Validating the Idea: Reducing Uncertainty, Not Proving You're Right

"Validate your idea" is advice everyone gives and almost nobody explains.

It sounds like a test you pass. Do some interviews, build a landing page, get a green light, start building. That's not how it works, and thinking of it that way leads people to fool themselves.

What this means

Validation means reducing uncertainty. It doesn't mean proving your idea will succeed.

Nothing proves that in advance. What you can do is take your biggest open question, find the cheapest way to get real evidence about it, and see whether that evidence makes you more or less confident.

It's the same experiment habit from Think Like a Builder, applied to the most important question: does anyone want this?

What I learned

The most useful thing I've learned about validation is that evidence has different weights. Not all "yes" answers are equal.

Roughly, from weakest to strongest:

1. Compliments. "Cool idea!" costs the person nothing, so it tells you almost nothing.
2. Predictions. "I'd definitely use that." People are bad at predicting what they'll do.
3. Past behavior. "Last month I spent an hour on Pinterest looking for haircuts." This one is real, because it already happened.
4. Small commitments. An email, a waitlist signup, time spent in a prototype.
5. Money. A pre-order, a deposit, a purchase. The strongest signal there is.

Most people validate at levels 1 and 2 and then feel surprised when nobody buys. The evidence was weak the whole time. It just felt strong because it was positive.

Real example

HairOver is a useful example because it's already past the "idea" stage. That lets me show what each kind of evidence looks like for a real product, and which ones I'm still missing.

What I have:

• Past behavior around the problem. People search for things like "hairstyles for a round face" and "best haircuts for a receding hairline" before haircuts. That's why the HairOver website is built around guides for exactly those searches. Searches show the problem is real. They don't show people will pay for my solution.
• Existing competitors. There are many AI hairstyle apps on the Play Store. That's weak but real evidence of demand. Companies don't keep building products nobody downloads.

What I'm still missing:

• The strongest evidence of all: people paying. The app is live and the pricing is planned out, but "will strangers pay for this?" is the question I'm still working to answer. Here's the pricing math.

I find this list more useful than a "we validated it" story. It shows the idea has some evidence behind it, and it shows exactly which uncertainty is still open.

The main validation tools, simply

Conversations. Talk to people who have the problem. Ask about the past, not the future. "What did you do last time?" beats "would you use this?" every time.

Landing pages. Describe the product, make one promise, and ask for one action: an email, a signup, a pre-order. What you're measuring is whether strangers take the action, not how nice the page looks.

Waitlists. Useful, but weak on their own. Signing up costs nothing. A waitlist becomes real evidence when you contact those people later and see how many actually show up.

Prototypes. A rough version people can click through or use. You're watching for whether they get what it's for and whether they want to keep going.

Pre-sales. Asking for money before the product fully exists. The strongest test, and the one that feels the most uncomfortable to ask for.

Small live experiments. A post, a short video, a small ad. Measure clicks or signups, not likes.

Fake demand vs real demand

Fake demand feels great. Lots of likes, lots of "I need this!", friends saying it's brilliant.

Real demand is quieter. A stranger signs up without you asking twice. Someone asks when they can pay. Someone uses the rough version three times in a week.

A simple test I use: would this person be disappointed if the product disappeared? If yes, that's demand. If they'd shrug, it was interest.

When to continue, and when to change direction

Decide this before you run the test. Write it down:

• "If fewer than X people out of Y sign up, I'll change the promise."
• "If nobody I talk to has tried to solve this before, I'll look for a different problem."

Then, when the result comes in, follow your own rule. The biggest risk isn't a failed test. It's a failed test you explain away because you like the idea.

Changing direction doesn't mean giving up, either. Usually it means keeping the problem and changing something smaller: the audience, the promise, or the price.

What to do

1. Name your biggest uncertainty in one sentence.
2. Pick the strongest evidence you can realistically get for it this week. Aim higher than compliments.
3. Write your continue/change rule down before you start.
4. Talk about the past, not the future, in every conversation.
5. Ask for a real commitment: an email, time, or money.

Mistakes to avoid

Counting compliments as validation. They feel good and tell you almost nothing.

Asking "would you use this?" You'll get polite yeses.

Building a waitlist and never testing it. Contact those people. See who's still interested.

Treating one good signal as proof. Validation reduces uncertainty. It doesn't remove it.

Running a test without a stop rule. Then any result can be read as success.

My current thinking

For HairOver, the problem side feels well supported. People clearly search for this, and plenty of apps exist in the space. The open question is the one that matters most for a business: will enough people pay, and keep paying?

I don't have that answer yet. The next few chapters on business models and first customers are, in a way, about how I'm trying to find it.

Key takeaway

Validation is about reducing uncertainty, not proving you're right. Rank your evidence (money beats commitments beats past behavior beats compliments), decide in advance what result would change your mind, and follow that rule when the answer comes in.

Next: Starting Before You Feel Ready.