How to pressure-test a product idea in a week (without writing code)

Building has never been cheaper or faster. Which makes skipping validation more tempting than ever, and more dangerous. 

35–42% of startups fail because nobody wanted what they built. And when it comes down to it, the issue is always that as founders, we make assumptions that we don’t end up testing.

Here’s how to test those product assumptions in less than a week ⬇️

Day 1: Figure out what you're betting on

Every product idea is held together by assumptions you haven't exposed yet. 

• Who the customer is
• How much the problem hurts
• Whether they'd pay for a solution
• Whether you can deliver one

Write them all down. Then ask, for each one: if this turns out to be wrong, does the whole idea fall apart? Test the ones where the answer is yes first. 

Days 3: Talk to strangers

Not friends, or your co-founder network. People who have no reason to be nice to you. Set up ten to fifteen conversations with people who match your ICP. 

The goal isn't to pitch them, it's to find out whether the problem you're solving is real to them, and how real. 

❌ Don’t open with "would you use something like this?" That’s bound to get a yes from everyone. 

✅ Try: "Can you walk me through the last time that happened?"

Listen for whether they've already tried to fix it. Someone who's burned through workarounds and is still frustrated is worth more than a hundred people who say your idea sounds cool. Behaviour tells you more than opinions ever will.

Also, ask what they currently pay to deal with the problem: in money, time, or hassle. That's your pricing anchor.

Days 4: Build a fake door

A one-page site, a clear headline, or a rough visual of what the product looks like. One clear button: "Get early access" or "Pre-order."

Then put $200–500 behind it in ads to your exact target customer. Cold traffic only, because people who know you aren't a real signal. 

Above 15% conversion on the CTA is worth paying attention to. Below that, either the problem isn't resonating or the message isn't.

While you're at it, test three price points. If everyone gravitates to the cheapest option, you're probably underpriced. If no one clicks anything, the value isn't coming through.

Days 6: Do it manually

Before you build the thing, be the thing. 

If you're building an AI research tool, do the research yourself for your first five users. 

If it's a data pipeline, do it in spreadsheets. 

A week of manually delivering the outcome teaches you more about what people need than months of building toward what you assumed they need.

Charge something, even if it's small. Free gets you interest. Money gets you signal.

Day 7: Read what people did, not what they said

Waitlist signups don't tell you much. What tells you something: someone tried to pay. Someone followed up without prompting. Someone introduced you to a colleague with the same problem.

Before you run any of this, write one sentence: "If we don't get [specific action] by [date], we stop." Then hold yourself to it. Your brain will manufacture reasons to keep going when the data says otherwise, and past-you is the only person who can protect future-you from it.