How to Validate a Business Idea
A four-rung ladder from "interesting idea" to "validated business" — with the right experiment for each rung and the pass/fail threshold that tells you to climb, pivot, or stop.
Why Validate Before You Build?
Validation means gathering evidence that strangers will pay for your solution — before you've spent months building it. Each rung of the ladder below tests a different assumption, in order of how likely it is to kill the business: first that the problem is real, then that there's demand for your solution, then that people will pay, and finally that the business model underneath can actually work.
The order matters because each rung is cheaper than the one above it. A week of interviews can save you a quarter of building; a landing page can save you a year. Sketch your assumptions on a Lean Canvas first so you know exactly what each experiment is testing.
Problem Validation
Does this problem exist, and does it hurt?
The experiment: 15-20 problem interviews with your target segment. Ask about the last time they hit the problem and what they did about it — never pitch your solution.
The threshold: Pass: a majority describe the problem unprompted and have actively tried to solve it (workarounds, spreadsheets, paying for partial fixes). Fail: they agree it's "annoying" but have done nothing about it.
Demand Validation
Will people act on a promise to solve it?
The experiment: A smoke test: landing page describing the solution + a real call to action (join waitlist, book a demo, start trial), driven by a small ad spend or community posts.
The threshold: Pass: visitors convert on the CTA at a rate meaningfully above noise for your channel, and follow-up conversations confirm intent. Fail: traffic arrives, nobody commits anything — not even an email.
Willingness-to-Pay Validation
Will they exchange money — before it's finished?
The experiment: Pre-sales, paid pilots, refundable deposits, or signed letters of intent at a real price. In B2B, ask for the meeting with the budget holder — that request is itself a test.
The threshold: Pass: a handful of strangers (not friends) commit money or contractual intent. Fail: everyone loves it at "free" and vanishes at a price.
Business Model Validation
Do the unit economics of delivering it work?
The experiment: Serve your first paying customers manually if needed (concierge MVP) and instrument everything: what it costs to acquire them, serve them, and keep them. Test each block of your canvas — channel, pricing, delivery — as its own hypothesis.
The threshold: Pass: a credible path where gross margin and acquisition cost can work at scale, and early customers stay and expand. Fail: every sale is heroic, delivery costs eat the price, or churn erases growth.
Testing the Business Model, Block by Block
Rung 4 deserves its own discipline. A business model is a bundle of assumptions — about the customer, the channel, the price, and the cost to deliver — and each block of your Business Model Canvas can be tested independently. Three lenses cover them all:
- Desirability — do customers want it? (Tested on rungs 1-2: interviews, smoke tests.)
- Viability — can it make money? (Tested on rung 3 and 4: pre-sales, pricing experiments, early unit economics.)
- Feasibility — can you deliver it? (Tested by delivering manually first: if a concierge version can't satisfy customers, automation won't save it.)
Work through the riskiest lens first for your model: a marketplace usually dies on desirability of the second side; a services business on viability; a hardware product on feasibility.
The 4 Validation Mistakes That Produce False Positives
Pitching instead of interviewing
The moment you describe your idea, every answer becomes polite encouragement. Validate the problem first; reveal the solution only at the demand stage.
Counting compliments as evidence
"I would definitely use that" costs the speaker nothing. Only behavior counts: time given, email shared, money committed.
Validating with friends and family
They're answering a different question — "do I support you?" Recruit strangers from the actual target segment, even though it's slower.
Stopping at demand and skipping the model
Plenty of ideas have eager users and no viable business underneath (the cost to serve exceeds any plausible price). Rung 4 exists because rungs 1-3 can all pass on a business that still doesn't work.