Entrepreneurship KB

Powered by Agentic Workers

The Validation Stack: From Hypothesis to Product-Market Fit

4 min read

The Validation Stack: From Hypothesis to Product-Market Fit

A synthesis of validation frameworks, mapping the journey from raw idea to proven product-market fit. Each layer builds on the one below it — skip a layer and the whole stack collapses.

Layer 1: The Idea Hypothesis

Framework: Paul Graham’s ideation principles

Every startup begins with a hypothesis: “I believe [these people] have [this problem] and will [pay/use] [this solution].”

The strongest hypotheses come from personal experience — Graham’s “organic ideas” that are noticed, not invented. The Well test filters early: “Who wants this so badly they’ll use a rough version from unknown founders?”

Output: A clearly stated hypothesis about customer, problem, and solution. Risk: The hypothesis is wrong. (This is expected — most are.)

Layer 2: Customer Discovery

Framework: Steve Blank’s customer-development

“There are no facts inside the building.” Take your hypothesis outside and test it against reality through structured customer conversations.

Key activities:

  • Interview 30-50 potential customers before writing code
  • Test each component of the hypothesis separately (Is this the right customer? The right problem? The right solution?)
  • Track what surprises you — surprises reveal where your hypothesis is wrong
  • Use the scientific method: hypothesize → experiment → learn → iterate

Output: A validated (or invalidated) understanding of your customer and their problem. Risk: Confirmation bias — hearing what you want to hear instead of what customers actually say.

Layer 3: The Minimum Viable Product

Framework: Eric Ries’ lean-startup methodology

Build the simplest possible experiment to test whether your solution actually solves the validated problem. The minimum-viable-product is NOT a low-quality product — it’s the minimum experiment needed for maximum learning.

Types of MVPs (from least to most effort):

  1. Landing page — Test demand by measuring signups
  2. Concierge — Deliver the service manually to validate the value proposition
  3. Wizard of Oz — Looks automated, but humans do the work behind the scenes
  4. Piecemeal — Combine existing tools to simulate the experience
  5. Single-feature — Build only the core feature (Instagram stripping Burbn to photo-sharing)

The Build-Measure-Learn loop:

  1. Identify the riskiest assumption
  2. Build the minimum experiment to test it
  3. Measure with actionable metrics (not vanity metrics)
  4. Learn: pivot or persevere?

Output: Evidence that customers will use (and ideally pay for) your solution. Risk: Building too much before testing, or testing the wrong assumption.

Layer 4: Product-Market Fit

Framework: Andreessen’s product-market-fit + Vohra’s PMF Engine

The moment when your product satisfies strong market demand. Andreessen says you can “always feel” it — but Vohra’s engine lets you measure it.

The Sean Ellis Test

Ask users who’ve engaged at least twice: “How would you feel if you could no longer use [product]?”

  • ≥40% “very disappointed” = you have PMF
  • <40% = keep iterating

The PMF Engine (4 steps)

  1. Segment: Find which user personas score highest → define your High-Expectation Customer
  2. Analyze: Study what “very disappointed” users love; focus on “somewhat disappointed” users who share your core value prop
  3. Roadmap: 50% doubling down on strengths, 50% removing barriers for the “somewhat disappointed” group
  4. Repeat: Survey continuously, rebuild roadmap quarterly

Rachleff’s Value vs Growth Hypothesis

  • Value hypothesis: What to build, for whom, with what business model
  • Growth hypothesis: How to cost-effectively acquire customers
  • Value MUST come before growth. Growth without value = flameout.

Output: A product that a defined market genuinely needs, with a quantitative score to prove it. Risk: Premature scaling — trying to grow before you’ve truly achieved fit.

The Stack in Practice

LayerQuestionMethodMetric
1. IdeaIs this worth exploring?Well test, personal experienceFounder conviction
2. DiscoveryDo customers have this problem?30-50 interviewsPattern recognition
3. MVPWill they use our solution?Build-Measure-LearnUsage, willingness to pay
4. PMFDo they love it?Sean Ellis survey≥40% “very disappointed”

Common Failure Patterns

  • Skipping Layer 2: Building before talking to customers → building something nobody wants
  • Skipping Layer 3: Going straight from interviews to full product → wasting months/years
  • Declaring PMF prematurely: Early traction from friends/network ≠ real PMF
  • Reversing the order: Trying to grow (Layer 4+) before validating (Layers 1-3)
  • Ignoring the data: Continuing to build despite evidence that the hypothesis is wrong (pivoting is not failure — it’s the validation stack working as designed)

See Also

Sources

Graph View

Backlinks