The Indie Hacker's Guide to Validating a Product Before You Build It

Most validation advice is either too slow or too expensive for a solo maker. Interview fifty strangers before you write a line of code. Run a concierge MVP for six weeks. Meanwhile you have a day job and the idea is burning a hole in your head. Here is the ladder I would climb in 2026: four rungs, cheapest first, and you only spend real money on humans once the cheap tests have nothing left to teach you.

Rung 1: Write the landing page first

Not after the product, before it. If you cannot explain the value prop on one page, the product will not fix that for you. Then run the 5-second test on it: does a stranger understand what this is, and who it is for, within five seconds of the page loading?

The cheapest way to check is the free roast on IndieCru.sh. Paste your URL, no account needed, and three AI personas review the page in about 90 seconds. They score it and name the top issues: vague headline, buried call to action, a feature list where the promise should be. Rewrite until the roast stops finding obvious problems. This rung costs zero dollars and kills more bad ideas than any other.

Rung 2: Build the smallest prototype with a real signup

Not a clickable Figma demo. A live page where someone can actually create an account and do the one thing your product exists to do. The signup flow matters because that is where abandonment hides. Fake-door tests measure clicks, and clicks are cheap lies. A real flow measures follow-through.

• One core action, done end to end. Everything else is stubbed or handled manually behind the scenes.
• A real signup with the real fields you plan to ask for. Every extra field is a dropout risk you want to measure, not hide.
• Days of work, not weeks. If the scope creeps, cut the product, not the flow.

Rung 3: Run an AI user panel on that flow

For $39 one-time, ten AI users review your app, and three of them actually sign up and walk your core flow, screenshotting every step and quoting exactly where and why they would abandon. Results land in about 15 minutes, and reruns are free for 30 days, so you can fix and re-test without paying again.

One honesty check, because it changes how you read the results: these are simulated personas, clearly labelled as such, not real people. An AI user cannot tell you whether anyone will love your idea. What it shows you, with embarrassing precision, is where people get stuck. Confusing copy, a button that promises the wrong thing, a form that asks for a phone number before it has earned any trust.

"I stopped at the plan picker because I could not tell what I was paying for, so I would have closed the tab."

That class of feedback is a fix you can ship the same night, then confirm with a free rerun.

Rung 4: Now pay for real humans

Only now do you invest in actual user interviews. A moderated session costs $60 or more per person, and that money is wasted if your subjects spend the hour stuck on a broken signup. Fix the obvious breaks first, then spend expensive human time on the one question simulation cannot answer: does anyone actually care. Do they have this problem, how are they solving it today, what would make them switch.

Both questions matter, and the order matters too. Friction first, desire second. Desire research done on a broken flow just tells you that people hate broken flows, which you already knew.

Start at the bottom rung tonight

Paste your landing page, even a rough draft on a free subdomain, into the free roast and see what three AI personas say in 90 seconds. If it survives, climb the ladder. If it does not, you just saved yourself a month of building the wrong thing. When you are ready for the full walkthrough of your signup flow, the details are on pricing.