You're Short on Evidence, Not Ideas

You have plenty of ideas. You’ve never been short on ideas.

What you’re short on is evidence, and the two feel similar from the inside because both arrive as a kind of certainty. The idea feels true. The conviction that people want it feels true. Neither has been outside your own head.

So you do the thing that feels like checking: you build it. You tell yourself you’ll find out once it’s real, once people can actually use it. That’s the most expensive way to ask a question that a twenty-minute conversation could have answered for free.

Why building feels like validating

Building is the only test a technical founder fully trusts, because it’s the only one they’re sure they can run. You can’t make a stranger want your product, but you can make the product. So “will anyone want this?” quietly gets swapped for “can I build this?”, and the second question gets a confident yes, which your brain banks as progress on the first.

It isn’t. You proved you can build it. You already knew that. The build was never the risky part.

And it’s getting less risky every month. When standing up a working product takes a weekend, “just build it and see” sounds almost reasonable: the cost looks low. But the cost was never mostly the build hours. It’s the weeks you spend believing a thing that a single hard conversation would have killed, and the company you don’t start because you’re busy maintaining the one you already know won’t work.

The load-bearing belief

Every plan rests on one belief that, if it’s false, the whole thing falls over. Find it. Then find the cheapest thing that could tell you it’s false.

Most plans have several assumptions, and founders instinctively test the comfortable ones. Can I build it? Yes. Is the market big? Sure, look at this report. Could it work technically? Obviously. All true, all irrelevant, because none of them is the load-bearing one.

The load-bearing belief is almost always about behaviour, not capability: that a specific person will change what they currently do, pay what you’ll need to charge, and keep doing it. That’s the one you skip, because the only way to test it is to go talk to that person and risk hearing no.

So run it deliberately. Write down the single belief that, if wrong, makes everything else pointless. Then ask: what’s the cheapest thing that would prove me wrong? If the cheapest test you can come up with is “build it and launch,” you haven’t found the real assumption yet. Keep going until you land on one a conversation, a pre-sale, or a fake door could puncture in a week.

A test that can’t fail isn’t validation. It’s reassurance with extra steps.

What this looks like

Think of an engineer two months into building a scheduling tool for allied-health clinics, full of conviction because three friends in the industry said “oh, that’s a good idea.” Those got heard as evidence. They were politeness.

The one that mattered wasn’t “is this a good idea”. It was “a clinic owner will switch off the booking system their whole front desk already knows, for mine.” The cheapest test of that wasn’t a product. It was asking five clinic owners to show their current setup and say, specifically, what would make them rip it out. That’s a week’s work. Four of them, it turns out, hate their system but won’t change it mid-financial-year, and the fifth just signed a two-year contract. That’s not a no, but it completely changes when and how you’d have to sell, and you learned it for the price of five coffees instead of another four months.

This is sharper in a smaller market. A founder in the US can ship something mediocre into an enormous addressable market and get bailed out by sheer volume: enough strangers stumble in that a weak assumption survives. In Australia your niche is smaller and your channels are thinner. You don’t get rescued by volume. You have to be right about the behaviour earlier, because there isn’t a long tail of strangers to hide a wrong guess inside.

What to do Monday

Don’t open the editor. Write one sentence: the belief this whole thing depends on, phrased so it could actually be false. Then list three ways to test it that don’t involve building anything, and do the cheapest one this week: a message to a real potential buyer, a request to see how they do it now, an offer to pre-sell.

You’re allowed to be wrong. That’s the entire point of the exercise: to be wrong now, cheaply, instead of later, expensively. The founders who make it aren’t the ones with better ideas. They’re the ones who found out faster.

Next: once you know what to build, the trap flips. You build too much of it.