How I pressure-test an idea before building it

The most expensive thing a founder can do is build the right product for a problem nobody has. It looks like progress the whole way down — there are designs, sprints, a demo, maybe a launch — and only at the end does the silence tell you the demand was never real. So before I commit to building anything, mine or a client's, I try to kill the idea on purpose, as cheaply as possible.

Start from the problem, never the solution

If an idea opens with the technology — "we use AI to…" — I get nervous. Technology is a how, and a how can't tell you whether anyone cares. I make myself write the idea the other way around: [a specific person] struggles to [do something], and today they cope by [current workaround], which is bad because [real consequence].

If the honest workaround is "nothing, it's not a big deal," I've usually found a vitamin, not a painkiller. Vitamins are lovely and nobody switches for them. The products that win solve something people were already paying for — in money, time, or frustration.

Find the assumption that would be fatal

Every idea is a tower of assumptions, and one of them is load-bearing. The skill is finding it before your bank balance does. I sort assumptions into three piles:

• Do they want it? (desirability — usually the riskiest)
• Can it make money and reach the people who'd pay? (viability)
• Can it actually be built to work reliably? (feasibility)

Then, for each, I ask a blunt question: if this were false, is the product dead? The assumption where the answer is "yes" and I'm least sure — that's the one to test first. Everything else can wait, because if the load-bearing assumption fails, the rest was wasted effort anyway.

Test the assumption, not the product

Here's the move most people skip: you can almost always test the risky assumption without building the product. If the risk is do people want this, you don't need software — you need conversations, a landing page, a waitlist, a fake door, an offer someone can say yes or no to. If the risk is can this be built reliably, you prototype the single hardest piece, not the whole system.

The goal of each test is a clear signal, cheaply bought. I'd rather spend a week and a small budget learning the idea is wrong than spend six months building a beautiful monument to it.

Watch what people do, discount what they say

People are kind. Ask if they'd use your thing and they'll say yes to spare your feelings. So I weight behavior over opinion: did they give me their email, their time, a small payment, a spot on a waitlist? A signup is worth a hundred "sounds great"s. Enthusiasm is free; commitment costs something, and only the things that cost something predict the future.

The point isn't to be right — it's to be wrong quickly

Validation done well feels a little disappointing, because its whole purpose is to expose weakness while weakness is still cheap. The founders who last aren't the ones who avoid being wrong. They're the ones who arrange to be wrong fast, in small ways, before the stakes get real.

If you've got an idea and you want a hard, friendly second opinion on where it's fragile, send it my way — I do this all day.