Most startups do not fail because the founders could not build. They fail because the founders built something nobody needed — and found out too late. Idea validation is the discipline of collecting evidence of real demand before you invest months of engineering time in a solution that may not fit.
This guide covers what validation actually means, the cheapest experiments that produce meaningful signal, how to talk to the right people, and the common mistakes that leave founders six months in with nothing to show for it.
What validation really means
Validation is not about proving that your idea is good. It is about disconfirming the hypothesis that it is bad — cheaply, quickly, and repeatedly. A validated idea is one where you have collected evidence, not opinions, that a specific group of people experience a painful problem and will take real action to solve it.
The key word is evidence. Opinions are easy to collect. You can get five people in a room, pitch your idea, and walk away with five enthusiastic nods. Those nods cost nothing and predict nothing. Evidence requires that someone does something: pays a deposit, books a call, joins a waitlist with real intent, or changes their behaviour in response to your offer. The gap between "sounds interesting" and "here is my credit card" is where most startup ideas go to die.
Validation also has a scope. You are not trying to validate that a market exists in the abstract — you are validating that your ICP, with their specific context, experiences the problem acutely enough to seek a solution and pay for one. A useful mental model: you are trying to find the people for whom your solution would be their first choice today, not the people who might eventually use it if the price were right.
Talk to your ICP before you build anything
The most reliable foundation for validation is direct conversation with people who match your Ideal Customer Profile. Not friends. Not colleagues. Not entrepreneurs who think about products for a living. The specific type of person who would be a paying customer in the first cohort.
The goal of these conversations is not to pitch — it is to understand the problem's shape, frequency, and priority in their real workflow. The framework from Rob Fitzpatrick's The Mom Test is the right anchor: ask about past behaviour, not future intentions. "Walk me through the last time you dealt with [problem]" produces more useful data than "would you use a tool that...?" because past behaviour is real, while future predictions are optimistic fiction.
Our guide on customer discovery interviews covers the full arc of question design, recruiting, and synthesis in detail. Before you invest in any of the experiments below, run at least ten to fifteen conversations with genuine members of your target segment. What you hear will reshape your experiments.
Cheap experiments that produce real signal
Once you have a problem hypothesis grounded in real conversations, the next step is to design experiments that test willingness to pay or willingness to change behaviour — not just willingness to express interest. Here are the four that consistently produce the clearest signal at the earliest stage:
Landing page with a friction step
A one-page description of the problem you solve and the outcome you deliver, followed by a concrete next step, is the cheapest experiment you can run. The critical detail is the friction step. A "notify me" email box is nearly friction-free and produces weak signal. A form that asks two qualifying questions, or a button that books a 20-minute call, introduces just enough friction to filter intent from curiosity.
Drive targeted traffic from communities where your ICP already gathers, not from broad social posts. Measure the conversion rate from visit to completed friction step, not just from visit to email submission. A 5–10% conversion rate on a qualifying form from a targeted audience is a strong positive signal. A 40% email capture rate from a vague audience tells you almost nothing.
Problem interviews with a pre-sell close
After fifteen conversations about the problem, you will know which interviewees felt the pain most acutely. Go back to the five or six who were most animated and offer them early access at a founding-member price. You do not need to have built anything. The offer can be explicit: "I am building a solution to this exact problem. If I can show you a working version in eight weeks, would you put down a refundable deposit today?"
The number of people who say yes to a refundable deposit is one of the most honest signals available at the pre-product stage. It separates the people who are genuinely frustrated by the problem from the people who enjoyed talking about it.
Concierge MVP
A concierge MVP delivers the outcome of your proposed product manually, without any of the product. If you are building a tool that automatically categorises support tickets, spend two weeks categorising tickets manually for three paying customers. You learn whether the outcome is valuable, what the edge cases are, and what your customers actually do with the output.
Concierge MVPs are underused because founders find them embarrassing. That embarrassment is misplaced. A paying customer who receives a valuable outcome manually is stronger validation than a thousand free users who clicked through your Product Hunt listing.
Wizard of Oz prototype
A step beyond concierge: you build just enough interface to make the product feel real, but humans power the backend. The user interacts with what looks like an automated system; you or your team deliver the result. This tests the product experience — including the interface and the delivery format — without building the automated system.
This approach is particularly useful when the manual concierge version is logistically difficult (e.g. the customer expects real-time results) but the automated version would take months to build.
Reading the signals: what matters and what does not
Not all positive responses mean the same thing. Part of becoming a good validator is learning to weight signals correctly.
Weak signals — things that feel encouraging but do not predict success: email signups without qualifying questions, verbal agreement in a conversation, social media likes and shares, praise from people who are not in your ICP, people who say they would "definitely use it" when asked.
Strong signals — things that require real commitment from the person responding: money exchanged (even a small deposit), a scheduled onboarding call the person showed up for, an unsolicited referral to someone else in the same ICP, a detailed description of exactly what they would change about their current workflow if your product existed, or someone switching from their existing solution mid-conversation.
The principle is simple: the easier it was for someone to take an action, the less the action tells you. Design your experiments to make the signal-producing action meaningful, not frictionless.
The role of ongoing feedback in validation
Early validation is a sprint, but the deeper question — whether your product creates enough value to retain users over time — can only be answered by watching real people use it over months, not days.
This is where the structure of ongoing monthly video feedback from genuine users becomes a validation instrument in its own right. When a user records a screen-share walkthrough every month explaining what they like, what is confusing, and what is missing, you receive twelve rounds of honest signal from someone who has lived with your product through their real workflow. That kind of depth is qualitatively different from a one-time onboarding interview or a feature request in a support ticket.
The goal is not just to validate that the problem exists — it is to validate that your product solves it well enough that people keep choosing it. Retention, not activation, is the real validation test for a SaaS product.
Validation and messaging: two sides of the same process
The language your interviewees use to describe their problem is the most valuable output of your validation process beyond the go/no-go signal itself. When you hear the same phrase in three separate conversations — "I spend half my day on this", "it's a black box", "I never know where things stand" — you are being handed your headline copy.
Before you write a word of marketing, read through your interview transcripts with a highlighter and collect every phrase your ICP used to describe the problem, the frustration, and the desired outcome. Then look at your current homepage. If the language does not match, your messaging is not validated, even if your product idea is.
Validated messaging reduces the cost of every subsequent acquisition channel because your copy resonates immediately with the people you are trying to reach. It is the compounding return on the investment you made in early conversations.
Common mistakes that waste months
The most expensive validation mistakes are the ones that feel like progress while they are happening:
- Talking to the wrong people. Getting encouraging feedback from people who are not in your ICP is worse than silence, because it is misleading. Every interviewee should be a genuine candidate for your first paid cohort — not a former colleague doing you a favour.
- Pitching instead of listening. If you spend more than a third of your interview time explaining your idea, you are not doing discovery. The goal of early conversations is to understand their world, not to sell them on yours.
- Conflating interest with intent. Someone who says "this is a great idea" and someone who says "I will pay you for access next month" are in completely different categories. Treat them accordingly.
- Building before you pre-sell. The risk of building first is that you spend months on something nobody will pay for, then have to learn that lesson from the product rather than from a conversation. Pre-selling — even at a conceptual level — forces the truth to surface early.
- Using a broad ICP definition. If your ICP is "small business owners" or "marketing managers," your validation findings will be too diffuse to act on. Tighten your ICP until you are being exclusionary, then validate within that segment.
When validation leads to first10
Once you have validated that a problem exists and that your ICP will pay to solve it, the next challenge is getting enough genuine users using your product consistently to reach product-market fit. That requires not just one-time validation experiments but an ongoing feedback structure — real people, from your actual ICP, using your product over months and telling you what they experience.
first10 is built for exactly this transition. Founders who apply get matched with ICP-fit users who receive a free 12-month subscription and commit to monthly video feedback for the full year. The platform handles the matching, the incentive structure, and the feedback cadence — so you can focus on what the feedback is telling you, not on wrangling it out of unwilling testers.
Validation is how you decide what to build. first10 is how you find out whether you built it right.
Validation does not stop when you ship. Apply to first10 and get matched with ICP-fit genuine users who give structured monthly video feedback for a full year — so you keep learning long after your first experiments are done.