How to Validate a Side Project Before You Build It

Most side projects fail before launch. Not because the builder lacked skill, but because they built something nobody wanted. The good news: you can find that out in a week, not six months.

Here is a repeatable process for validating a side project idea before you write a line of code or spend a cent.

Start With a Problem, Not an Idea

The instinct is to start with a cool thing you want to build. Flip that. Start with a problem you have personally, or one you have heard repeated by at least five other people.

"I want to build a SaaS dashboard" is not a problem. "Freelancers on Upwork lose 3-4 hours a week reconciling invoices across platforms" is a problem.

Write the problem in one sentence. If you need more than one sentence, the problem is not clear yet. Clarity here saves you weeks later.

Check Search Volume and Forum Activity

Before anything else, see if people are already searching for a solution.

Use Google Keyword Planner or Ahrefs to check monthly search volume for your problem phrase. You are looking for at least 500 monthly searches globally. Below that, the market is probably too thin or people are not actively seeking a fix.

Then go to Reddit, Hacker News, and niche forums. Search for the problem. Look for threads where people are frustrated, asking for tool recommendations, or describing workarounds. A thread with 200 comments about a problem is more useful than keyword data because it shows emotional investment.

If you find both search volume and forum frustration, that is a real problem.

Build a Fake Door

A fake door test is a landing page that describes your product as if it exists and has a call-to-action - usually "Join the waitlist" or "Get early access."

You do not need to build the product first. You need:

Tools like Carrd or Framer let you ship this in under two hours. Buy a domain. Set up the page. Done.

Your target conversion rate is 20% of visitors signing up. If you drive 200 people to the page and fewer than 40 sign up, something is off - either the problem is wrong, the audience is wrong, or the copy is weak.

Drive Targeted Traffic Fast

A landing page with no traffic tells you nothing. You need at least 200 visitors before you can trust the conversion rate.

Three ways to get there quickly:

Reddit and forums. Post in the exact communities where you found the frustrated threads. Do not pitch. Share the landing page in the context of asking for feedback on a solution you are exploring. Be upfront. Most communities respond well to honest builders.

Twitter/X DMs. Find people who have tweeted about the problem. Send 20 short, specific DMs. Say you are building a tool for the exact thing they complained about and ask if they would look at it. Expect a 10-15% reply rate.

A small paid ad test. Spend $50 on Google or Reddit ads targeting the exact search phrase you validated earlier. If your cost per signup exceeds $10, the organic routes will not save you - the economics are broken.

Talk to 10 People

Email signups are weak signals. Conversations are strong ones.

Pick 10 people from your waitlist and ask for a 20-minute call. Use a Calendly link. Expect 3-5 to actually show up.

On the call, ask:

You are not pitching. You are listening. The answers tell you whether the pain is real, what the existing alternatives are, and what feature actually matters.

If nobody can name what they currently pay to solve the problem, monetisation will be hard.

The Pre-Sale Test

This is the highest-signal test. Ask people to pay before the product exists.

Set a price. Even $29 or $49. Send an email to your waitlist explaining that you are building this, here is exactly what it will do, and early supporters can lock in a discounted price now.

If 2-3% of your list buys, you have a viable project. If zero people buy, you have a real answer before you wasted four months building.

Use Stripe and a simple form. You do not need a payment system built into a product that does not exist yet.

When to Move Forward

Move forward when you have:

Hit four of those five and build the simplest version of the solution. Not the full product - the smallest thing that solves the core problem.