An idea validation framework is a fixed order of tests. You write the pass mark down before you run them. This one has four gates in a row: problem, buyer, commitment, money. Each gate names the proof that counts, the number to hit, and what you’ll spend. Fail a gate and you stop, or change one named thing.
TL;DR
- Validation rarely fails on the tests. It fails because nobody set the pass mark first. So whatever comes back reads as good news.
- Four gates, in order. Is the problem real? Can you name and reach the buyer? Will they give up something that costs them? Will they pay?
- Every gate has a different bar for proof. Enthusiasm from people who know you clears none of them.
- Decide the numbers and the budget before you start. Every other framework leaves that part out.
The number nobody writes down
Twelve interviews in. An honest founder can tell you exactly what she heard. Eight people said the problem was real. Four said they’d probably pay. Nobody laughed her out of the room. So she built for four months. The product landed to silence.
Nothing in that story is a bad test. The interviews were fine. What was missing is a number, written down first. Eight out of twelve means go. Six means stop. Without it, results don’t get judged. They get read. And you always read them in the way that suits the thing you want to build. “No market need” is still the most common cause of startup death. It sits at the top of CB Insights’ post-mortem analysis. Those founders were rarely lazy. Most of them tested. They graded their own homework.
So the first move is paperwork, not strategy. Open a doc. Write the four gates. Fill in the pass column before you talk to anyone.
The four gates
Run them in this order. The order isn’t random. Each gate is cheaper than the one after it. And each one kills a different kind of idea. Skip ahead and you pay for a test. An earlier gate would have answered it free.
| Gate | The question it settles | What counts as evidence | Pass mark (you set this first) | Budget |
|---|---|---|---|---|
| 1. Problem | Does this cost them enough that they’ve already tried to fix it? | Unprompted accounts of the problem in the last 90 days; a workaround they built, bought, or hired for | e.g. 7 of 10 target-buyer conversations describe it without you naming it | ~1 week |
| 2. Buyer | Can you name one segment and reach it repeatedly? | Named places where they gather: specific communities, lists, events, search terms with real volume | e.g. 3 channels you can reach twice a week without paying a gatekeeper | ~2 days |
| 3. Commitment | Will a stranger give up something that costs them? | Email from cold traffic, a booked call, a waitlist deposit, a signed letter of intent | e.g. 20 signups and 5 booked calls from 200 targeted visitors | ~1 week + a small ad budget |
| 4. Money | Will they pay before the thing exists? | Pre-order, deposit, paid pilot, a card charged | e.g. 3 pre-orders at the real price, not a discount | 1–2 weeks |
The pass marks above are examples, not rules. Nobody can hand you the right number for your market. Pick yours from what the decision costs. If failing means you lose a weekend, set a loose bar. If it means you quit your job, set a bar that scares you.
Gate 1: the problem, not the fix
The proof that clears this gate is always about the past. What did they do about it last time? How long did it take? What did it cost them? Rob Fitzpatrick’s The Mom Test is the whole method in one book. Ask about past behaviour. Never ask about future intent. Future intent is free to promise. One person built a spreadsheet to survive your problem. She tells you more than ten people who say the idea sounds useful.
Gate 2: one buyer you can reach
Founders skip this one and pay for it at Gate 3. That’s when they find there’s no way to put the page in front of the right people. Being reachable is a real limit, not a marketing detail. If your buyer is “small business owners,” you don’t have a buyer; you have a label, not a person. Narrow it until you can name three real places. Places you could show up on Tuesday.
Gate 3: something that costs them
Attention is free, so it proves nothing. The cheapest thing that costs them is an email address. It has to come from someone who found you cold. The strongest is a booked call. A calendar slot costs time, and you can’t get time back. This is the gate where a demand page does its job. Send targeted traffic to a page that asks for a commitment. Then count. The tactics live in our guide to validating a startup idea.
Gate 4: money before the product
Charging early isn’t about revenue yet. A person who pays turns into a different kind of source. They know what they want. They’re in a hurry. And they’re honest about what’s missing. If a pre-order is impossible in your market, swap it out. Use the nearest costly thing instead. A signed pilot deal with a start date will do. Swapping in interest is a way to skip the test. It’s how founders talk their way through this gate without passing it.
What doesn’t count
“I’d definitely use that” is future intent. It’s free to offer, and it’s usually meant kindly. Friends, workmates and your own followers are responding to you, not to the idea. Survey answers about what people might buy are the same trap. Ask a hundred people if they’d buy, and you’ll learn about their manners.
Free signups from untargeted traffic don’t count either. A viral post brings tourists. Gate 3 counts strangers who came from a channel you can use again. Advice from other founders doesn’t count, even good advice. They’re guessing about your market, and they know less about it than you do.
Copy this
Paste it and fill the pass column. Don’t edit that column after the data arrives.
IDEA: ________________________ DATE: ________
BUYER (one segment, specific): ________________
GATE 1 — PROBLEM pass mark: ____ / ____ conversations result: ____ □ pass □ fail
GATE 2 — BUYER pass mark: ____ reachable channels result: ____ □ pass □ fail
GATE 3 — COMMITMENT pass mark: ____ from ____ cold visits result: ____ □ pass □ fail
GATE 4 — MONEY pass mark: ____ paid at $____ result: ____ □ pass □ fail
TOTAL BUDGET BEFORE DECIDING: ____ days, $____
IF A GATE FAILS TWICE, I STOP. Signed: ________
The signature line is not a joke. It’s there because of the you who reads the results. That you is more hopeful than the you who wrote the plan.
Where this breaks
Long enterprise sales cycles bend Gate 4. You can’t get a pre-order out of a six-month buying process in two weeks. The honest swap is someone inside the buyer who will put a pilot date in writing. Brand-new categories bend Gate 1. People can’t describe a problem they’ve never had a name for. So look for the workaround, not the complaint. Regulated markets bend it all. There the real limit is approval, not demand.
What doesn’t bend is the pass mark. If you set the pass mark after you’ve seen the result, you’re not running a framework. You’re building a case.
If you’d rather not run the gates by hand, compare the tools that do parts of this. We do it honestly, and we say where each one beats us.
Some tools print a score that is a model’s opinion. Ours is built from five counted checks. Every point sits beside the count it came from. The report ends with up to five changes to your brief. Each one comes with the quote that prompted it and the check it helps. The full comparison is ProofMachine vs ValidatorAI.
When Gate 3 passes, you have a new problem. You must recruit the people who cleared it. That’s getting your first users.
Steve Blank’s customer development writing is still the best free read on why this works.