Most SaaS ideas die not because they were wrong, but because nobody pressure-tested them before three months of engineering had been poured in. The framework below is what we use inside the studio — a fortnight, a fixed budget, and a written go/no-go decision at the end. It works on 80% of B2B SaaS concepts and saves the rest of the world a lot of wasted code.
Why 14 days, not 90
The classic "lean validation" timeline of three months has a problem: the longer you take to validate, the more you fall in love with the idea. Sunk-cost bias creeps in by week three. By week eight you're rationalising.
Fourteen days is short enough to stay honest. It's also long enough to do real work: two weekends of customer calls, a paid signal test, a landing page, and a commitment conversation with at least three prospective buyers. Anyone telling you they validated their idea "over a weekend" validated their excitement, not the market.
Step 1: Write the kill-or-keep hypothesis (Day 0)
Before any work starts, write one paragraph in exactly this format:
Example for a fictional product:
"We believe that solo accountants serving 20–60 small businesses spend 6+ hours a week chasing client documents and would pay a monthly subscription for an automated client portal that does it for them. We will kill this idea if, after 14 days, we cannot find 5 such accountants who will pay today for the first month of access."
Write the kill clause first. If you can't articulate what would make you walk away, you don't have a hypothesis — you have a hope.
Week one: signal & conversations
Build the landing page
One page. Clear headline naming the customer and the problem. Three benefit bullets. One call-to-action ("Book a 20-minute call" or "Get early access" — pick the one matching your kill criterion). Don't add a video, don't add a logo carousel, don't write a manifesto.
Run a small paid signal test
Run a small paid ad spend on LinkedIn or Google targeting your exact ICP. The metric you care about is not conversion rate or CPC — it's whether anyone with a real name and real job title takes the action. Three real conversions from a small test budget is a strong signal. Zero conversions is a kill signal worth paying attention to.
Twelve customer calls
Reach out cold to 60 people in your ICP. Aim for 12 conversations of 20–30 minutes. Ask about the problem, not the product. The script: "I'm researching [problem area]. Walk me through how you solved it last week." Take notes. Don't pitch.
Week two: commitment & verdict
Build a clickable mock
Take what you heard in week one and build a Figma prototype showing the three core workflows. No code. Two to three days, max. The mock should be specific enough that a customer can imagine using it tomorrow.
The commitment conversation
Go back to the 12 customers from week one. Show them the mock. Ask one specific question: "If this existed today at [your price], would you pay for the first month right now to be in the beta?" Take their card or their PO. Anything less than a real commitment is theatre.
Verdict
Read your kill clause out loud. Count the commitments. Make the decision. Write it down — including the date — so future-you can't rewrite the past.
What counts as a 'pass'
From running this framework across more than 30 ideas, here's what actually predicts a market exists:
- At least 3 paid commitments (not "I'd be interested" — money or signed POs)
- At least 6 detailed problem stories from your customer calls, each describing the problem in their own words within the last 30 days
- A repeated phrase — when 4+ customers describe the problem in similar language, you've found language you can sell with
- An honest "I'd switch from [current solution] for this" from at least 2 customers — switching cost is the truest signal of pain
If you got all four, build. If you got three, keep validating for another fortnight with a sharper version. If you got fewer than three, kill it.
Three mistakes founders make in this framework
1. Pitching instead of asking
The customer calls are research, not sales. The moment you start saying "and what would be great is if we also built…" you've stopped learning. Sit on your hands. Ask one more question.
2. Counting compliments as commitments
"That sounds really useful" is not data. "Here's my card" is data. The distinction is the entire framework.
3. Moving the goalposts
If you wrote "5 paying customers" as the kill criterion and you have 2 at day 14, the answer is no — not "let me extend by a week." The framework only works if you respect the verdict you set before you started.
FAQ
Does this work for consumer SaaS too?
Adjust the commitment bar: for consumer, a waitlist with paid early-bird access (a token deposit) is a reasonable proxy for paid commitment. The 14-day structure still applies.
What if my customer can't decide alone?
If your ICP is enterprise and the commitment requires committee approval, set a different commitment criterion: "an LOI committing to a 30-day paid pilot from at least 2 buyers." Don't lower the bar — change the metric.
I can't afford a test ad spend.
Skip the paid signal and replace it with a cold outbound campaign — 200 personalised emails to your exact ICP, measuring book-rate. The signal is similar; the work is harder.
What if I'm a technical founder and want to "just build it"?
Then build a Figma prototype in two days, not code in three months. The cost of validating a wrong idea early is hours; the cost of building one is your runway.
Want a studio to run this validation with you?
Our two-week validation sprint is a fixed-fee engagement: hypothesis, research, signal test, commitment calls, written verdict. You'll know whether to build by day 15.