How to Validate a Startup Idea Before Writing a Line of Code

I've watched 30+ founders spend 3-6 months building MVPs for problems that didn't exist. Each one had ample 'validation' upfront — surveys, friend feedback, encouragement from their cofounder. None of it was real validation. Here's what actually works.

Two weeks. Twenty interviews. One landing page. A $9 preorder. If you can't get five strangers to commit nine real dollars, you don't have a problem worth building for. This post is the playbook.

What 'validation' actually means

Validation is evidence that strangers will pay you money to solve a problem they have today. It is not survey results. It is not friends saying 'cool idea.' It is not VCs nodding politely. The only valid signals are time, money, and unprompted action.

• Time: a stranger spends 30+ minutes on a call, follow-up email, or testing prototype.
• Money: a stranger pays a non-trivial dollar amount before the product exists.
• Unprompted action: a stranger refers another person, reschedules a missed call, or follows up unprovoked.

The 14-day validation playbook

This is the exact sequence I run with founders pre-MVP. Two weeks, no code, hard yes/no decision at the end. Roughly 60% of ideas fail this test, which is the point.

1. Day 1: write a one-paragraph problem statement. Who exactly has this problem (job title, company size, geography). What they're doing today instead. What it costs them. If you can't write this in 100 words, you don't understand it yet.
2. Day 2-3: build a target list of 50 candidates who match the problem statement. Use LinkedIn Sales Navigator, Apollo, or your existing network. Tier them: 20 'cold,' 20 'warm,' 10 'hot' (already complaining about the problem in public).
3. Day 4-9: run 20 customer discovery interviews. 30 minutes each. Use the Mom Test rules below. Record (with permission). Take notes on actual quotes, not your interpretations.
4. Day 10: synthesize the interviews. Look for the same phrase appearing 3+ times. That's a real problem. If you find no repeated phrasing, your problem statement is wrong — go back to day 1.
5. Day 11: build a one-page landing site on Framer or Webflow. Hero, three benefits, three FAQs, a $9 'preorder' button via Stripe Checkout that captures email and charges. Total time: 4 hours. Don't optimize for design.
6. Day 12-13: drive 500 targeted visitors. Options: $300 LinkedIn ads to your ICP, 50 cold DMs to 'hot' candidates, posts in 5 relevant subreddits or Slack communities. Track UTMs.
7. Day 14: tally results. Decide go/no-go using the threshold below. Refund preorders if no-go. Build if go.

Customer interviews: the Mom Test rules

Rob Fitzpatrick's The Mom Test is the canonical interview methodology. The book is 130 pages and worth the four hours. The compressed version: ask about specific past behavior, never about future intentions. Three rules guide every question.

1. Talk about the customer's life, not your idea. Bad: 'Would you use a tool that did X?' Good: 'When was the last time you did X manually? What happened?'
2. Ask about specifics in the past, not generics or opinions. Bad: 'How often do you have this problem?' Good: 'Walk me through the last time it happened. When? Who was involved? What did you do?'
3. Talk less, listen more. Aim for 70% of the airtime on the customer. If you're explaining your idea for more than 90 seconds total, you're contaminating the data.

Six questions I ask in every customer interview, in order:

• 'Walk me through what you did yesterday at work.' (gets them in narrative mode)
• 'When was the last time [specific situation related to the problem]? What happened?' (extracts a real story)
• 'What did you try first to solve it?' (reveals existing alternatives)
• 'What was hard about that?' (uncovers pain)
• 'How much time/money did that cost you?' (quantifies the pain)
• 'Who else have you talked to about this?' (gets you the next interview)

Landing page test: $9 preorder mechanics

The $9 preorder test is where validation becomes real. Survey results have an 80% deception rate. Email signups have a 50% deception rate. Money exchanged has a 5% deception rate. Charge $9 because it's small enough to be impulsive but large enough to filter tire-kickers.

The page itself is six elements: a headline, a subheadline, three benefit bullets, three FAQs, social proof if you have any (logos of who you've talked to is fine), and a prominent CTA button to a Stripe Checkout page. No login. No 'learn more.' No multi-step flow.

The headline should match exact phrasing from your interviews. If three customers used the phrase 'losing track of expense reports,' your headline is 'Stop losing track of expense reports.' Don't write clever copy. Mirror their language.

The go/no-go decision criteria

After 14 days, you have data. Here's the exact threshold I use to recommend whether to start building.

Real example: a B2B founder I worked with

A founder came to me with a 'tool for managing freelancer payments.' Generic, broad, lots of competition. We ran the 14-day playbook. Twenty interviews surfaced one specific phrase: '1099 reconciliation at year-end is a nightmare.' That's not freelancer payments. That's tax compliance.

Repositioned the landing page around 1099 reconciliation. Charged $19 instead of $9 (more specific, higher willingness to pay). Drove 300 visitors via LinkedIn to CFOs of agencies. Got 22 preorders in 5 days. Conversion: 7.3%. Build go.

We then did the [MVP Build Sprint](/services/mvp-sprint) in 12 days. They had 50 paying customers within 60 days of launch. The whole process from idea to revenue was under 90 days, with two weeks of validation upfront preventing what would have been three months of wrong building.

Common validation traps and how to avoid them

Five validation anti-patterns I see weekly. Avoid these and your signal-to-noise ratio doubles.

• Talking only to friends, family, and your existing network. They want to support you. They will lie to you with kindness. Always include 50%+ strangers.
• Pitching the solution before extracting the problem. The moment you mention your idea, the interview is over. Save it for the last 5 minutes if at all.
• Counting email signups as validation. Email is free. Money is not. Charge $9 minimum.
• Skipping the landing page test because 'we already have interest.' Verbal interest is fiction. Run the test anyway.
• Validating with founders or VCs instead of customers. They have different incentives than buyers. Talk to buyers.

How long should validation take?

Two weeks is the right ceiling. More than that and you're avoiding the build, which is its own form of procrastination. Less than that and you don't have enough signal.

If you cannot do 20 interviews and a landing page test in 14 days, the limiting factor is not your idea — it's your access to the target market. That's a different problem and you should solve it first. Sales-to-customers is harder than build-the-product. Don't skip the access question.

Once you've validated, the build comes next. For scope, see [MVP features you should cut](/blog/mvp-features-cut-list). For stack, see [the MVP tech stack for 2026](/blog/mvp-tech-stack-2026). For timing, see [how long does it really take to build an MVP](/blog/mvp-timeline-realistic).

What to do if validation fails

Validation failing is a win. You saved 3-6 months of building the wrong thing. Three options after a failed validation:

1. Pivot the problem statement. Same target customer, different problem. The interviews surfaced a real pain — but it wasn't the one you assumed. This is the most common outcome.
2. Pivot the customer. Same problem, different customer. Maybe SMBs don't care but enterprises do, or vice versa.
3. Kill the idea cleanly. Refund preorders. Send a thank-you email. Take 2 weeks off. Start the next validation cycle. The interview list and the warm 5-10 paying preorders carry forward.