strategyplaybook.aiSqwyz
Sign in
← All 124 tools
Research · Diagnose

Initial hypothesis

The answer you would give today if forced to. Writing it down turns research from a trawl into a test, and it is uncomfortable precisely because it can be wrong.

The printable canvas and the handoff into Claude Code are part of a paid plan. See what a plan includes. It needs the Claude desktop app on this machine, and your team and the prompt library already installed in that project — we cannot see your disk, so open it there.

Stage
01 Diagnose
Works at
All levels
Maturity
Any
Time
Ninety minutes, before any research starts

What it is

A written, falsifiable statement of what the answer probably is, made before the work that tests it. It is not a prediction to be defended — it is a target for evidence to hit, and the value is entirely in how quickly it gets corrected.

The stated answer
One paragraph, specific enough to be wrong. "Margin is the problem" is not a hypothesis; "margin is falling because trade customers now cost twice what they did to serve" is.
The supporting logic
The two or three steps that get from what is known to the claim. Each step is a place the hypothesis can break, which is what makes them worth writing separately.
What would have to be true
The conditions the answer depends on, written as things that can be checked rather than things that can be believed.
The disconfirming test
The single cheapest piece of evidence that would kill it. If nobody can name one, the hypothesis is not falsifiable and is therefore not useful.
The date
Hypotheses are meant to change. Dating it makes the change visible, instead of letting the story quietly rewrite itself into one that was right all along.
The mistake people makeWriting a hypothesis so hedged that no evidence could contradict it. A statement that survives anything has told the team nothing and cost them a morning.
What it’s forBefore research starts, when the team knows enough to have a view and has not admitted to one.
What it’s not forYou genuinely know nothing about the business or the market. Read for a day, then come back.

How you run it

  1. Have everyone write it alone firstNinety seconds, no discussion. Spoken hypotheses converge on whoever speaks first; written ones show the actual spread of belief in the room.
  2. Make it specific enough to be wrongName the number, the segment or the mechanism. Vagueness is how a hypothesis survives a year of contradicting evidence.
  3. Write down what would have to be trueTwo or three conditions, each checkable. These become the first branches anybody actually researches.
  4. Name the cheapest way to kill itIf disconfirming it costs a quarter, the hypothesis is unusable as written. Find a cheaper test, or narrow the claim until one exists.
  5. Revisit it in writing, keeping the old versionThe trail from first guess to final answer is the most persuasive thing in the finished document, and it only exists if nobody overwrote it.

The prompt

Two ways to run it

Run this tool in your own Claude

The short prompt starts your partner against the library on your disk. The long one carries everything with it and needs nothing installed.

🔒 Prompt locked

Everything above is free — what this tool is, what it is for, what it is not for, and how to run it as a workshop. Three tool prompts a month are free with an account; beyond three, and for the printable canvas, it is the paid part.

See plans — from £19 a month

Your playbook

It lands in the earliest stage this tool suits. Move it on the Playbook page.

You’ll need

  • The issue tree, or at minimum the question
  • What the team already believes, said out loud
  • The evidence that exists today, however thin

You’ll end up with

  • A stated answer with its reasoning, dated
  • The two or three things that would have to be true for it to hold
  • The cheapest evidence that would kill it, and who is looking for that

Runs well with

Business Model CanvasCapability maturity assessmentCompetitor profilingComplementary assets — who captures the value