How to Validate a Business Idea Before Building It
To validate a business idea before building it, check five things in public evidence: that buyers describe the problem in their own words, that many people talk about or search for it, that someone already pays for a fix, that there is room for a newcomer, and that you can reach the buyers. Then run one cheap test that would prove demand in a week. If the evidence is thin on pain or payment, change the idea before you write a line of code or order a single unit.
Most ideas don’t fail because the builder was slow. They fail because nobody wanted the thing enough to pay for it. The good news: the market leaves evidence everywhere, and reading it takes hours, not months.
Why does validating before building matter?
Building is the expensive part. A physical product needs samples and inventory, a SaaS needs weeks of code, an app needs a store listing and reviews. Validation moves the hard question (“will anyone pay?”) to the front, where the answer is cheap.
An AI chatbot alone won’t do this for you. Ask Claude or ChatGPT if your idea is good and it will usually say yes and hand you a plan. It hasn’t read this year’s threads, seen which ads keep running or checked what competitors charge. Validation means looking at what buyers actually do, not at what a model hopes they do.
What should you check, and where?
Five checks cover most ideas. Each has a place to look and a signal that means yes.
| Check | The question | Where to look | A yes looks like |
|---|---|---|---|
| Pain | Do buyers describe the problem themselves? | Reddit threads and comments, X posts | People describe failed fixes and ask for help |
| Demand | Do many people talk about or look for it? | Reddit, TikTok and YouTube views, search suggestions | Many threads, videos with real views |
| Pay | Does anyone already pay for a fix? | Competitor prices, TikTok Shop sales, long-running ads | Products sell at real prices; ads run for months |
| Room | Is there space for a newcomer? | Complaints in comments, unserved groups | Recurring complaints about what exists |
| Reach | Can you find these buyers cheaply? | Subreddits, creators, search terms | Active communities that point at this buyer |
Start with pain. If you can’t find buyers describing the problem in their own words, the rest rarely matters. Our guide on how to find pain points on Reddit walks through the search terms and the comment patterns that show real pain.
How do you tell real pain from polite interest?
Real pain has three marks: the person describes what they already tried, they say how it failed, and they ask for something better. A comment saying the idea “would be cool” is interest. A comment from someone who tried three planners and quit each one within a week is pain.
Read comments, not just thread titles. Titles get written for attention; comments are where people explain their situation. Count how often the same complaint shows up across different threads and different subreddits. One loud thread is an anecdote. The same complaint in five places is a pattern.
How do you know people will pay?
Good proof is that someone already sells something close at a real price. That feels backwards, since competitors seem like bad news. But a market with no paid products is often a market with no buyers.
Three places show payment:
- Competitor prices. List what close products cost. See how to find competitors for a product idea for where to look by kind of product.
- Ads that keep running. An ad that has run for a month or more is someone paying to sell this, every day. The Facebook Ad Library guide shows how to read run length.
- Units sold. For physical products, TikTok Shop shows sold counts. The TikTok Shop research guide covers what those numbers mean.
When prices exist, they also tell you where to price. How to price a new product turns a competitor price range into a number to test.
What does a real example look like?
Here is one example check Vettoz ran on October 18, 2026, for an undated paper planner for adults with ADHD (physical product, US). It isn’t proof that every planner idea works; it shows what strong evidence looks like.
- 15 Reddit threads with 561 comments
- 12 TikTok videos with 9.4 million plays and 10 YouTube videos with 9.5 million views
- 960 active Meta ads, 12 of them running for 30 days or more
- 10 priced competitors from $2.20 to $63, with a median of $11.49
The check scored 86 out of 100, a “go”. The weakest dimension was room, at 69: the market is busy, so the product would need a clear angle. Compare that with a team hydration-reminder SaaS checked the same day: it scored 46, a “rethink”, because most of the evidence was about something else (threads about electrolyte drinks) and no competitor showed a price.
Which low-cost test settles it?
Evidence tells you whether to test. A test tells you whether to build. Pick a test that costs little and would prove demand in 7 days, and decide the number that means go before you start.
| Kind of idea | Low-cost test | A number that means go |
|---|---|---|
| Physical product | Pre-order or waitlist page plus one short video | Pre-orders or signups you set in advance |
| Digital product | Pre-sale page or a low-priced first version | Paid orders, not likes |
| SaaS | Landing page with a waitlist and a small ad budget | Signups at a cost you can live with |
| Mobile app | Landing page with a TestFlight waitlist | Waitlist signups from the right audience |
The kind of product changes the details. There are deeper guides for validating a SaaS idea, validating a digital product idea and app store research for app ideas.
What are the warning signs to stop?
Some patterns mean the idea needs to change before any test:
- Nobody describes the problem, or only you do.
- The conversation exists, but it’s about a different problem.
- Free tools and DIY dominate, and nothing sells at a real price.
- A few giant brands own every search and every ad slot.
Signs your product idea is bad goes through each one with what to change instead of giving up.
FAQ
How long should validation take?
Reading the evidence takes a few hours by hand, or about a minute with a tool that collects it for you. The test that follows takes about a week.
Do I need to talk to customers?
Conversations help, especially for SaaS. Public evidence comes first because it shows you who to talk to and what to ask.
Is a crowded market bad?
Not on its own. Competitors prove payment. A crowded market needs a sharper angle, and the comments usually tell you which one.
Can ChatGPT or Claude validate my idea?
On their own they answer from memory. Connected to a research tool, they can read live evidence and write the report from it. See validating a business idea with ChatGPT.
What if the score is low?
Change the buyer, the angle or the kind of product, and check again. A low score on one version doesn’t kill the problem.
Check your idea with Vettoz
Vettoz does the evidence part inside Claude, ChatGPT or Cursor. He reads Reddit, X, TikTok, YouTube, active Meta ads, Google, and TikTok Shop or the App Store depending on the kind, scores the idea on pain, demand, pay, room and reach, and your AI writes the report with links. See how it works or check 2 ideas free.