Validation Planner

Overview

This skill turns a chosen opportunity into a real demand test. The goal is not to produce activity. The goal is to create a short validation plan with numeric targets that can prove, narrow, or kill the idea quickly.

The plan should focus on the single biggest unknown, not on validating everything at once.

Load references/experiment-patterns.md when selecting the right validation motion. Load references/validation-metrics.md when defining thresholds.

When to use

• When a user already has a best bet or chosen opportunity
• When an idea needs validation before building, scaling, or quitting a path
• When a service-first, audience-first, or product-first wedge needs concrete testing steps

Do NOT use when:

• The user still needs help choosing the opportunity itself
• The problem is purely implementation and demand is already proven

Workflow

1. Identify the real unknowns

Name the key risks behind the idea, such as:

• demand risk
• buyer-access risk
• willingness-to-pay risk
• workflow-repeatability risk
• legal-scope risk

Then choose the single biggest unknown. Name it explicitly and explain why it matters more than the others right now.

2. Choose the right validation motion

Pick a test that matches the opportunity:

• Warm outreach and paid pilot
• Marketplace or directory test
• Audience pre-sell or waitlist
• Async product test
• Constraint-change plan for no_go opportunities

Use the strongest available commitment signal the user can realistically ask for, such as:

• deposit
• paid pilot
• pre-order
• signed implementation slot
• completed checkout

Do not stop at interviews, clicks, opens, or waitlist signups when a stronger signal is feasible.

3. Define numeric success and failure criteria

Every plan should include numbers where possible:

• outreach count
• interview count
• conversion target
• payment target
• activation target
• stop or pivot threshold

Use decision-grade thresholds:

• go
• revise
• kill

Briefly justify the numbers based on channel quality, audience size, price point, access level, or sales cycle. Avoid fake precision.

4. Produce the plan

Return both:

• a readable validation plan
• a JSON block using assets/validation-plan-template.json

Default sections:

• Validation Thesis
• Key Unknowns
• 7-Day Plan
• Success Criteria
• Kill Criteria
• Structured Plan

Also include a brief Why Not The Lower-Signal Alternative explanation when there is an obvious weaker motion, such as interviews instead of paid pilots.

Checklist

• The plan is testing the biggest unknown, not generic activity
• Numeric success and failure thresholds are explicit
• The motion matches the opportunity and the user's constraints
• The strongest feasible commitment signal is used
• The plan isolates one primary unknown instead of many
• The response includes readable plan and structured JSON

Examples

Example:

Input: Best bet is an HVAC missed-call recovery service with warm owner access.

Output: A 7-day plan with owner interview counts, pilot target counts, an offer statement, and kill criteria tied to urgency and willingness to pay.

Common mistakes

Mistake	Fix
Writing generic activity lists	Tie the plan to a specific unknown and numeric target
Validating everything at once	Choose the single biggest risk first
No stop condition	Add clear pivot or kill thresholds
Using weak signals when stronger ones are feasible	Ask for money, commitments, or defined next steps

Quick reference

Operation	How
Pick motion	Match the test to the opportunity and buyer access
Set targets	Use numeric thresholds from references/validation-metrics.md and justify them briefly
Format result	Human-readable plan plus JSON using assets/validation-plan-template.json

Key principles

1. Test the real risk — Validate the biggest uncertainty, not the easiest activity.
2. Numbers beat vibes — Use thresholds, counts, and time windows.
3. Use the strongest available signal — If money or commitment can be asked for, do not hide behind soft interest metrics.
4. Kill fast when needed — A weak idea should be narrowed or stopped, not endlessly studied.