PromptKit
← All prompts
WorkflowBuild · ProductPro

Run a lightweight product experiment from hypothesis to decision

Most "experiments" never had a real hypothesis, which is why they never produce a real decision.

Outcome
A specific, falsifiable hypothesis, a minimum test to check it, and a decision rule you commit to before seeing results.
Time
30–60 minutes to plan, before you build anything

This workflow is part of PromptKit Pro

A handful of prompts and workflows stay free for everyone — Pro unlocks the full library, including this one, plus the guided runner.

Unlock with Pro →

Most "let's test it and see" plans skip the part that makes an experiment worth running: a hypothesis specific enough to fail, and a decision rule set before results arrive. Without that, any outcome gets rationalized into "promising, let's keep going," and the experiment produces activity instead of a decision. This playbook forces the hypothesis, the minimum test, and the decision rule into existence before anything gets built.

When not to use this

For something you're building regardless of outcome — a compliance requirement, an obvious fix — skip the ceremony and just build it. This is for genuinely uncertain bets where you could be wrong and want to find out cheaply.

Pro

The full step-by-step workflow is part of PromptKit Pro.

Unlock with Pro

Did this work?

You have a hypothesis specific enough to be proven wrong, a test cheap enough to actually run, and a decision rule you wrote down before seeing any results.

Tested on claude-opus-5. Evidence status is draft; it moves to battle-tested only on recorded runs, never by hand.

Consense — Big decision? Compare and combine multiple LLM responses.