Validate Your Idea

What this phase is about

Before you open Claude Code, before you scaffold a single file, you need one thing: a clear idea. Not a vague direction. Not "something like Airbnb but for dogs." A real, specific idea that you can explain in two sentences and defend under pressure.

This phase exists because half-baked ideas produce half-baked products. The build process is fast now. Claude Code can scaffold a working app in hours. That speed amplifies everything — including your mistakes. If you start with a fuzzy idea, you will ship a fuzzy product.

The goal of Phase 0 is simple: leave with a one-liner that explains what you are building, who it is for, and why they would pay for it.

Why this matters more than ever

AI-assisted development has removed most of the technical barriers to building. Anyone can ship an app. That means the differentiator is no longer "can you build it" — it is "should you build it, and does it solve a real problem for real people."

The teams that are winning with AI tools are not the ones who build the fastest. They are the ones who know exactly what they are building before they start.

The half-baked idea test

Answer these four questions honestly. If you cannot, your idea needs more work before you write a line of code.

1. What is the specific problem you are solving? (Not a category of problems — the exact problem.)
2. Who has this problem? Describe one real person, not a demographic.
3. Why does this problem exist today? Why has no one solved it well?
4. How will you know when you have solved it? What does success look like for your user?

If you got through all four, you have a real idea. If you got stuck, that is the work — figure out the answer before moving forward.

The one-liner format

Write your idea as a single sentence using this structure:

"We help [specific user] who [problem they face] by [what you do], unlike [current alternative] which [why the alternative falls short]."

Example: "We help solo consultants who lose track of client follow-ups by sending them a daily action digest, unlike CRMs which require manual data entry that no one actually does."

If your sentence is vague, abstract, or covers multiple user types, keep editing. You are not done until it is specific and honest.

Validate before you build

You do not need to run a formal research project. You need to talk to five people who have the problem you are solving. Not your friends. Not your family. People who are actually in the situation you are describing.

In each conversation, ask:

• Tell me about the last time you dealt with [the problem].
• What did you do about it?
• What was frustrating about that solution?
• How much time or money does this cost you?

You are listening for patterns. If three out of five people describe the same frustration in similar words, that is a signal. If everyone describes a different problem, your idea needs to narrow down.

What you need to have before moving on

Before you start Phase 1 (Idea to PRD), you should have:

• A one-liner that passes the format above
• At least three conversations with real people who have the problem
• A clear sense of what success looks like for your first user
• Confidence that you are solving a real problem, not a problem you invented

If you have all four, open the next doc. If you are missing any of them, that is your Phase 0 work. Do it now — it will save you weeks of building the wrong thing.

A note on perfectionism

Phase 0 is not about having a perfect idea. It is about having a clear one. The idea will evolve. The product will change. That is normal. What you are trying to avoid is starting the build without knowing who you are building for or why.

Clear is not the same as certain. You can be clear about your current hypothesis and still expect it to change. That is healthy. Vague is the problem, not uncertainty.