?
2. What is your deadline and team capacity for this research?
3. What is your default action if you do no further research?
4. What internal data or customer access do you already have?

Once I reply, follow these rules:
- Write three competing hypotheses, including a status-quo hypothesis.
- Define one concrete disconfirming test per hypothesis that would prove it wrong.
- List only minimum-viable data sources that fit my stated timeline and access.
- Set numeric, pre-committed decision thresholds tied to a specific action.
- Never invent statistics, benchmarks, or external citations.

Output format:## Decision Frame
## Competing Hypotheses and Disconfirming Tests (table: Hypothesis | Core Assumption | Disconfirming Test | Minimum-Viable Source)
## Pre-Committed Decision Thresholds## Execution Sequence

End with one line naming the biggest assumption or missing input that could invalidate this plan.

Example output

At a glance3-week plan to test the self-serve tierDecision thresholds are set before any data comes in
  1. Run smoke test and log downgrade requestsWeek 1Launch paid search test; reps log self-serve downgrade asks
  2. Onboard 15 small-fleet pilot usersWeek 2Onboard manually and track every support touchpoint
  3. Audit results against thresholdsWeek 3Check funnel, CRM and Zendesk data, then finalize the roadmap vote

Example for: a 40-person B2B logistics software company deciding whether to launch a self-serve tier for small fleets within a 3-week research window.

Decision Frame

Decide whether to fund a 6-month engineering build for a self-serve SMB tier ($99/month) or keep all product capacity on mid-market enterprise accounts ($1,800/month average). Default action without new evidence: stay focused on mid-market.

Competing Hypotheses and Disconfirming Tests

| Hypothesis | Core Assumption | Disconfirming Test | Minimum-Viable Source |
| --- | --- | --- | --- | |
| H1: Self-serve unlocks profitable net-new demand | Small fleets (5-20 trucks) will convert without sales calls | Fewer than 8% of landing-page visitors from targeted ads start a paid trial or book setup | 2-week paid search smoke test ($1,500 budget) + analytics funnel |
| H2: Self-serve cannibalizes mid-market pipeline | Existing mid-market prospects downgrade to avoid annual contracts | More than 15% of active mid-market deals ask to switch to the $99 tier during discovery | Structured sales-call log across 25 live deals + CRM win/loss notes |
| H3: Status quo wins due to support drag | Small fleets require high onboarding hand-holding that breaks unit economics | Median onboarding tickets per pilot user exceed 4 in the first 14 days | Concierge pilot with 15 small-fleet signups tracked in Zendesk |

Pre-Committed Decision Thresholds

  • Go (build self-serve): Smoke-test paid intent >= 10%, cannibalization signal <= 10% of mid-market deals, and concierge pilot requires <= 2 support tickets per account.
  • No-Go (stay mid-market): Paid intent < 6% OR support tickets > 4 per account in the first 14 days.
  • Revise price/packaging: Paid intent >= 10% but cannibalization > 15%; gate features by fleet size before building.

Execution Sequence

  1. Week 1: Launch paid search smoke test and brief sales reps to log self-serve downgrade requests.
  2. Week 2: Onboard 15 small-fleet respondents manually and track every support touchpoint.
  3. Week 3: Audit funnel conversion, CRM logs, and Zendesk ticket counts against the pre-committed thresholds and finalize the roadmap vote.

Key risk: This plan assumes two weeks of search traffic reflects year-round fleet buying cycles and excludes seasonal freight slowdowns.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Requiring a status-quo hypothesis and explicit disconfirming tests prevents teams from cherry-picking data that confirms a pet idea.
  • Pre-committed numeric thresholds stop stakeholders from moving the goalposts after the research results come back.
  • Constraining sources to the user's real timeline and data access produces an executable plan instead of an academic wish list.

Use this before commissioning market research or a strategy memo in Claude, Gemini, Perplexity, or ChatGPT. Have your rough strategic question, deadline, and available internal data ready to answer the intake questions.

Common edits

  • Add a rule to score each hypothesis by cost of being wrong if you are weighing high-stakes capital allocation decisions.
  • Ask Perplexity or Gemini with web search enabled to suggest public industry datasets alongside your internal data sources.

Make it repeatable

Keep using this prompt without pasting it in every time.

Keep exploring

More in AI Research