Short answer: Prove the problem is real before you build. That is how you validate a startup idea. Interview 10–20 target users. Ask what they do today, not what they might buy one day. Then put a live landing page in front of strangers. Ask for a commitment: an email, a pre-order, a deposit. Commitments validate. Compliments don’t.
TL;DR
- 42% of failed startups die from “no market need,” the top cause, ahead of running out of cash. (CB Insights)
- Validate the problem first. A painful problem with a so-so fix beats a slick fix nobody needs.
- Ask people about past behavior. “Tell me about the last time this happened” beats “would you use this?”
- A landing page beats an MVP. Dropbox’s 3-minute video took its waitlist from 5,000 to 75,000 overnight, before the product existed.
- Take money early. Only ~9% of founders pre-sell. The ones who do get the truth everyone else pays for at launch.
Why most startup ideas die (and how to stop it)
When CB Insights analyzed the post-mortems of failed startups, the most common cause of death was “no market need”: 42% of them built something nobody wanted. (CB Insights) Running out of cash gets blamed more often, but the cash usually ran out because not enough people cared. Validation finds that out early. The test costs a weekend, not two years. Full breakdown in why startups fail.
Here’s the trap. Most first-time founders walk into it. They fall in love with a solution. Then they go hunting for a problem it fits. Building feels like progress. Here’s the discipline. You sit in “I don’t know if anyone wants this yet” long enough to get a real answer. Built the whole thing first, then went looking for users? Most of us have done it. Some of us twice.
What does it actually mean to validate a startup idea?
To validate a startup idea, you gather real proof. You gather it before you build. The proof has to show one thing. A specific group of people has a problem. It hurts enough that they’ll pay to make it go away. A poll of your friends doesn’t count.
Three things people blur together:
- Market research tells you the market exists and how big it is. Useful, but a big market full of people who won’t switch is a graveyard.
- Idea validation tells you that this group has this pain and will act on it.
- Product-market fit comes after you build. It’s when demand starts pulling the product out of your hands.
Step 1: Validate the problem, not your solution
Write down the problem, not the product. Open with “I’m building an app that…” and you have skipped the question. Here’s the question. Is the pain frequent and costly enough? Are people already trying to solve it today?
You’re hunting for what some founders call a Tier 1 problem. The pain is so bad that people have already built a workaround. An ugly one. A tangle of spreadsheets, maybe, or a part-time hire doing it by hand. Duct-tape solutions are the strongest green flag there is. They prove that people already spend money and time on this.
Now the flip side. If nobody solves it today, there is often no problem. “No competitors” is the scariest phrase in a pitch deck.
Step 2: Interview 10–20 people the right way (The Mom Test)
Customer interviews are where you win or lose it. They’re also where founders fool themselves the most. People are nice. Ask “would you use an app that does X?” and they’ll say “sure, sounds great,” because saying no to your face is awkward. The result is a folder of compliments. You’ll mistake it for signal.
The fix is The Mom Test (Rob Fitzpatrick). Ask the questions even your mom couldn’t lie about. Ask about the past and the present. Never ask about the future they imagine.
- Bad: “Would you pay for a tool that does this?” You’ll hear “yeah, probably!” Worthless.
- Good: “Walk me through the last time you dealt with this. What did it cost you?” You’ll hear a real story with real stakes.
Field rules:
- Ask what they did, not what they think. Last week’s actions predict the future. What they say they’ll do doesn’t.
- Talk less than they do. If you’re pitching, it’s a sales call. Shut up and dig.
- Chase the money and the workaround. “What are you using now? What does it cost you, in dollars or hours?” Money already flowing is the loudest signal.
- Talk to skeptics. The people who won’t switch tell you more than the ones who love you.
On volume: a pattern shows up around 5 conversations, pricing intuition around 10, real confidence past 20. Immad Akhund had roughly 90 conversations. That was all before he wrote a line of Mercury’s code. He talked to founders, investors, and lawyers. He wanted to know if the rules would even allow it. Karri Saarinen did this before Linear. He spent about a year in chats with coworkers. They worked at Uber, Airbnb, and Coinbase. He kept hearing the same complaint. Software was slow. That became Linear’s wedge.
Step 3: Get a commitment, not a compliment
“Cool idea!” costs nothing to give, and that’s exactly what it’s worth. A commitment costs the giver. An email address, or money down. The cost is what makes it data.
Bob Moore (Crossbeam) saw this. He was pitching early ideas to other founders at the time. One signal was worth trusting. Someone stepped up to introduce him to a customer. That put their reputation on the line.
Rank commitments by what they cost the giver:
| Weak signal (cheap) | Strong signal (costly) |
|---|---|
| “Great idea!” | Gives you their email for early access |
| A thumbs-up on your post | Books a call / joins a design-partner program |
| “I’d definitely use that” | Pre-orders, or puts down a deposit |
| A survey response | Introduces you to their boss / their team |
Step 4: Ship a landing page and test real demand (the smoke test)
Most guides mention this step and walk right past it. It’s the fastest honest read on demand you can get.
A smoke test is a real landing page for a product that doesn’t fully exist yet: a clear headline, the promise, maybe a mock screenshot, and a single call to action (“get early access,” “join the waitlist”). Drive cold traffic to it and measure whether strangers act. It’s also called a fake-door test. Clicking a button is a small commitment. A small commitment is worth more than a big compliment.
Here’s the classic example. Drew Houston put up a 3-minute demo video for Dropbox. He did it before he built the hard sync engine. The beta waitlist jumped from 5,000 to 75,000 people overnight.
What a good smoke test needs:
- One promise, one CTA. The single outcome your user wants, and one button.
- Real traffic. Post where your target users already are. Or spend $50–100 on ads to buy honest strangers. Warm traffic from friends lies.
- A pass bar you set before you look. Does cold traffic give you an email below roughly 2–5%? Then the message isn’t landing. The problem or the audience is off. Above it, you’re onto something.
This is where an idea-validation tool earns its keep: instead of guessing at a pass bar, run the search against real threads and get the verbatim buyer evidence plus the communities to send traffic to. An afternoon instead of a fortnight. Want to shop around first? We line up the best idea validation tools, and we say where each one beats us. Two matchups get searched most. One is ValidatorAI alternatives. The other is ValidatorAI vs VenturusAI. We put ourselves in ProofMachine vs ValidatorAI.
Step 5: Pre-sell (the highest-fidelity test there is)
A landing page measures interest. A pre-sale measures belief. MicroConf’s survey work found the number. Only about 9% of founders pre-sell. They do it before building. Read that as an edge. The ones who ask for money early get the truth. Everyone else pays for it after launch.
Failory calls the reason the “Niceness Gap”: people will tell you your idea is great to spare your feelings, but almost nobody hands over a credit card to be polite. Money burns the niceness away.
Proof you don’t need scale:
- Kettle & Fire: Justin Mares spent $50 on Bing ads driving to a pre-order page and booked $500 in pre-sales in two weeks. He had a real bone-broth business before any stock.
- Steph Smith pre-sold a product to 86 buyers in under 24 hours (~$1,270), to an audience she’d built.
- Maven: Gagan Biyani ran a “Minimum Viable Test,” a live cohort course co-taught with Sam Parr, and pulled ~$150,000 in first-cohort revenue with a 9/10 rating, before building the platform.
You don’t need a Stripe integration either. “I’m taking $20 deposits for the first 10 spots, refundable if I don’t ship” is a pre-sale, and the number of yeses is worth more than 100 survey responses.
The scorecard: how to read the signal
Founders collect a pile of notes and page numbers. Then they stall on the decision. Score the idea on five dimensions, each with a concrete pass bar. Rate 1–5:
| Dimension | Fails (1–2) | Passes (4–5) |
|---|---|---|
| Problem severity | “Minor annoyance, I guess” | People already pay for or hack together a workaround |
| Willingness to pay | “I’d use a free version” | Pre-order, deposit, or existing spend on an inferior tool |
| Demand signal | You have to explain the problem first | Cold landing-page traffic converts above your pass bar |
| Reachable audience | Scattered, no shared watering hole | Concentrated in findable communities you can name |
| Message resonance | People nod politely | People finish your sentence and ask “when can I get it?” |
Two of them predict failure best. Willingness to pay, and demand signal. A 1–2 on either kills the idea. It doesn’t matter how good the rest looks. Fix the problem or the audience first, then build. Want the way to decide, not the tactics? We do that in our idea validation framework: four gates in a row. The pass mark gets written before the test.
[OUR DATA: e.g. "Across N ideas run through ProofMachine, the most common failing dimension was X (Y%)." — GATED: do NOT fill until N ≥ ~100 runs. See proprietary-data.md publish gate.]
How long should validation take?
Validation is a ladder. You climb only as far as the signal takes you. Time-box it:
- A weekend (~2–4 hours): desk research. Is anyone solving this today, and what do they charge? Where does the audience gather? Kill obviously dead ideas here for free.
- Two weeks (~20 hours): 10–20 Mom Test interviews. Plus a live landing page with a single CTA and real traffic. This is where most ideas should live.
- A month: a pre-sale or a Minimum Viable Test. Real money or real commitments from strangers.
Write production code only after that. Each rung’s job is to give the next rung permission to exist.
Common mistakes that quietly kill good ideas
- Pitching friends and family. They love you, so they lie. Talk to strangers who have the problem.
- Asking hypothetical questions. “Would you use this?” invites a comfortable yes. Ask about the last time it happened.
- Counting praise as commitment. Enthusiasm is free. Emails and money are not.
- Skipping willingness to pay. Love ≠ revenue. Test the wallet early.
- Validating forever. Set your pass bars, hit your time-box, then make the call: build it or kill it.
You’ve validated — now what?
A validated idea with no users is still just an idea. You built a landing page to test demand. Now it’s the page you send your first real users to. Don’t tear it down. Point traffic at it.
Your first users are already in a room. They’re arguing about the problem in there. So getting your first users from Reddit is the next step. It’s the natural one after a green light. To find that room without reading a thousand threads yourself, run an idea validation search and start from what it brings back.
Sitting on an idea and not sure it clears the scorecard? Run it through the idea validation tool. You get the threads where people describe the problem in their own words, the communities they’re posting in, and a verdict that is allowed to say no. About ten minutes.