How to validate a business idea before you build it

Most projects fail for a single reason: they answer a need nobody really has. You spend weeks building, you launch… and it's silence. Validating a business idea means reversing the order: check the demand before you produce, not after. Here's a concrete method to do it in under an hour, without writing a line of code.

43%

of startups that shut down cite poor product-market fit among the causes of failure.

Why validate an idea before building the product

Building first means betting hundreds of hours on a hunch. Validating first turns that hunch into measurable signals. The question is not “is my idea cool?” but “are people already looking for a solution, and would they pay for it?”.

Good news: those people leave traces everywhere — in their Google searches, on forums, on YouTube. You just have to read them.

The 4 signals to measure when validating an idea

An idea is judged on four signals crossed together, never on one alone:

• Demand — how many people actively search for the topic?
• Trend — is interest rising, flat or collapsing over 12 months?
• Competition — is the field empty, healthy, or completely saturated?
• Purchase intent — do the searches contain words that smell of a wallet (“price”, “buy”, “course”, “software”)?

Strong demand with no purchase intent = traffic that never pays. Strong purchase intent with no demand = a market that is too small. The right signal is the balance of all four.

The 5-step validation method

1. State your idea in one sentence

No business plan. One sentence describing who + which problem + which solution. Example: “Notion coaching for burnt-out freelancers”. The sharper it is, the cleaner the signals.

2. Measure real demand

Type the core of your idea into the Google search bar and look at the autocomplete suggestions: these are real searches, ranked by frequency. Many varied suggestions = a rich demand field. No suggestion at all = a warning sign.

3. Check the 12-month trend

Run your main keyword through Google Trends. A rising curve is a market coming towards you. A collapsing curve is a train leaving the station — even with demand today, you'll arrive late. Also spot the related queries with strong growth: those are tomorrow's angles of attack.

4. Probe the community and the content

A living niche has a community (active forums, specialist groups) and content (YouTube videos with views). Search your topic on forums: are people asking questions, complaining about a problem? On YouTube: is there an audience, or a desert? A community that struggles is a gold mine of pains to solve.

5. Assess purchase intent

Go back to the searches you collected in step 2 and count those containing commercial markers: price, pricing, buy, course, software, best, review. The more there are (and the “hotter” they run), the more money there is in the niche. A search for “how to do X for free” is not worth a search for “best X software price”.

How to read the validation verdict

Once the signals are crossed, file your idea into one of three boxes:

• GO — solid demand, healthy trend, manageable competition, purchase intent present. Ship an MVP this week.
• REFINE — the vein exists but one signal is off. Tighten the angle or the segment before charging ahead.
• PIVOT — too little demand or a dead market. Don't burn your energy here.

The mistakes that distort validation

• Asking people close to you. They lie out of kindness. Search data doesn't.
• Confusing interest with intent. “That's interesting” is not “I would pay for that”.
• Ignoring the trend. A free-falling niche can look big… right before it disappears.
• Over-building. The point of validation is to spend as little as possible to learn as much as possible.

Turning signals into an experiment

Desk research reduces uncertainty, but it doesn't replace a real commitment. Write down your three riskiest assumptions: the problem exists, the segment is reachable, and the value justifies a price. Then attach an expected proof to each one. For instance, ten professionals accept an interview, five share their current process and two ask to test a solution.

Test the assumption that could kill the project first. If regulation blocks access to the data you need, improving the landing page is pointless. This ordering stops you from stacking up easy proofs while ignoring the central risk.

Worked example: an AI newsletter for lawyers

The initial wording is too broad. So you observe searches around legal monitoring, the existing offers and conversations between professionals. The interviews may reveal that the problem is not receiving more news, but spotting quickly the rulings that affect one precise type of case. The idea then becomes “a weekly digest of the developments that matter to employment lawyers”, with a verifiable source and a verifiable time saving.

A page presents a full example and offers a paid four-week pilot subscription. Email opens measure interest, but payments and renewals measure value. If readers browse without renewing, you need to understand whether the content genuinely replaces a task or is simply pleasant to read.

Keeping a validation log

For every test, note the date, the assumption, the segment, the channel, the threshold you set before seeing the result, and what you observed. Finish with “what this experiment does not prove”. This log stops you from reinterpreting a mediocre result as a win, and makes comparing several ideas easier. Serious validation is not certainty: it is a series of better-informed decisions.

The shortcut: validating an idea in 30 seconds

Running these 5 steps by hand takes 30 to 60 minutes per idea. Sondari automates the whole thing: it cross-references Google Autocomplete, Google Trends and YouTube, computes a Sondari Score across those four dimensions, and hands you the GO / REFINE / PIVOT verdict directly — in 30 seconds. Ideal for testing ten ideas in an afternoon instead of one in a week.

For what comes next, see our guide on finding a profitable niche.