How to Validate a Startup Idea Before You Bet Everything on It
Most new products do not fail because the team could not build them. They fail because, once built, not enough people wanted them badly enough to pay. Idea validation is the unglamorous work of finding that out early, while being wrong is still cheap.
This guide is the backbone of Try Validator. It sets out a simple loop you can run in weeks rather than months, with links to the deeper articles on each step. None of it guarantees success. What it does is change the odds, and in a game where the base rate of failure is high, that is the whole point.
Start with the base rate, not the dream
Founders tend to judge an idea from the inside: how clever it is, how much they would use it, how big the market looks in a slide. Forecasters start from the outside, with the base rate. Most startups do not return their investors' money, and post-mortem studies such as CB Insights' widely shared analyses of failed companies have repeatedly put "no market need" at or near the top of the list of reasons. That is not a reason to give up. It is a reason to treat every idea as a bet whose odds you do not yet know.
Professional gamblers think in exactly these terms. A sports bettor does not ask whether a team is good; they ask whether the price on offer is better than the true probability of a win. A founder should ask the same thing of an idea: given what it will cost me in time and money, is my chance of success high enough? We work through the arithmetic, with expected value, bankroll sizing and the poker player's habit of separating decisions from outcomes, in startups as bets: expected value and portfolio thinking.
Step 1: write the idea as a set of assumptions
An idea is really a stack of claims. "A scheduling app for independent physiotherapists" assumes that physios struggle with scheduling, that the pain is frequent and costly, that they will switch from what they use now, that you can reach them affordably and that they will pay a price that covers your costs. Write each claim down as a sentence that could turn out to be false.
- Problem: who has it, how often and what it costs them today.
- Customer: the specific segment that feels it most, not "small businesses".
- Solution: why your approach is better than the current workaround.
- Willingness to pay: what they would pay and who signs off.
- Channel: how you will reach them at a cost below what they are worth to you.
Step 2: find the riskiest assumption
Rank the assumptions by two questions: how uncertain is it, and how fatal would it be if it were false? The one that scores highest on both is your riskiest assumption, and it is where your first test should go. For many consumer ideas it is demand. For B2B tools it is often the buying process: the user loves it, but the person with the budget does not care. For marketplaces it is usually supply.
Step 3: gather evidence, cheapest first
Evidence comes in grades. Opinions ("I would definitely use that") are the weakest. Past behaviour is stronger: what people have already tried, paid for or hacked together. Commitments are stronger still: an email address, a booked call, a deposit, a signed letter of intent. Money changing hands is the strongest of all.
| Evidence | Typical method | Cost | Strength |
|---|---|---|---|
| Stated interest | Surveys, social posts | Very low | Weak |
| Past behaviour | Customer interviews | Low | Medium |
| Small commitment | Landing-page smoke test, waitlist | Low | Medium |
| Real usage | Concierge or Wizard of Oz MVP | Medium | Strong |
| Payment | Pre-sales, deposits, paid pilots | Medium | Strongest |
Run the cheap tests first, but do not stop at them. A pile of positive survey answers is not validation; it is a hypothesis worth testing with something more expensive.
Step 4: set the pass mark before you look
Decide in advance what result would make you continue, change direction or stop. "If fewer than 10 of 30 interviewees describe this problem without prompting, we drop the segment." "If the landing page converts below 3% of targeted visitors, we rethink the offer." Writing the threshold first protects you from the most common validation error, which is reading every result as encouraging. We collect the other traps in validation mistakes and false positives.
Step 5: decide, then raise the stakes
Validation is not a single gate. Each passed test justifies a slightly bigger bet: from interviews to a smoke test, from a smoke test to a concierge service, from a concierge service to a paid pilot, and only then to a full build. Think of it as a poker hand played across several streets. You put in a little to see the next card, and you fold the moment the odds turn against you. Folding early is not failure; it is the cheapest outcome available.
The validation loop at a glance
- 1. Assumptions
- Write the idea as claims that could be false
- 2. Riskiest first
- Most uncertain and most fatal
- 3. Evidence
- Cheapest test that could prove you wrong
- 4. Pass mark
- Set the threshold before seeing results
- 5. Decide
- Continue, pivot or stop, then raise the stake
How long should validation take?
For most software ideas, a first loop of interviews and a smoke test fits into two to six weeks of part-time work. Hardware, regulated industries and anything involving licences take longer; if you are looking at payments, health or gambling products, see our notes on validating iGaming and sports-betting ideas, where the licensing hurdle can be the riskiest assumption of all.
The aim is not certainty. You will never get that before launch. The aim is to make your big bet with better information than your competitors have, and to make sure it is a bet you can afford to lose.
