Most idea validation founders do is theater. They talk to friends who say "cool idea!" They post polls that everyone smashes "yes" on. They send out surveys and read the answers charitably. They interview 5 potential customers who politely agree the problem exists.
None of that is validation. Validation is the process of trying to KILL your idea and failing. If your validation effort can only produce "yes, this seems good" — you haven't validated anything. You've collected encouragement.
Real validation asks: "what would make me stop building this?" and then actively tests for that. Here are the 5 methods that actually falsify ideas, and the specific signals that mean "stop" — the ones I use with founders before they spend money on a build.
First: understand what you're actually testing
An idea has three separate validation questions:
1. Is the problem real and painful enough to spend money on? (customer discovery) 2. Will your specific solution solve it well enough? (solution validation) 3. Can you reach the people with the problem economically? (market/channel validation)
Most founders test #2 without ever testing #1 or #3. That's why so many well-built products fail — the founder validated the solution people liked in theory, without validating whether they'd actually pay for it or how you'd find them at scale.
Test in that order: #1 first, then #2, then #3. Don't skip.
Method 1: The pre-sales test (the strongest signal that exists)
What it is: Before building, try to sell the product. Real money, real commitment. If you can't, you don't have validation. Full stop.
The specific move: 1. Write a 1-page description of what your product will do (specific enough that a customer would understand what they're getting) 2. Set a real price. Not "free trial to build interest" — actual dollars. 3. Reach out to 20-50 people in your target customer segment personally (see the first-10-customers post for the outreach playbook) 4. Ask them to pre-pay (or sign a Letter of Intent to pay when it's ready) 5. Track: how many said yes, how many hesitated, how many said no, and WHY
What the signals mean:
- 5-10 of 20 pre-pay: you have strong validation. Build.
- 1-3 pre-pay: weak signal. Investigate what specifically made them commit vs the others.
- 0 pre-pay but many "sounds interesting": the idea sounds nice but isn't painful enough to pay for. Danger sign.
- 0 pre-pay AND everyone politely declines: the idea is dead. Kill it.
Why this is #1: pre-payment is the ONLY validation that separates "people say they want this" from "people will actually pay for this." Every other method has some risk of collecting encouragement without commitment. Pre-payment doesn't.
The uncomfortable truth: most founders skip this because they're scared of hearing "no." That fear is the signal you MOST need to overcome, because it's the same fear that will kill your business later when you should have listened.
Method 2: The competitor cost-of-workaround audit
What it is: find out what your target customers are ALREADY doing to solve the problem, and how much time/money they spend on that workaround.
The specific move: 1. Identify 10 people in your target segment 2. Ask each: "How do you currently deal with [the problem]?" 3. Follow up: "How much time does that take you per week?" and "What do you pay for tools/services related to this?" 4. Add up their total pain (hours × their hourly rate + tool spend)
What the signals mean:
- They currently spend 5+ hours/week or $200+/month on workarounds: strong pain. There's likely willingness to pay.
- They currently spend 1-2 hours/week and no money: weak pain. Hard to convert unless you're 10x better.
- They don't do anything about it currently: the problem isn't painful. This is the killer signal that most founders miss — if people aren't ALREADY paying to solve it in some form, they probably won't pay you to solve it either.
The specific test: if your prospects can't clearly articulate their current workaround AND its cost, they don't feel the pain enough to pay for a solution. Move on.
Method 3: The 10-user manual delivery (Wizard of Oz test)
What it is: deliver the OUTCOME your product promises, manually, before building the product. If people won't pay for the manual version, they won't pay for the automated version.
The specific move: 1. Land 5-10 real customers who will pay for the outcome 2. Deliver that outcome yourself, by hand, over 30-60 days 3. Track: retention (do they keep paying?), NPS (would they refer?), and the specific parts of the workflow that were valuable vs unused
Example: you want to build "AI-powered contract review software." Instead of building, offer "we review contracts in 24 hours and send you a report." Do the reviews yourself. Charge $200 each.
What the signals mean:
- Customers stay for 3+ months and refer others: validated. Now build the product to scale what you've proven works.
- Customers cancel after 1-2 months: the outcome isn't as valuable as you thought, or your delivery quality is inconsistent. Investigate before building.
- You can't even find 5 people willing to pay for the manual version: the outcome isn't valuable enough. Kill the idea.
Why founders resist this: it feels inefficient. "I could build it in 3 months and then sell to 100 people." No, you couldn't — you'd build it, find out no one wanted it, and lose the 3 months. Manual delivery is the fastest way to test if there's a business here.
Method 4: The specific ad test (channel + demand validation combined)
What it is: spend $500-$2,000 running highly targeted ads to a landing page that describes your product, and measure the specific conversion behaviors.
The specific move: 1. Build a one-page landing page describing the product + a "Get Early Access" or "Notify Me" button 2. Run $500-$2,000 of ads targeted to your specific customer profile (LinkedIn if B2B, Meta if consumer, Google if high-intent search) 3. Track: cost per click, click-through rate on ads, and conversion rate on the landing page
What the signals mean:
- CTR > 2% on ads AND landing page conversion > 5%: strong signal that the pitch resonates. Now test if signups convert to paying.
- CTR < 1% but landing page conversion > 5%: the audience isn't reachable via this channel, but they ARE interested. Try different channel.
- CTR > 2% but landing page conversion < 2%: the pitch attracts curiosity but doesn't sell. Refine the pitch.
- Both metrics low: either wrong audience, wrong pitch, or wrong product. Investigate.
Warning: signups are NOT validation. Willingness to leave an email is a very weak commitment. Follow up 2 weeks later with a real offer and see how many convert to paying. If <5% of email signups eventually pay when the product is ready, you got interest without demand.
Method 5: The "cold reject" test (the fastest way to falsify)
What it is: find 20 people in your target segment and specifically try to make them say NO to your idea. Give them every reason NOT to be interested.
The specific move: 1. Contact 20 people who are ideal candidates 2. Instead of pitching, ask: "Here's what I'm thinking of building [description]. Honestly, is there any reason this would NOT be useful for you?" 3. Actively look for the objection. Don't defend. Just listen and note. 4. See what patterns emerge across the 20 responses.
What the signals mean:
- Most objections are minor (price, feature X, timing): those are solvable. The idea has legs.
- Most objections are fundamental ("I already have this in [existing tool]", "That's not really a problem for me", "I'd never trust a small startup with this"): the fundamentals are broken. Reconsider.
- Everyone gives different reasons for saying no: either your target audience is too broad, or the idea has multiple problems.
- You can't even get 20 people to hear the pitch: if your audience isn't willing to spend 15 minutes even hearing about it, you'll never sell to them at scale.
Why this is powerful: most founders' "customer discovery" interviews are actually pitch sessions where the founder subtly encourages the interviewee to agree. The cold reject test flips the incentive: you're actively trying to get NO. Any "yes" you get after actively trying to get "no" is much more real.
The signals that mean STOP (regardless of method)
Beyond the individual method signals, watch for these red flags across ALL your validation work:
Red flag 1: You can only find people to talk to who are already predisposed to like the idea.
If the only "validated" customers are friends, connections through your existing network, or people you already had access to — you haven't validated at all. Real validation includes cold audiences.
Red flag 2: The value proposition changes every time you explain it.
If your description of what the product does drifts across 5 different conversations, you don't yet have a clear idea. Building something with an unclear identity is expensive.
Red flag 3: The excited responses are all future-tense ("this would be great!") not present-tense ("I need this now").
Future-tense enthusiasm is often social — people want to encourage you. Present-tense need is validation. When someone says "when can I have this? I want to pay you today" — that's real. When they say "if this existed, I might use it" — that's not.
Red flag 4: You can't articulate ONE specific user in ONE sentence.
If your target is "small businesses" or "professionals" or "people who want X" — that's not specific enough to build for. You should be able to name a real person by role, industry, company size, and specific job-to-be-done in one sentence.
Red flag 5: You're building for a problem YOU have, and refusing to check if others have it.
Some of the best startups came from founders scratching their own itch. But some of the worst came from founders who assumed everyone shared their itch without testing. The difference: real validation of others, not just introspection.
Red flag 6: The market doesn't exist yet at meaningful scale.
"I'm going to create a new category" is a very small percentage of successful startups. Most successful ones enter existing markets with a specific differentiation. If nobody is spending money in your space today, it's unlikely they'll start when you show up.
When to stop validating and start building
Validation isn't infinite. At some point you have enough signal to commit. Here's the threshold:
Green light to build:
- 5+ people have pre-paid or committed to pay real money
- You can explain the value in one sentence and it consistently lands
- Your target customer is one specific person type, not a category
- You know at least 2 channels where you can reliably find your target customer
- The manual/Wizard-of-Oz version has run for 30+ days and customers stayed
Yellow light — keep validating:
- You have signups/interest but no paid commitments
- The pitch is landing with some people but not others
- You know your customer but not your channel
Red light — stop, reconsider:
- Nobody will pre-pay
- Interviews are all polite encouragement, no committed intent
- The idea keeps changing shape
- You can't find enough people to even test with
A specific example
I worked with a founder who had an idea for a B2B tool. Common story. Here's how validation unfolded:
Week 1-2: ran the cold reject test with 30 target customers. 22 said the problem was real. 8 pre-committed to $200/month if the tool existed. Strong signal.
Week 3-4: built a minimal Wizard-of-Oz version — she did the work herself, in a Google Sheet, updated manually. Charged the 8 people $200/month. All 8 stayed for month 1.
Week 5-8: 6 of 8 stayed for month 2. 2 referred colleagues (now 12 customers).
Week 9: started building the real product with confidence — she'd already proven the demand, the price point, and the retention. The build was funded by the 12 paying customers.
Contrast: she'd previously spent 6 months building a different idea without doing this validation. That product launched to 200 signups and 4 paying customers, then stalled and shut down.
Same founder. Same skill. Same building capacity. The difference was validation-first vs build-first.
What to do this week
If you have an idea but haven't validated:
- Start with Method 5 (cold reject test) — it's the fastest and requires no investment
- Contact 20 people in your target segment. Actively try to get NO.
- Report back what you heard, not what you hoped to hear.
If you have positive interviews but no revenue:
- Try Method 1 (pre-sales). If nobody will pay, you don't have validation.
- The gap between "sounds interesting" and "I'll pay for it" is huge. Cross it before building.
If you have paying customers already:
- Move on from validation. Focus on retention and growth — see the 5 metrics post.
- If those metrics are healthy, the SaaS onboarding fixes post covers what to work on next.
If you're still figuring out the shape of the idea:
- The MVP planner tool on this site helps articulate scope in a way that makes it validate-able
- The refactor vs rewrite framework applies if you're staring at an existing product that hasn't validated — sometimes it needs a strategic pivot rather than more building
The founders who validate well aren't the ones with the best ideas. They're the ones willing to hear "no" and act on it. Every "no" you accept honestly during validation is 3 months of building you don't waste.
---
If you have an idea and want an outside opinion on how to validate it — specifically which of the 5 methods to run first, and what to watch for — reach out via the contact page with a paragraph about the idea and target customer. I'll spend 30 minutes on it and give you a specific plan.