Validation timelines
How long does it take to validate a startup idea?
A realistic calendar: how many hours each test takes, a 7-day sprint for one hypothesis, a 30-day plan for a GO/NO-GO, and what to do when day 90 arrives without a decision.
Reviewed October 28, 2026. Timings assume evenings and free tools — field-realistic, not best-case.
Quick answer
A weekend per hypothesis. About 30 days for a defensible answer. Day 90 without a decision is itself a decision.
One validation question — will tutors pay for scheduling software? — fits inside a week: a landing page takes an evening, ads need about seven days to produce a readable signal, and five customer interviews fit in four to five days of calendar. A full GO/NO-GO on the idea usually takes 20-40 hours spread over a month, because two or three tests need to agree. And if you are 90 days in with no test that changed your mind, the evidence is probably telling you something you don't want to hear.
Key takeaways
How the time actually goes
- Speed comes from running tests in parallel, not from rushing any single one: build the landing page while interviews book, run ads while you read replies.
- Ad-driven tests need about seven days of data before the signal is readable. Three days of clicks is noise with a chart.
- Interviews are calendar-bound, not effort-bound: five of them spread across 4-7 days because that is how scheduling works.
- The expensive delay is not testing — it is polishing a landing page for two weeks and calling it validation.
- About half of new US establishments don't reach year five, per BLS data. Waiting three months to learn what a week could have told you compounds that risk.
Why the clock matters
The calendar matters more than the budget
The U.S. Bureau of Labor Statistics' Business Employment Dynamics program tracks what happens to new establishments: roughly 78% survive their first year, about 51% reach year five, and around 35% reach year ten. Most of that attrition is decided in the earliest months, while the business is still mostly a hypothesis.
Every week spent building before testing pushes the same learning risk further out, at a higher cost. The tests below are ordered by how fast they return a signal — start with the fast ones.
Test times
How long each test takes
Setup assumes evenings and free tools. Wait time is calendar time, not hours worked — run tests in parallel and the total is shorter than any single test.
| Test | Setup | Wait for signal | What it tells you |
|---|---|---|---|
| Problem interviews (five) | 2-3 hours prep | 4-7 days to schedule and run | Does the problem come up unprompted, in their words? |
| Landing page + waitlist | One evening | 3-7 days of traffic | Do strangers hand over an email address? |
| Fake-door / smoke test with ads | 1-2 evenings | 7 days of ad data | Click-through and sign-up rates, audience by audience |
| Willingness-to-pay question or pre-order | One evening to add | 2-7 days | Do people commit money, not just interest? |
| Concierge test (do it by hand) | 1-3 days | 1-2 weeks serving first users | Do users come back without you chasing them? |
| Prototype in front of users | 3-7 days | 1-2 weeks of sessions | Do they use it correctly without help? |
These are the same tests described across this site. Pick two or three; a verdict needs them to agree.
The 7-day plan
Seven days, one hypothesis
For a founder with a job, evenings only. This plan answers one question — not the whole business.
Pick the riskiest assumption
Write down the one belief that, if wrong, kills the idea. Not "people like the idea" — "tutors will pay $15/month for scheduling." One sentence, falsifiable.
Book five interviews
Post in two communities where your audience already gathers. Offer $25-50 gift cards if they're strangers. Expect some no's and some no-shows.
Build the landing page
One evening: Carrd, one promise, one call to action, a Tally form behind the button. No carousel, no roadmap section, no logo.
Interviews one and two
Ask about their current workaround, not your idea. The Mom Test script on this site is the checklist.
Turn on $5-10/day of ads
Meta or Google, two audiences, one ad each, both pointed at the landing page. Don't touch the copy for seven days.
Interview three
By now the pattern usually shows up in their words, not yours.
Read the week
Five conversations, a few hundred ad impressions, a sign-up count. One of three things happened: the problem showed up unprompted, it didn't, or your five people were too polite. Write down which.
This won't produce a verdict on the business. It produces a next step with evidence behind it — which is what day 7 is for.
The 30-day plan
Thirty days to a GO/NO-GO
Four weeks, two gates, part-time. Each gate is a written decision — continue, adjust, or stop.
Week 1 — problem evidence
Five problem interviews plus a landing page built by the weekend. You're collecting evidence that the problem is real and urgent for a specific group of people.
Gate: at least three of five interviewees describe the problem unprompted. Fewer than three — rewrite who it's for before spending a dollar.
Week 2 — demand evidence
Seven days of $5-15/day ads on the landing page. Two audiences, an honest promise, a real sign-up form. No tweaking mid-week.
Gate: a sign-up rate you'd be comfortable explaining to a stranger, or a clear winner between the two audiences. Neither — the offer is the problem, not the traffic.
Week 3 — money evidence
Add a willingness-to-pay step: a price on the page, a pre-order button, or a direct question inside two more interviews. Interest is free; commitment is data.
Gate: at least one person has committed something scarce — money, a deposit, a signed design-partner note. Zero commitments after two weeks of signals is itself a finding.
Week 4 — the decision
Put the three gates side by side and decide: GO with a 30-day build scope, NO-GO with reasons written down, or PIVOT with the one assumption you'd retest. Whatever you decide, it's a decision — not a mood.
Write it down. A GO without a scope is how day 90 happens.
The 90-day trap
What it means when day 90 arrives
If week 4 produced a decision, day 90 finds you mid-build or glad you stopped. But if you're 90 days in and still "gathering data," something else is going on. Three patterns cover most of it:
- The tests were vague. "Getting feedback" is not a test; "will five tutors pay $15/month" is. Vague tests can't fail — which is why they run forever.
- Every result got reinterpreted as encouragement. When a low sign-up rate becomes "the landing page needs better copy" four times in a row, the copy was never the question.
- Stopping feels like losing. It isn't. A written NO-GO with reasons is evidence too — and the cheapest tuition available.
The fix at day 90 is the same as at day 7: one falsifiable question, one small test, one written decision. If you need permission to stop, write the kill criteria in advance. Your future self will thank you.
FAQ
Timeline questions founders actually ask
- How long does it take to validate a startup idea part-time?
- Most people can run one full test cycle in two to three weeks of evenings and reach a defensible GO/NO-GO in about 30 days. The constraint is scheduling interviews and letting ad data accumulate — not the work itself.
- Can you validate an idea in one weekend?
- One hypothesis, yes: a landing page, a handful of community replies, maybe one interview. That's a useful signal, not a verdict. Treat the weekend as the first of several small tests rather than the whole validation.
- How long should a smoke test run?
- About seven days of ad data. Clicks in the first two or three days are too thin to trust; a week gives you enough to compare two audiences honestly. If nothing converts by day 7, extend once with different copy, then decide.
- How many interviews can I realistically do per week?
- Three to five is a full pace alongside a day job. Scheduling, not stamina, is the limit: expect no-shows and reschedules. Five good conversations beat fifteen rushed ones.
- When is validation actually done?
- When you can write a one-page decision with evidence attached: the problem is real for a specific group, strangers act on it, and at least one person has committed something scarce. That's a month for most ideas — and "done" doesn't mean "every question answered."
- Is it bad if validation takes longer than 90 days?
- It depends what the 90 days contained. Three completed tests with a documented reason to continue is diligence. Three months of vague feedback with no written gates is avoidance — usually a NO-GO in disguise.
- Should I build an MVP while validating?
- Build the smallest thing that makes the next test possible — a concierge version done by hand often beats code. A full MVP is what you build after a GO, scoped to the promise the evidence supported.
- Where can I see the survival data you mention?
- The figures come from the U.S. Bureau of Labor Statistics' Business Employment Dynamics program. We summarize the survival numbers, with links to the source tables, on our startup statistics page.
Related guides
- howLongToValidate.relatedcheapestTestTitlehowLongToValidate.relatedcheapestTestBodyRead the guide →
- howLongToValidate.relatedcostTitlehowLongToValidate.relatedcostBodyRead the guide →
- howLongToValidate.relatedinterviewCountTitlehowLongToValidate.relatedinterviewCountBodyRead the guide →
- howLongToValidate.relatedmomTestTitlehowLongToValidate.relatedmomTestBodyRead the guide →
- howLongToValidate.relatedfakeDoorTitlehowLongToValidate.relatedfakeDoorBodyRead the guide →
- howLongToValidate.relatedkillCriteriaTitlehowLongToValidate.relatedkillCriteriaBodyRead the guide →
Give the calendar a job.
One question, one test, one written decision — then repeat. The analyzer can score your idea and name the riskiest assumption before day one.