Why you can't validate your startup idea, and what to do instead

In this post, Charles Hope – four-time founder, startup advisor and mentor to more than 700 founders – challenges one of the most persistent myths in startup building.

You might have met Charles as one of the speakers at Founder Growth Rooms, the SeedLegals mini accelerator where founders pressure-test their strategy, sharpen their pitch and get ready to win over investors.

Here, he cuts through the compliments, surveys and waitlists to reveal what meaningful validation actually looks like. And how three conversations could save you months of building the wrong thing.

Founders tell me all the time that they have validated their idea. Then they show me the evidence.

The evidence is usually a compliment in a nice outfit.

I am not being unkind. I have done exactly the same thing myself. I once walked out of a room glowing, certain I had proof. I had no proof. I had a lovely afternoon.

So let us look at what validation usually means. Then let us look at what actually works.

What founders usually mean by “validated”

The evidence normally arrives in one of 4 shapes.

30 people saw it and loved it. 200 people joined a waitlist. A survey came back 80% positive. Or an experienced mentor said it was a great idea.

Every one of those feels like progress. That is exactly the problem.

Now look at what they have in common. Not one of them describes something that already happened. Each one predicts a future that has not arrived.

“Loved it” is a feeling. “Would use it” is a forecast. A waitlist signup costs 4 seconds and no money.

So you have collected opinions about a hypothetical. That is not nothing. It is simply not evidence.

Why a yes tells you so little

There are a few reasons, and only one of them is anybody’s fault.

First, people are kind. You sat in front of someone, visibly hopeful, and asked what they thought. That is not a research question. You asked them to choose between honesty and your feelings.

Most people choose your feelings. They are being polite, not dishonest. Unfortunately, politeness sounds exactly like validation.

Second, the question usually contains its own answer. “Wouldn’t it be useful if this happened automatically?” Well, obviously. Nobody in recorded history has said no to that.

Third, saying yes costs them nothing. Free signals are worth roughly what you paid for them.

Then there is the awkward reason. You went looking for a yes.

You are invested. You have told your friends. You may have left a job for this. So you asked in a way that made yes easy, you heard it, and you stopped digging.

That is confirmation bias. It does not care how clever you are. Mine certainly does not.

You can’t validate an idea. You can only validate a problem.

Here is the reframe that changes everything.

Your idea is your guess at a solution. It lives in your head. Until you build it, there is no fact out in the world to check it against.

The problem is different. The problem is already happening.

Right now, somebody is losing an hour a week to it. Somebody else is paying badly to work around it. All of that is true whether you exist or not.

Facts can be checked. Guesses can only be admired.

This changes the job in front of you. To test an idea, you must describe it. Describing it means selling it, and selling it spoils every answer you get.

To test a problem, you never mention your idea at all. Not once.

Instead, you go and find out what is already true. Honestly, that is the easier job. It is also far less flattering, which is probably why fewer founders do it.

The rule that protects every conversation

Never talk about your idea, your product, or your solution.

I know what I am asking here. I am telling a founder not to mention their startup. It is a bit like asking someone not to mention their new puppy.

But the moment you describe your solution, the conversation changes shape. They stop telling you about their life. They start reacting to yours.

After that, everything you hear is a response to your pitch. Later, you cannot separate the clean answers from the contaminated ones. In your notes, it all looks the same.

They will ask what you are building. So have a short line ready.

Try something like this. “I am looking at how people handle this. I am still early, so I would rather hear how it works for you first.”

Then ask your next question and move on. Almost everybody lets it go. Learn first, share afterwards.

Ask about the last time, not the next time

We are all unreliable narrators of our own future.

Not dishonest, just human. Ask me in January how often I will visit the gym in February. My answer will be sincere and completely wrong.

The past behaves much better. The past already happened, so it leaves evidence.

There are calendar entries, spreadsheets, invoices and slightly annoyed messages to colleagues. You can check those. So can they.

The 3 questions that do most of the work

Once you are in the conversation, 3 questions carry most of the weight.

• Tell me about the last time this happened.
• What did you do about it?
• What did that cost you, in time, money, or something you gave up?

Everything else is follow-up. Afterwards, ask why gently, 2 or 3 times. The first answer is the tidy one. The third is usually the true one.

Want the longer version? I wrote a guide on how to validate your idea properly.

How to read the answers

Every answer helps, even the disappointing ones.

If they cannot remember the last time, the problem is not memorable. That is an answer.

If they remember it clearly but did nothing, they are living with it comfortably. Also an answer, just a quieter one.

But if they built something ugly at 11 at night to cope, pay attention. You have found something real.

People don’t build workarounds for problems they don’t have. A workaround is somebody voting with their own evening. That beats a hundred survey responses.

Where to start this week

You do not need a research plan. You need 3 conversations.

First, write down who has the problem. Use one sentence you could paste into a search bar.

Describe behaviour and situation, not demographics. “Small business owners” is not a customer. “People who send 5 invoices a month and chase the late ones themselves” is a customer.

Next, find 5 of those people. Then book 3 conversations.

When you get there, ask about the last time. Ask what they did. Ask what it cost them.

Do not show them anything. Do not pitch. Just find out what is already true.

You might learn the problem is real and painful, and then you get to build. You might learn it belongs to somebody else entirely, and you adjust.

Or you might learn that nobody is bleeding. That is not failure.

That is the cheapest good news you will ever receive. Finding out over 3 conversations beats finding out over 18 months and your savings.