How to Validate an Idea:Set the Bar Before the Evidence
Validate an idea by setting the bar before you see the evidence: write it down and match it to what being wrong costs you in evenings. Kind words do not count.
To validate an idea, decide the bar before you collect the evidence, and write it down. An idea is validated when it clears the standard you set in advance, not when you feel confident. If you never stated the bar, it will quietly move to wherever your wishes already were.
How to validate an idea without fooling yourself
You already know the feeling. You had a few chats, people nodded, someone said "I'd use that". You went home buzzing. You called it validated.
It wasn't. It was a good mood.
To validate an idea is not to feel sure and not to take a vote. It is a comparison between what you found and a standard you fixed before you went looking. No standard, no validation. Just a story you told yourself with a few quotes attached.
Here is the part most advice skips. The danger is not a bar that is too low. The danger is a bar that doesn't exist. An unstated bar always moves to accommodate what you already wanted to do. You wanted to build it. Funny how the evidence turned out to agree.
Three things that look like proof and are not
Run your last "validated" idea through this. Be honest.
- Three friendly conversations. Friendly is the problem. People who like you are generous in a chat and stingy with their time and money.
- Your own opinion. You know the market. You have good instincts. Your instincts are also the thing under test.
- A poll where everyone said yes. Saying yes costs nothing. A poll measures politeness at scale.
None of these would survive a sceptical question from a stranger. "Would they have said that if it cost them something?" Most of the time, no.
So what counts? Behaviour. What someone did about the problem last month. What they tried. What they paid. The more it cost them, the stronger it is. A small commitment beats a large compliment every time.
Free guide · 21 pagesDiscovery and delivery, one person.The Dual-Track Playbook: how to keep discovering while you keep building.Get the free guide →Set the bar in both directions
Before the test, answer two questions in writing.
First: what result would make me commit evenings to building this? Second: what result would make me drop it?
Both. Not just the yes. If you only define success, every ambiguous result becomes a reason to carry on. Defining the no is what gives the test teeth. It's the same trick as deciding in advance what you'll do if the answer is negative: if you wouldn't actually stop, the test is theatre. Our piece on testing an idea cheaply goes into that side of it.
Then write it where you will see it. Not in your head. In your head the bar is made of rubber.
A good bar is specific enough that a stranger could read your notes and say whether you passed. "People seemed interested" fails that test. "Several people from the target group did the small thing I asked" passes it.
Match the bar to the cost of being wrong
Here is the relief. The bar does not have to be high. It has to fit the bet.
A small, reversible change needs very little. Try it, watch what happens, undo it if it flops. You don't need a research programme to move a button or reword a page. A month of building needs far more, because a month of evenings is the only budget you can't top up.
So scale the standard to what you stand to lose:
| The bet | What being wrong costs you | Evidence that is enough |
|---|---|---|
| A small change you can undo | An afternoon | A reasonable hunch and a look at the data you already have |
| A new offer or page | A few days and some credibility | People from the target group doing something small without being begged |
| A new feature or product line | Weeks of evenings | Past behaviour that proves the problem, plus a real commitment to the solution |
| Something that changes how the business runs | A quarter of focus | All of the above, repeated, from people who match the customer you want |
Notice what the table does not say about how to validate an idea. It never says "lots of positive feedback". Volume of nice words is not on the scale anywhere.
Ask whose evidence it is
One more filter. Who did you hear it from?
Whoever happened to be available is not a sample. Your newsletter's most loyal reader, your cousin, the person who replies to everything. They are easy to reach and easy to please. The customer you actually want may think very differently.
Count only people who match the target. If your evidence is thin on those, the honest verdict is "not yet", and that is a perfectly good answer. There is a whole post on asking questions that don't lead, because leading questions are the quietest way to fake a pass.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Write the rule for when you validate an idea
Validated means one thing in practice: this moves from discovery to building. So you need a rule for the move.
Without one, two things happen. Either everything graduates, and your discovery is a formality with extra steps. Or nothing does, and you pile up well-researched ideas that never ship.
Since you are both the researcher and the builder, you will be arguing with yourself. The version of you that wants to build usually wins. So write the rule while you're calm. Something like:
- The riskiest assumption has been tested, not the convenient one.
- The result cleared the bar I wrote down before the test.
- I can state what outcome this is expected to produce.
- I can say what I still don't know.
Four lines. If graduating an item takes longer than running a small test, the process has become the point. Keep it light.
And one rule for the other direction: nothing graduates just because the build queue is empty. An empty queue is a reason to do maintenance or a known improvement. It is never a reason to lower the bar.
Give yourself two permissions
You need to be allowed to say two sentences out loud.
"Not yet." That is a real result. It means the evidence is thin and you know what to test next.
"I know enough. Stop researching." That is also a real result. There is a hazard here: it always feels responsible to learn a little more. Timebox the question. At the end, decide with what you have, even if the decision is "I still don't know, and here is the one thing I'd test next". Both sentences stop a loop. One stops the building loop. The other stops the research loop. A solo operator needs both, because nobody else is there to say them.
Free guide · 21 pagesDiscovery and delivery, one person.The Dual-Track Playbook: how to keep discovering while you keep building.Get the free guide →What to do on Monday
Pick the idea you are most excited about right now. The one already half built in your head.
- Write what would make you commit evenings to it.
- Write what would make you drop it.
- Write who counts as evidence, and who doesn't.
- Choose the cheapest test that could move you. Our post on finding your riskiest assumption helps you pick the right one.
- Set a date to decide.
That takes half an hour. It will save you from the most expensive thing a one-person business can do: build the wrong thing beautifully. AI makes that easier than ever. It will happily help you ship a polished answer to a question nobody asked.
And if the idea fails the bar? Good. You've just got a month of evenings back. Record what you learned and move on. A killed idea is output, and killing ideas on purpose is a skill you can learn.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Frequently asked questions
How many conversations do I need before an idea is validated?
There is no magic count, and be wary of anyone who gives you one. What matters is who you spoke to and what they did, not how many. A few people from the target group who acted on the problem are worth more than a crowd of polite strangers.
Is a pre-order or a paid test the only real validation?
No, but a commitment is stronger than a compliment. Past behaviour, a small action, time given or money handed over all count. Which one you need depends on what you are risking. A small reversible change needs far less than a month of building.
What if the evidence is mixed?
Then you probably didn't set the bar clearly enough, or you tested the wrong assumption. Say "not yet", name the one thing you would test next, and give it a deadline. Mixed is not a licence to build.
Can I validate an idea by looking at data I already have?
Sometimes, yes. Looking at existing data is one of the cheapest tests available. It works best for small, reversible bets. For bigger ones, pair it with what real people have actually done.
Who decides it is validated if I work alone?
You do, and that is the risk. The fix is writing the bar down first, so the you who wants to build is held to a standard set by the calmer you. That is how you validate an idea honestly when you work alone.
The full method, with the checklists, is in The Dual-Track Playbook, a free 21-page playbook.
How trapped are you, really?
Take the Corporate Suffocation Index, the free solopreneur readiness assessment.