How to validate a SaaS idea before you build it

Step 1: Write the idea in one sentence

Who is it for, and what problem does it fix? 'A tool that turns logos into laser cutter files for hobby makers' is testable. 'An AI platform for creators' is not. If you cannot name the buyer, every later step will give a vague answer.

Stuck for ideas to test? The SaaS idea generator and the Reddit digests pull problems from real posts.

Step 2: Check if people search for it

Search demand is the cheapest proof that a problem exists. Type the problem the way a buyer would into Google and read the autocomplete suggestions. Then put the main phrases into a keyword tool to see monthly searches and difficulty.

• Look for a cluster of related searches, not one big keyword. Ten phrases with 100 searches each are easier to win than one with 10,000.
• Prefer words that show intent to act: 'convert', 'generator', 'template', 'alternative', 'for' plus a job title.
• Zero results everywhere is a signal too. It usually means people do not think of this as a problem they can solve with software.

Step 3: Look at Google page 1

Search your main phrases and look at who ranks. This tells you whether a new site has any chance.

• Forums, Reddit threads and thin blog posts on page 1: good sign. Nobody has built a proper answer yet.
• Small tools and indie sites: fine. You can compete on being better or narrower.
• Only big brands and directories: hard. Look for a narrower phrase instead.

Step 4: Find competitors that already make money

A competitor with revenue is the best news you can get. It proves people pay for this. Search for existing tools, check their pricing pages, and look for verified revenue where it is public (for example on revenue leaderboards like TrustMRR).

No competitors at all is usually not a gap in the market. More often it means there is no market. Be suspicious of ideas where you cannot find a single paying product nearby.

Step 5: Read the complaints

Read reviews of the competitors and threads where people talk about the problem. You are looking for repeated, specific complaints: too expensive, missing one feature, too complicated for a small team. A repeated complaint is a way in.

Write down the exact words people use. They become your page headline and the search phrases you target.

Step 6: Ask for money with a fake door

Everything above tells you a market exists. It does not tell you that people will pay you. For that, put up a page that describes the product, gives a real price and has a buy button. When someone clicks, tell them it is not ready and let them leave an email. That is a fake door test; there are three real examples with numbers here.

Write the decision rules before launch. Ours: wait 28 days; under 30 visitors means extend 14 days, and still under 30 at 42 days means close it; a buy-click rate under 2% means kill, 2 to 5% means talk to people first, over 5% means keep going; no more than 3 tests at once.

What happens when you skip step 6

I built a resume tool and ran it for 8 weeks: 23 signups, 0 paid. In one stretch 6 of 7 signups reached checkout and none of them paid. Signups felt like validation. They were not.

Two things I do differently now: ask for money from the first version of the page, and read the funnel step by step (visit, signup, checkout, pay) so you can see exactly where people stop.

And one thing the current tests taught me: traffic is the first problem. Our three fake doors had 42 visitors in the first five days, which is not enough to judge anything. Plan where visitors will come from before you build the page.

Do it yourself, free, or have it done

You can do steps 1 to 5 by hand with Google and a keyword tool. To speed it up, IdeaPanner walks you through the same checks for free: target searches, who holds page 1, rivals and whether they make money, then a build, pivot, wait or drop verdict.

If you would rather have a person do it, the Idea Check is done by hand with real data, not an AI score.

Questions