How to Validate a Startup Idea Before You Spend a Dollar on Development

Direct answer: Validate a startup idea by getting strangers to take a real action, talk about a real problem, hand over money, or sign a letter of intent, before you write a line of code. Run 2 to 3 of 7 low-cost methods (interviews, a landing page, a concierge test, a Wizard of Oz test, pre-sales, community research, or a competitor teardown) over 2 to 4 weeks, then score the results against a go/no-go threshold.

The 7 Validation Methods, Ranked by Cost

1. Problem interviews: free, days
2. Community and forum research: free, days
3. Competitor teardown: free, days
4. Landing page smoke test: $50 to $500, 1 to 2 weeks
5. Concierge test: your time, 1 to 3 weeks
6. Wizard of Oz test: your time, 1 to 3 weeks
7. Pre-sale or letter of intent: free, hardest to get

Run the free methods first (1 to 3), then layer in one or two that ask for a real commitment (4 to 7): replace an opinion with evidence before it gets expensive to be wrong.

Why Validation Comes Before Code

Direct answer: Code is the slowest, most expensive way to test demand, and it's the default move for most first-time founders, since writing software feels like progress while talking to strangers doesn't. An analysis of 431 VC-backed shutdowns found 43% failed because the product didn't fit a real market need, ahead of running out of cash (29%) and broken unit economics (19%) (source: CB Insights, Why Startups Fail). Every method below exists to catch that before a budget goes into building the wrong thing.

How to Run a Problem Interview That Doesn't Lead the Witness

Direct answer: Ask 15 to 20 people about a specific past instance of the problem, never about your solution or their future intentions. The goal is a stranger talking about their life for 80% of the conversation while you listen, not a friendly nod at your pitch.

Most interviews fail the same way: the founder asks a question the other person can only answer politely. "Would you use an app that does X?" gets a "yeah, probably," worthless, people are bad at predicting their own behavior and good at being nice. This is the exact failure The Mom Test by Rob Fitzpatrick fixes: talk about their life, not your idea; ask about the past, not hypotheticals; let them talk (source: The Mom Test by Rob Fitzpatrick).

Leading questions to cut, and what to ask instead:

A script for almost any idea: open with "Tell me about the last time you dealt with [problem]," follow with "what did you do about it," and close with "who else deals with this?" (how one interview becomes three). If nobody describes a real, recent instance of the problem, there's no urgent problem to build for, no matter how many say the idea "sounds interesting."

How Effective Is a Landing Page Smoke Test, and What Counts as a Good Conversion Rate

Direct answer: Ship a one-page description of the product with a real call to action (waitlist, "request access," or a fake "Buy Now" button), drive a small amount of real traffic to it, and measure who takes the action. A 5 to 10% sign-up rate is a reasonable bar for genuine interest from cold traffic, and 25%+ conversion on a waitlist CTA is a strong signal (source: CXL, Guide to Smoke Testing; Indie Hackers, What's a Good Signup Rate on an Idea Validation Page?).

Build it in a day with Carrd or Framer, write one clear sentence about the problem, and put $50 to $200 behind Google or Meta ads pointed at the audience from your interviews. Below 3 to 5% sign-ups, the page is telling you the problem isn't urgent, not that the headline needs a redraft. Traffic quality beats page design: 200 visitors who match your interview subjects outperform 2,000 random clicks.

What Is a Concierge MVP and When It Beats a Landing Page

Direct answer: A concierge test delivers the outcome by hand, spreadsheets, manual emails, you personally doing the software's job for 3 to 5 customers, while they know a human is behind it. It's slower to set up than a landing page but tells you which parts of the workflow are actually painful enough to automate first, which a click on a button never will.

The textbook example is Food on the Table: before building software, founder Manuel Rosso personally shopped groceries for early users to test whether they wanted a custom meal-planning service, learning which features and price point mattered before automating anything (source: Shortform, What's a Concierge MVP?). Keep it deliberately small. If you can't keep up manually with 3 to 5 customers, that tells you where the real bottleneck sits before code hides it.

What Is a Concierge MVP and When It Beats a Landing Page (AI-generated illustration)

How a Wizard of Oz Test Works Without Building the Backend

Direct answer: Build only the front end a customer sees, a form, a page, or a chat window, and have a human manually perform the "automated" work behind the scenes without the customer knowing. It looks and feels like a real product while costing a fraction of what the actual backend would take to build.

Zappos is the standard case: founder Nick Swinmurn tested demand for buying shoes online, before building any inventory system, by photographing shoes at local stores and personally buying and shipping them per order (source: Learning Loop, Concierge vs. Wizard of Oz). Early Q&A service Aardvark ran the same play, routing questions to staff who typed answers manually, no algorithm underneath. Use this over concierge whenever the product needs to feel automated, since a visible human would undercut what you're testing.

How to Get a Pre-Sale or Letter of Intent Before the Product Exists

Direct answer: Ask for a deposit, a signed contract, or a written letter of intent before the product exists, targeting the same people from your problem interviews. Money and signatures are the only validation signal that can't be faked by politeness, an enthusiastic "I'd definitely use that" is worth far less than a card on file or a signature on a page.

This is the hardest method to pull off, which is exactly why it's the strongest signal. For a B2B idea, ask a prospect who described the problem clearly to sign a one-page letter of intent: what you'll deliver, by when, and what they'll pay, no obligation if you don't ship. For a consumer idea, a real pre-order or refundable deposit does the same job. As a rule of thumb, 3 to 5 people willing to commit money or a signature beats 50 people saying they like it.

How to Validate Demand by Lurking in Communities Before You Build Anything

Direct answer: Search Reddit, niche Slack groups, and industry forums for people already describing your target problem, before you post a single feature idea. Communities of 10,000 to 500,000 members tend to give the best signal: large enough for real data, focused enough to avoid generic noise (source: PainOnSocial, How to Validate Startup Ideas on Reddit).

Search phrases like "wish there was a tool that" and "does anyone know a way to" inside relevant subreddits, sort by "Top" over the past year, and read the comments, since they reveal what people already tried and where it failed. A pain point worth building for shows up across multiple unrelated communities: if freelancers, agencies, and solo founders all complain about the same invoicing headache in three different subreddits, that's a broader market than one thread suggests. It costs nothing but time, and it surfaces the exact wording you can reuse in a landing page headline.

How to Validate an Idea by Studying Competitors You Haven't Talked to Yet

Direct answer: Read the negative reviews of every existing competitor on G2, Capterra, and app stores before assuming there's no competition or that competition means the idea is dead. The "what do you dislike" and "reasons for switching" fields on review sites surface objection patterns, pricing friction, and feature gaps that rarely come up in a friendly sales call.

How to Validate an Idea by Studying Competitors You Haven't Talked to Yet (AI-generated illustration)

Pull the last 12 months of reviews for your top 5 competitors, cluster complaints into 5 to 8 recurring themes (setup, pricing, integrations, support), and flag any competitor scoring below 7.5 on a feature category, a repeatable weakness, not a one-off complaint (source: Steve, G2 and Capterra Reviews as Competitive Intelligence). Existing competitors are good news: they prove people already pay for this. Your job is finding the slice they're visibly failing.

How to Score Your Validation Results and Decide Go or No-Go

Direct answer: Score each method you ran on a simple point scale, add them up, and use fixed thresholds instead of a gut feeling. A rough score of 6 or higher out of a possible 15 to 18 (depending on how many methods you ran) means build a v1; below that, keep testing or drop the idea.

Weight pre-sales and LOIs highest, they're the only signal that survives contact with someone's wallet. A high interview score with zero willingness to pay is common and not, by itself, a green light. In a gray zone, run one more method instead of talking yourself into an answer.

The 4-Week Startup Validation Timeline

Direct answer: Spend week 1 on interviews and community research, week 2 on a landing page, weeks 3 to 4 running a concierge or Wizard of Oz test with the most interested people, then score the results and decide before writing production code.

Two or three methods run well beat all seven run badly. A B2B idea leans on interviews, pre-sales, and teardowns; a consumer idea leans on community research, a landing page, and a Wizard of Oz test.

When to Kill the Idea, Even Though Sunk Cost Makes That Hard

Direct answer: Kill or pivot the idea when validation produces polite interest but no money, no signatures, and no repeat manual usage, after genuinely running 2 to 3 of the methods above. The hardest part isn't running the tests, it's acting on a result that contradicts months of invested belief.

Sunk cost is the bias that keeps founders shipping a validated "no" anyway: it feels more painful to walk away from time already spent than to have never spent it. A 2026 survey of 200 founders found 81% had pivoted from their original idea at least once, and 42% wished they had pivoted sooner (source: Wilbur Labs, Why Startups Fail (2026)). Most founders do eventually pivot, but the delay is expensive. Set the go/no-go threshold before running the tests, and treat "the numbers don't support it" as a finding, not a failure.

What to Do After You Validate: Scope the MVP

Validation gets you to one decision: build or don't. Once the answer is yes, the next failure mode is over-scoping v1 into a shrunk five-year vision instead of a product that does one job well. That step, plus pricing and the first ten customers, lives in How to Build a SaaS Product. For what a properly-scoped v1 costs by build type, see MVP Development Cost in 2026. If you've validated demand and need a lean v1 without overbuilding it, that's what software development for startups is for.

Frequently Asked Questions

What's the cheapest way to validate a startup idea?

Problem interviews cost nothing but time. Talk to 15 to 20 people about a specific past instance of the problem, never your solution, and let them do most of the talking. Community and forum research is a close second and surfaces the exact language real customers use.

How long should startup idea validation take?

A focused sprint takes 2 to 4 weeks: interviews and community research in week 1, a landing page and competitor teardown in week 2, a concierge, Wizard of Oz, or pre-sale push in weeks 3 to 4. Running it longer usually means avoiding a decision, not gathering signal.

What's a good conversion rate for a startup idea landing page?

A 5 to 10% sign-up rate from targeted, cold traffic is a reasonable bar for genuine interest, and 25%+ on a waitlist CTA is a strong signal. Below 3 to 5% usually means the problem isn't urgent enough yet.

How many customers do I need before building an MVP?

There's no fixed number, but 3 to 5 people who commit real money or a signed letter of intent beat 50 people saying "I'd use that." Money and signatures are signals that survive contact with reality; enthusiasm in an interview often doesn't.

When should I give up on a startup idea?

Give up, or pivot, when 2 to 3 methods produce polite interest but no money, no signatures, and no repeat usage of a manual version. Founders who wait past that point are usually fighting sunk cost, not new evidence: 42% of founders in a 2026 survey wished they'd pivoted sooner.

Do I need a landing page to validate a startup idea?

No. Problem interviews, a concierge test, community research, and pre-sales each validate a different part of the idea without any ad spend. Most founders get a clearer picture from combining two or three methods than from any single one.

What's the difference between a concierge MVP and a Wizard of Oz test?

In a concierge test, customers know a human is manually delivering the service. In a Wizard of Oz test, customers believe they're using automated software while a human works behind the scenes. Use concierge when the manual process builds trust; use Wizard of Oz when the product needs to feel automated.