Guide
How to Validate a Startup Idea Before Building
Validate a startup idea by finding where people already discuss the problem publicly, interviewing 10–15 of them about what they’ve tried and failed at, and only starting to build once you see the same pain point and workaround show up repeatedly.
Why this matters
An idea feeling obviously good to you is not evidence. It’s common to spend months building something nobody asked for because the idea made sense in the abstract. Validation replaces that gamble with a cheap, fast check: does this problem actually show up, unprompted, in how real people already talk and behave?
The validation method
- 1
Write the problem down in one sentence, not the solution
"People struggle to stick with a new habit after the first week" is a problem statement. "An app with streak tracking" is a solution. If you skip straight to the solution, you’ll only find people who agree with your framing.
- 2
Find where people already complain about this problem
Search forums, subreddits, review sections, and support threads for the problem in their own words, not your product’s vocabulary. You’re looking for unprompted complaints, not answers to a survey.
- 3
Talk to 10–15 people who’ve lived the problem
Ask about the last time it happened to them, not whether they’d use your product. Past behavior is a far more reliable signal than a hypothetical "would you..." answer.
- 4
Look for a repeated pattern, not a single enthusiastic reply
One person loving your idea is encouragement. The same pain point, workaround, or objection showing up across most conversations is evidence.
- 5
Decide what would have to be true to build
Before writing code, write down the specific pattern that would justify building (e.g. "8 of 12 people describe this exact failure mode and are paying for a worse workaround"). If you don’t see it, narrow the problem and go back to step 2.
Common mistake
Asking "would you use this?" instead of "tell me about the last time this happened to you." Hypothetical interest is cheap; people are polite and will usually say yes. Past behavior is what actually predicts future behavior.
How FirstSight helps
FirstSight automates the first two steps: describe the problem in a sentence, then let it search public discussion for people already describing it. It surfaces recurring pain points and existing workarounds, then generates specific interview questions for you to use in step 3. It provides an honest read on whether the public signal is strong enough to proceed. It does not replace customer conversations; it helps you know what to ask.
Ready to move from guessing to evidence?