---
title: How to validate a product idea | AskTheCustomer
description: Test your riskiest assumption first. A step-by-step guide to validating product ideas with existing evidence, customer interviews and low-cost experiments.
image: https://www.askthecustomer.ai/hubfs/blog-images/p1-og.png
---

[AskTheCustomer](https://www.askthecustomer.ai/)

[The problem](https://www.askthecustomer.ai/#problem) [How it works](https://www.askthecustomer.ai/#how) [Use cases](https://www.askthecustomer.ai/#use-cases) [Who it's for](https://www.askthecustomer.ai/who-its-for) [Blog](https://www.askthecustomer.ai/blog) [Early Access](https://www.askthecustomer.ai/early-access)

[Home](https://www.askthecustomer.ai/)/[Blog](https://www.askthecustomer.ai/blog)/Article

# How to validate a product idea before building it

OS [Olha Solonenko](https://www.askthecustomer.ai/blog/author/olha-solonenko) 4 October 2026 11 min read Updated 20 October 2026

![Social share image: How to validate a product idea before building it.](https://www.askthecustomer.ai/hs-fs/hubfs/blog-images/p1-og.png?width=1200&height=630&name=p1-og.png)

Most products fail because nobody needed them, not because they couldn't be built. Here's how to test the riskiest assumption first, using evidence you already have and cheap experiments.

Key takeaways

- No market need is the most-cited reason startups fail (42% of 101 CB Insights post-mortems).
- Split "validation" into four risks: value, usability, feasibility and business viability. Test value first.
- Start with evidence you already have: sales calls, support tickets, past research and reviews.
- Interview about past behaviour, not your idea. Commitments beat compliments.
- Climb the evidence ladder (say, do, pay) only as high as the decision needs.

*How to… · Product*

## The short answer

Validate the riskiest assumption first, using the cheapest method that produces real evidence. In practice that means four moves: write down what has to be true for the idea to work, check the customer evidence you already have, talk to customers about their past behaviour (not your idea), and only then run small tests that ask people to *do* something, like click, sign up, pay or switch.

The goal isn't to prove your idea is good. It's to find out, as cheaply as possible, whether it's wrong.

This guide walks through each step, compares the main methods by cost and strength of evidence, and ends with a scorecard you can copy.

## Why it matters: most products fail on demand, not engineering

When CB Insights analysed 101 startup post-mortems, the most-cited reason for failure was "no market need", named in 42% of them. That was well ahead of running out of cash (29%) and having the wrong team (23%). Founders could cite more than one reason, so the numbers don't add to 100%. ([CB Insights, via Angel Capital Association](https://angelcapitalassociation.org/data/Secure/Leadership%20Workshop%20Library/2017-leadership-workshop/20CausesofFailureScorecard.pdf?rev=EE89))

![Bar chart: no market need is the most-cited reason startups fail, at 42% of 101 post-mortems.](https://www.askthecustomer.ai/hs-fs/hubfs/blog-images/p1-failure-reasons.webp?width=1600&name=p1-failure-reasons.webp)

*Source: CB Insights analysis of 101 startup post-mortems.*

The same pattern shows up inside established companies. Pendo analysed usage across 615 software products and found that 80% of features are rarely or never used, while about 12% of features generate 80% of daily usage. ([Pendo, 2019 Feature Adoption Report](https://www.pendo.io/resources/the-2019-feature-adoption-report/))

There's a newer reason to take this seriously. AI coding tools have made building faster and cheaper than ever. When building is cheap, the expensive mistake is no longer "we couldn't build it". It's "we built the wrong thing quickly".

## Step 1: Name what you're actually testing

"Validate the idea" is too vague to act on. Marty Cagan of SVPG splits product risk into four questions, and each one needs a different kind of test. ([SVPG, The Four Big Risks](https://www.svpg.com/four-big-risks/))

![The four big product risks: value, usability, feasibility and business viability, with how to test each.](https://www.askthecustomer.ai/hs-fs/hubfs/blog-images/p1-four-risks.webp?width=1600&name=p1-four-risks.webp)

*The four big risks, adapted from Marty Cagan (SVPG).*

Cagan argues value risk is usually the hardest of the four, and that teams should tackle value and business viability early. Usability and feasibility usually need something built, even a rough prototype. Value and viability can often be tested with conversations, existing evidence and a landing page.

So before any test, write your idea as a short list of assumptions, one per line, in the form *"We believe \[who\] has \[problem\] and would \[do what\] to solve it."* Then mark each one: how much does the idea depend on it, and how much evidence do you already have? Start with the assumptions that matter most and have the least evidence.

## Step 2: Check the evidence you already have

Most teams skip this step and go straight to new research. That's a mistake, because the cheapest evidence is the evidence you've already collected. Before you book a single interview, spend an afternoon searching what customers have already told you.

Places to look:

- **Sales call recordings and notes.** Search for the problem in customers' words, not your feature name. Lost-deal notes are especially useful.
- **Support tickets and chat logs.** Look for workarounds, complaints and "is there a way to…" questions.
- **Past interviews and research reports.** Old research often answers new questions if you reread it with a specific assumption in mind.
- **Feature requests and CRM notes.** Count *who* asked, not just how often. Ten requests from one account are not ten signals.
- **Public reviews and communities.** Competitor reviews show what the wider market complains about.

Keep two lists as you go: what your own customers say, and what the wider market says. They often disagree, and the gap is useful. If your customers never mention a problem the market talks about constantly, either your customers are unusual or you're not reaching the people who have it.

This step rarely proves an idea. What it does is tell you which assumptions are already supported, which are contradicted, and which have no evidence at all. Those last ones are what you test next.

## Step 3: Talk to customers about their past, not your idea

The most common validation mistake is pitching the idea and asking "would you use this?" People are polite and optimistic about the future. They'll say yes, and then not buy.

Rob Fitzpatrick's book *The Mom Test* gives three rules that fix most bad interviews: talk about their life instead of your idea, ask about specific things that happened in the past rather than opinions about the future, and talk less while listening more. ([Summary via Unusual Ventures](https://www.unusual.vc/rob-fitzpatricks-mom-test/))

| Instead of asking… | Ask… |
| --- | --- |
| "Would you use a tool that does X?" | "Tell me about the last time you had to do X." |
| "Is this a big problem for you?" | "What did it cost you last time? Time, money, a missed deal?" |
| "How much would you pay for this?" | "What are you paying for today, including workarounds?" |
| "Do you like this design?" | "What have you already tried to solve this? Why did you stop?" |
| "Would your team adopt this?" | "Who else was involved last time? Who signed off?" |

Listen for three signals that matter more than enthusiasm:

- **They've already tried to solve it.** Spreadsheets, scripts, an agency, a competitor. Effort spent is the strongest sign a problem is real.
- **It has a cost they can name.** Hours, money, a lost customer, a delayed launch.
- **They offer something.** An intro to a colleague, a follow-up call, a pilot, a deposit. Fitzpatrick calls these commitments, and they're worth more than compliments.

In B2B, talk to more than one role. The person with the problem, the person who'd use the product and the person who pays are often three different people, and each can kill the idea for different reasons.

For help with the wording itself, see our guide on [how to write customer interview questions that aren't leading](https://www.askthecustomer.ai/blog/how-to-write-customer-interview-questions-not-leading).

## Step 4: Climb the evidence ladder

Not all evidence is equal. What people *say* is weaker than what they *do*, and what they do is weaker than what they *pay*. Each rung up costs more, so only climb as high as the decision needs.

![The evidence ladder, from existing evidence to pre-sales: what people say, do and pay.](https://www.askthecustomer.ai/hs-fs/hubfs/blog-images/p1-evidence-ladder.webp?width=1600&name=p1-evidence-ladder.webp)

*The evidence ladder: start at the bottom and stop when you have enough to decide.*

Here's how the common methods compare. Time and effort are rough guides for a small team, not benchmarks.

| Method | What it tells you | Best for risk | Typical effort | Evidence type |
| --- | --- | --- | --- | --- |
| Review existing evidence | Whether customers already raised the problem | Value | Hours | Past behaviour |
| Problem interviews (5–10 per segment) | How painful the problem is, what they do today | Value | 1–2 weeks | Past behaviour |
| Fake door / landing page | Whether people click or sign up for the promise | Value | Days | Action |
| Concierge test | Whether the outcome is worth it when delivered by hand, customer knows | Value, viability | 2–6 weeks | Action, sometimes payment |
| Wizard of Oz | Whether the experience works when humans run it behind the scenes | Usability, value | 2–4 weeks | Action |
| Prototype usability test | Whether people can figure it out | Usability | 1 week | Observed behaviour |
| Pre-sale, paid pilot or letter of intent | Whether they'll commit money or a contract | Value, viability | Weeks | Payment |

A few notes on using them well:

- **Fake doors** show an entry point to a feature that doesn't exist yet, such as a button or a pricing tier, and measure who tries it. Tell people honestly what happened when they click, and don't overuse them, or users stop trusting your product. ([Scrum.org on product experiments](https://www.scrum.org/node/89427))
- **Concierge and Wizard of Oz** are often confused. In a concierge test the customer knows a human is doing the work. In Wizard of Oz the interface looks automated. Concierge suits workflow products; Wizard of Oz suits testing the experience. ([Thorgate](https://blog.thorgate.eu/blog/3-ways-how-validate-your-startup-idea-minimal-investment/))
- **Usability tests** need fewer people than you think. Jakob Nielsen's research found that about five users per distinct user group reveal roughly 85% of usability problems on a focused flow. Test five, fix, then test again. ([Nielsen Norman Group](https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/))

## Common traps, and how to avoid them

- **Collecting compliments.** "Love it!" is not evidence. Ask what they did the last time the problem came up, or ask for a small commitment.
- **Asking only your friendliest customers.** Your champions will say yes to almost anything. Include churned customers, lost deals and people who've never heard of you.
- **Treating one loud account as a pattern.** Note how many *different* customers raised something, and in which segments. Our guide on [how to prioritise feature requests from customers](https://www.askthecustomer.ai/blog/how-to-prioritise-feature-requests-from-customers) shows how to count them.
- **Mixing segments.** An idea can be essential for enterprise IT admins and irrelevant to small-business owners. Look at evidence per segment and role, not as one blended average.
- **Blending your customers with the market.** Internet opinion is useful context, but it isn't your customers. Keep the two sources separate so you can see where they agree and where they don't.
- **Testing the easy assumption.** Teams often test usability because it's comfortable, while the real risk is whether anyone needs the thing. Test the riskiest assumption first.
- **Never deciding.** Set your pass/fail criteria *before* you run the test, such as "8 of 10 interviewees describe a workaround they use today". Otherwise any result can be read as a yes.

## A validation scorecard you can copy

Use this to keep everyone honest about what you know and what you're guessing. Replace the example assumptions with your own, set the evidence level for each (Strong, Some, None yet or Contradicted), and decide your pass criteria before testing.

| Assumption | Risk | Evidence so far | Where it came from | Next test | Pass if… |
| --- | --- | --- | --- | --- | --- |
| \[Segment\] has \[problem\] often enough to care | Value |  |  | Problem interviews | Most interviewees describe a recent example unprompted |
| They already spend time or money on a workaround | Value |  |  | Interviews, support tickets | Most can name the workaround and its cost |
| The person with the problem can influence the purchase | Viability |  |  | Interviews with buyers and users | Buyer and user agree it's a priority |
| They'll act on the promise, not just praise it | Value |  |  | Fake door or landing page | Agreed click or sign-up rate from the right audience |
| They can use the first version without help | Usability |  |  | Prototype test, 5 per segment | Most complete the core task unaided |
| We can reach and sell to them affordably | Viability |  |  | Small outbound or ad test | Cost per qualified lead within budget |

If an assumption is marked "Contradicted", stop and rethink before testing anything further downstream.

## Frequently asked questions

### How many customer interviews do I need to validate an idea?

There's no magic number. Many teams start with 5 to 10 per segment and stop when answers become repetitive. For usability testing specifically, Nielsen Norman Group's research suggests about five people per distinct user group catches most problems on a focused flow.

### How long should validation take?

It depends on the size of the bet. Reviewing existing evidence takes hours, a round of interviews one or two weeks, and a concierge test or paid pilot several weeks. Match the effort to what you'd lose if you were wrong.

### Can I validate a product idea without talking to customers?

Partly. Existing evidence, competitor reviews and fake door tests can tell you a lot. But they rarely tell you *why* people behave as they do, so most ideas still need at least a handful of real conversations.

### Can AI or synthetic personas validate my idea?

Not on their own. A persona invented from a prompt reflects the model's assumptions, not your customers. AI is useful for synthesising the evidence you already have, preparing interview questions and spotting gaps. Treat its output as a hypothesis to check, not as proof. We compare the two approaches in [evidence-based vs prompt-generated personas](https://www.askthecustomer.ai/blog/evidence-based-vs-prompt-generated-personas).

### What's the difference between validation and an MVP?

A validation experiment tests whether an idea is worth building, often without building anything. An MVP is a first working version with only the essential features. Validate first so your MVP tests the right thing.

### What if the evidence is mixed?

Split it by segment and role. Mixed results often mean the idea is strong for one group and weak for another, which is a useful finding in itself.

## Sources

- [CB Insights: Top 20 reasons startups fail (summary via Angel Capital Association)](https://angelcapitalassociation.org/data/Secure/Leadership%20Workshop%20Library/2017-leadership-workshop/20CausesofFailureScorecard.pdf?rev=EE89)
- [Pendo: 2019 Feature Adoption Report](https://www.pendo.io/resources/the-2019-feature-adoption-report/)
- [Marty Cagan, SVPG: The Four Big Risks](https://www.svpg.com/four-big-risks/)
- [Unusual Ventures: Takeaways from The Mom Test](https://www.unusual.vc/rob-fitzpatricks-mom-test/)
- [Scrum.org: How to craft effective product experiments](https://www.scrum.org/node/89427)
- [Thorgate: Three ways to validate your startup idea](https://blog.thorgate.eu/blog/3-ways-how-validate-your-startup-idea-minimal-investment/)
- [Nielsen Norman Group: Why you only need to test with 5 users](https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/)

*Last updated: 20 October 2026*

Don't guess. Ask the customer.

OS

### Olha Solonenko

Olha Solonenko is the founder of AskTheCustomer, an evidence-based AI customer persona platform for B2B teams. Based in Ireland, she works on helping product, marketing and sales teams use real customer evidence in everyday decisions.

[Connect on LinkedIn](https://www.linkedin.com/in/olha-solonenko-3630684/)

On this page

Early Access

## Build personas from *real customer evidence.*

AskTheCustomer turns your interviews, calls and tickets into personas you can question, with sources for every answer. Early Access is free.

[Join Early Access →](https://www.askthecustomer.ai/early-access) [Is it a fit for your team?](https://www.askthecustomer.ai/who-its-for)

[AskTheCustomer](https://www.askthecustomer.ai/)

Stop guessing. *Ask the customer.*

© 2026 AskTheCustomer · Built in Ireland  
Page updated October 2026

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Olha Solonenko",
    "url" : "https://www.askthecustomer.ai/blog/author/olha-solonenko"
  },
  "dateModified" : "2026-10-20T00:46:22.098Z",
  "datePublished" : "2026-10-04T18:50:44.000Z",
  "headline" : "How to validate a product idea | AskTheCustomer",
  "image" : [ "https://www.askthecustomer.ai/hubfs/blog-images/p1-og.png" ],
  "mainEntityOfPage" : {
    "@id" : "https://www.askthecustomer.ai/blog/how-to-validate-a-product-idea-before-building-it",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject"
    }
  }
}
```

```json
{
  "@context" : "https://schema.org",
  "@type" : "FAQPage",
  "mainEntity" : [ {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "There's no magic number. Many teams start with 5 to 10 per segment and stop when answers become repetitive. For usability testing specifically, Nielsen Norman Group's research suggests about five people per distinct user group catches most problems on a focused flow."
    },
    "name" : "How many customer interviews do I need to validate an idea?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "It depends on the size of the bet. Reviewing existing evidence takes hours, a round of interviews one or two weeks, and a concierge test or paid pilot several weeks. Match the effort to what you'd lose if you were wrong."
    },
    "name" : "How long should validation take?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "Partly. Existing evidence, competitor reviews and fake door tests can tell you a lot. But they rarely tell you why people behave as they do, so most ideas still need at least a handful of real conversations."
    },
    "name" : "Can I validate a product idea without talking to customers?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "Not on their own. A persona invented from a prompt reflects the model's assumptions, not your customers. AI is useful for synthesising the evidence you already have, preparing interview questions and spotting gaps. Treat its output as a hypothesis to check, not as proof."
    },
    "name" : "Can AI or synthetic personas validate my idea?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "A validation experiment tests whether an idea is worth building, often without building anything. An MVP is a first working version with only the essential features. Validate first so your MVP tests the right thing."
    },
    "name" : "What's the difference between validation and an MVP?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "Split it by segment and role. Mixed results often mean the idea is strong for one group and weak for another, which is a useful finding in itself."
    },
    "name" : "What if the evidence is mixed?"
  } ]
}
```

```json
{
  "@context" : "https://schema.org",
  "@id" : "https://www.askthecustomer.ai/blog/how-to-validate-a-product-idea-before-building-it#article",
  "@type" : "BlogPosting",
  "author" : {
    "@id" : "https://www.askthecustomer.ai/#olha-solonenko",
    "@type" : "Person",
    "description" : "Olha Solonenko is the founder of AskTheCustomer, an evidence-based AI customer persona platform for B2B teams. Based in Ireland, she works on helping product, marketing and sales teams use real customer evidence in everyday decisions.",
    "name" : "Olha Solonenko",
    "sameAs" : [ "https://www.linkedin.com/in/olha-solonenko-3630684/" ],
    "url" : "https://www.linkedin.com/in/olha-solonenko-3630684/",
    "worksFor" : {
      "@id" : "https://www.askthecustomer.ai/#organization"
    }
  },
  "dateModified" : "2026-10-20",
  "datePublished" : "2026-10-04",
  "description" : "Test your riskiest assumption first. A step-by-step guide to validating product ideas with existing evidence, customer interviews and low-cost experiments.",
  "headline" : "<span id=\"hs_cos_wrapper_name\" class=\"hs_cos_wrapper hs_cos_wrapper_meta_field hs_cos_wrapper_type_text\" style=\"\" data-hs-cos-general-type=\"meta_field\" data-hs-cos-type=\"text\" >How to validate a product idea before building it</span>",
  "image" : {
    "@type" : "ImageObject",
    "height" : 630,
    "url" : "https://149472448.fs1.hubspotusercontent-eu1.net/hubfs/149472448/blog-images/p1-og.png",
    "width" : 1200
  },
  "inLanguage" : "en",
  "isPartOf" : {
    "@id" : "https://www.askthecustomer.ai/blog#blog",
    "@type" : "Blog",
    "name" : "The AskTheCustomer Blog",
    "url" : "https://www.askthecustomer.ai/blog"
  },
  "mainEntityOfPage" : "https://www.askthecustomer.ai/blog/how-to-validate-a-product-idea-before-building-it",
  "publisher" : {
    "@id" : "https://www.askthecustomer.ai/#organization"
  },
  "timeRequired" : "PT11M",
  "url" : "https://www.askthecustomer.ai/blog/how-to-validate-a-product-idea-before-building-it",
  "wordCount" : 2224
}
```

```json
{
  "@context" : "https://schema.org",
  "@id" : "https://www.askthecustomer.ai/#organization",
  "@type" : "Organization",
  "logo" : {
    "@type" : "ImageObject",
    "height" : 512,
    "url" : "https://149472448.fs1.hubspotusercontent-eu1.net/hubfs/149472448/atc-logo.png",
    "width" : 512
  },
  "name" : "AskTheCustomer",
  "url" : "https://www.askthecustomer.ai/"
}
```

```json
{
  "@context" : "https://schema.org",
  "@type" : "BreadcrumbList",
  "itemListElement" : [ {
    "@type" : "ListItem",
    "item" : "https://www.askthecustomer.ai/",
    "name" : "Home",
    "position" : 1
  }, {
    "@type" : "ListItem",
    "item" : "https://www.askthecustomer.ai/blog",
    "name" : "Blog",
    "position" : 2
  }, {
    "@type" : "ListItem",
    "item" : "https://www.askthecustomer.ai/blog/how-to-validate-a-product-idea-before-building-it",
    "name" : "<span id=\"hs_cos_wrapper_name\" class=\"hs_cos_wrapper hs_cos_wrapper_meta_field hs_cos_wrapper_type_text\" style=\"\" data-hs-cos-general-type=\"meta_field\" data-hs-cos-type=\"text\" >How to validate a product idea before building it</span>",
    "position" : 3
  } ]
}
```