customer-discovery（客户发现）

Customer Discovery

Systematically validate your business hypotheses before building anything. Master Steve Blank's Customer Development methodology that became the foundation of Lean Startup and YC's approach.

When to Use This Skill

• Starting a new venture to avoid building something nobody wants
• Before writing a line of code to validate problem-solution fit
• Pivoting decisions to systematically test new directions
• Early-stage fundraising to prove market validation
• Product roadmap planning to prioritize based on validated customer needs
• Entering new markets to understand unfamiliar customer segments

Methodology Foundation

What Claude Does vs What You Decide

What This Skill Does

1. Documents business hypotheses - Turns assumptions into testable statements
2. Designs validation experiments - Creates tests that can prove you wrong
3. Structures customer conversations - Gets real insights, not false positives
4. Evaluates problem-solution fit - Determines if you should proceed, pivot, or stop
5. Creates learning roadmap - Prioritizes what to validate first
6. Tracks validation progress - Maintains evidence-based decision making

How to Use

Start Customer Discovery for a New Idea

I'm starting customer discovery for [business idea].
Help me document my hypotheses and design validation experiments.
Use Steve Blank's Customer Development methodology.

Evaluate Customer Discovery Progress

Here's what I've learned from [X] customer interviews: [summary]
Evaluate my progress against Customer Discovery criteria.
Should I proceed, pivot, or go back to more interviews?

Design a Validation Experiment

I need to validate this hypothesis: [hypothesis]
Design a Customer Discovery experiment to test it.
What would prove it true? What would prove it false?

Instructions

When helping with Customer Discovery, follow Steve Blank's systematic approach:

Step 1: Document Your Business Model Hypotheses

## Business Model Canvas Hypotheses

Before talking to customers, document what you BELIEVE to be true.
These are GUESSES until validated.

### Customer Hypotheses
| # | Hypothesis | Confidence | Evidence |
|---|------------|------------|----------|
| C1 | Our target customer is [WHO] | Low/Med/High | [None yet] |
| C2 | They have this problem: [WHAT] | Low/Med/High | [None yet] |
| C3 | The problem is severe because [WHY] | Low/Med/High | [None yet] |
| C4 | They currently solve it by [HOW] | Low/Med/High | [None yet] |

### Value Proposition Hypotheses
| # | Hypothesis | Confidence | Evidence |
|---|------------|------------|----------|
| V1 | Our solution provides [BENEFIT] | Low/Med/High | [None yet] |
| V2 | Customers care about [FEATURE/OUTCOME] | Low/Med/High | [None yet] |
| V3 | We're differentiated by [UNIQUE VALUE] | Low/Med/High | [None yet] |

### Channel Hypotheses
| # | Hypothesis | Confidence | Evidence |
|---|------------|------------|----------|
| H1 | We can reach customers through [CHANNEL] | Low/Med/High | [None yet] |
| H2 | Customer acquisition will cost [ESTIMATE] | Low/Med/High | [None yet] |

### Revenue Hypotheses
| # | Hypothesis | Confidence | Evidence |
|---|------------|------------|----------|
| R1 | Customers will pay [PRICE] | Low/Med/High | [None yet] |
| R2 | Revenue model: [TYPE] | Low/Med/High | [None yet] |
| R3 | Lifetime value: [ESTIMATE] | Low/Med/High | [None yet] |

Critical Rule: You don't have a business until these hypotheses are validated with real evidence from real customers.

Step 2: Prioritize What to Validate First

## Hypothesis Prioritization Matrix

Prioritize by RISK and IMPACT:

| Hypothesis | If WRONG, Impact | Current Confidence | Priority |
|------------|------------------|-------------------|----------|
| [H1] | Fatal/Major/Minor | Low/Med/High | P1/P2/P3 |
| [H2] | Fatal/Major/Minor | Low/Med/High | P1/P2/P3 |

### Priority Rules:
- P1: Fatal if wrong + Low confidence → Validate FIRST
- P2: Major if wrong + Low/Med confidence → Validate SOON
- P3: Minor if wrong OR High confidence → Validate LATER

### Your P1 Hypotheses (Validate These First):
1. ________________________________
2. ________________________________
3. ________________________________

These are your "leap of faith" assumptions.

Step 3: Design Validation Experiments

## Experiment Design Template

### Hypothesis Being Tested:
"We believe [CUSTOMER SEGMENT] has [PROBLEM] and would [BEHAVIOR]."

### Experiment Type:
- [ ] Customer Interviews (Problem Discovery)
- [ ] Solution Interviews (Solution Validation)
- [ ] Landing Page Test (Demand Validation)
- [ ] Concierge MVP (Solution Validation)
- [ ] Smoke Test (Demand Validation)

### Success Criteria:
**Validated if:** [Specific, measurable outcome]
**Invalidated if:** [Specific, measurable outcome]

Example:
- Validated: 8+ of 10 customers describe this exact problem unprompted
- Invalidated: Fewer than 5 of 10 mention this problem

### Sample Size:
- Minimum: [Number] customers
- Target segment: [Description]

### Data to Collect:
1.
2.
3.

### Timeline:
- Start: [Date]
- End: [Date]
- Go/No-Go Decision: [Date]

Step 4: Conduct Customer Discovery Interviews

## Customer Discovery Interview Framework

### PHASE 1: Problem Discovery (First 5-10 interviews)

**Goal:** Understand the problem space, NOT pitch solutions

**Interview Structure:**
1. **Context** (5 min): Their role, background, day-to-day
2. **Problem Exploration** (15 min): Deep dive into the problem area
3. **Existing Solutions** (5 min): What they use today
4. **Impact** (5 min): How the problem affects them
5. **Wrap-up** (5 min): Referrals, next steps

**Key Questions:**
- "Tell me about your biggest challenge with [area]..."
- "Walk me through the last time you dealt with [problem]..."
- "What solutions have you tried? What worked/didn't work?"
- "How much time/money does this cost you?"
- "If this magically disappeared, what would change?"

**DO NOT:**
- Pitch your solution
- Ask leading questions
- Talk more than 30% of the time
- Ask "would you use/pay for..."

### PHASE 2: Solution Discovery (Next 5-10 interviews)

**Goal:** Test if your solution addresses validated problems

**Only proceed to Phase 2 if:**
- [ ] You've talked to 10+ potential customers
- [ ] Clear problem patterns emerged
- [ ] Problems are severe and frequent enough
- [ ] Customers are actively seeking solutions

**Solution Interview Structure:**
1. **Recap** (3 min): Confirm the problem you heard
2. **Solution Demo** (10 min): Show concept/prototype
3. **Reaction** (10 min): Their honest feedback
4. **Commitment** (5 min): Would they try/buy/refer?

**Key Questions:**
- "Does this solve the problem you described?"
- "What's missing?"
- "What would you pay for this?"
- "Would you be willing to pilot this?"
- "Who else should see this?"

Step 5: Evaluate Results and Decide

## Customer Discovery Scorecard

### Problem Validation

| Criteria | Threshold | Your Result | Pass? |
|----------|-----------|-------------|-------|
| # of interviews completed | 10+ | | Y/N |
| % who confirmed the problem unprompted | 70%+ | | Y/N |
| Problem severity (1-10 avg) | 7+ | | Y/N |
| Already spending money on solutions | 50%+ | | Y/N |
| Actively seeking alternatives | 30%+ | | Y/N |

**Problem Validated:** All Y → Proceed to Solution Validation
**Problem NOT Validated:** Any N → Pivot or dig deeper

### Solution Validation

| Criteria | Threshold | Your Result | Pass? |
|----------|-----------|-------------|-------|
| # of solution interviews | 10+ | | Y/N |
| Said "this solves my problem" | 70%+ | | Y/N |
| Willing to try/pilot | 50%+ | | Y/N |
| Willing to pay (some amount) | 30%+ | | Y/N |
| Gave commitment (deposit, LOI, intro) | 3+ people | | Y/N |

**Solution Validated:** All Y → Proceed to MVP
**Solution NOT Validated:** Any N → Iterate or pivot

### Decision Framework

IF Problem NOT Validated: → PIVOT: Different customer segment? → PIVOT: Different problem? → STOP: No market here?

IF Problem Validated BUT Solution NOT Validated: → ITERATE: Different solution to same problem → ITERATE: Different features/approach

IF Both Validated: → BUILD: Minimum Viable Product → SELL: Pre-sales before building

Step 6: Document Learnings and Update Hypotheses

## Customer Discovery Learning Log

### Round [#] Summary

**Dates:** [Start] - [End]
**Interviews Conducted:** [#]
**Customer Segments Covered:** [List]

### Validated Hypotheses ✅
| Hypothesis | Evidence | Confidence Now |
|------------|----------|----------------|
| [H1] | [Quotes/Data] | High |

### Invalidated Hypotheses ❌
| Hypothesis | Evidence | Next Step |
|------------|----------|-----------|
| [H2] | [Quotes/Data] | Pivot to... |

### Surprises / New Insights 🔍
| Insight | Implication |
|---------|-------------|
| [Learning] | [What this means] |

### Updated Business Model
[What changed based on learnings]

### Next Validation Round
| Hypothesis to Test | Experiment | Timeline |
|--------------------|------------|----------|
| [H3] | [Method] | [Dates] |

Examples

Example 1: B2B SaaS - HR Tech Startup

Context: Founder believes mid-size companies need better employee onboarding software.

Input:

"I want to build HR onboarding software for mid-size companies. Help me set up Customer Discovery."

Output:

Business Model Hypotheses:

P1 Validation Plan:

Experiment 1: Problem Discovery

• Target: 15 HR managers at 100-500 employee companies
• Method: 30-minute phone interviews
• Questions:

1. "Walk me through what happens when you hire someone..." 2. "What's the biggest headache in that process?" 3. "How much time does onboarding take per new hire?" 4. "What tools are you using today?" 5. "What's the cost when onboarding goes poorly?"

Success Criteria:

• Validated if: 10+ of 15 describe manual, time-consuming process; 8+ mention specific pain points
• Invalidated if: Most say "onboarding is fine" or <4 hours per hire

Sample Interview Guide:

Opening: "I'm researching how mid-size companies handle employee onboarding.
Not selling anything - just learning. Mind if I ask some questions?"

Problem Exploration:
1. "When you hired your last employee, walk me through the onboarding process..."
2. "What took the most time?"
3. "Where do things typically go wrong?"
4. "Tell me about a time onboarding didn't go well..."

Current State:
5. "What tools or systems are you using for this?"
6. "What do you like about your current process?"
7. "What's frustrating about it?"

Impact:
8. "How does onboarding quality affect employee retention?"
9. "If you could wave a magic wand, what would change?"

Close:
10. "Who else should I talk to about this?"
11. "Can I follow up in a few weeks to share what I'm learning?"

Example 2: Consumer App - Fitness

Context: Founder wants to build a fitness app for busy parents.

Input:

"I did 12 customer discovery interviews. Here's my summary. Evaluate my progress."

Interview Summary:

• 12 interviews with working parents, ages 30-45
• 10/12 said they want to exercise more
• 8/12 said lack of time is the main barrier
• 6/12 have tried fitness apps before
• Only 2/12 are currently paying for any fitness solution
• 4/12 said they'd "probably" use a new app

Output:

Customer Discovery Evaluation:

Analysis:

Good Signs:

• Strong problem confirmation (want to exercise, no time)
• Consistent pattern across interviews

Warning Signs:

• Only 17% paying for anything (low willingness to pay)
• "Probably use" is weak signal (opinions, not commitment)
• 6 tried apps, only 2 still paying (high churn)

Missing Data:

• How severe is the pain? (scale 1-10)
• What have they tried recently?
• What would they pay? (past behavior, not predictions)

Diagnosis: Problem validated, but MARKET not validated.

The problem is real (busy parents want to exercise). But evidence suggests:

• People SAY they want this but don't PAY for it
• High awareness but low conversion/retention
• "Nice to have" not "must have"

Recommendation: Go Deeper Before Building

Next 5 interviews, focus on:

1. "Tell me about a fitness product you actually paid for and kept using..."
2. "What made you stop using [previous app]?"
3. "What's different about the 2 of you who are paying now vs. when you weren't?"
4. "If I told you this app costs $20/month, what's your reaction?"
5. "What would make fitness a MUST instead of NICE TO HAVE?"

Decision: Do not build yet. Either find a more acute pain point within this segment, or find a segment that demonstrates stronger buying behavior.

Example 3: Pivot Decision

Context: Founder validated problem but solution bombed.

Input:

"Problem validated: Small business owners hate bookkeeping. But when I showed my automated bookkeeping tool, response was lukewarm. 3 of 10 were excited. Should I pivot?"

Output:

Situation Analysis:

Questions Before Pivoting:

1. Who were the 3 excited customers?

- What made them different? - Same problem, different intensity? - Different sub-segment?

1. Why were 7 NOT excited?

- Wrong solution to right problem? - Right solution, wrong positioning? - Problem not severe enough for them? - Already have adequate solution?

1. What specifically was lukewarm?

- "Too complex" - "Already have something" - "Doesn't solve my real problem" - "Price too high"

Pivot Decision Framework:

┌─────────────────────────────────────────────────────────┐
│                 SOLUTION FAILED                          │
│                                                          │
│  Is the problem still validated?                         │
│  └── YES → Solution Pivot (keep problem, change solution)│
│                                                          │
│  Were some customers excited?                            │
│  └── YES → Segment Pivot (narrow to excited segment)     │
│                                                          │
│  Did feedback suggest different problem?                 │
│  └── YES → Problem Pivot (keep customer, find new problem)│
│                                                          │
│  Is there any signal at all?                             │
│  └── NO → Consider stopping or major pivot               │
└─────────────────────────────────────────────────────────┘

Recommended Next Steps:

1. Interview the 3 excited customers deeply

- What specifically excited them? - What would they pay? - Would they commit to pilot?

1. Understand the 7 "no"s

- What would have made them excited? - What are they using instead?

1. Consider these pivots:

- Segment Pivot: Target the excited segment only - Solution Pivot: Same problem, different solution (maybe simpler?) - Channel Pivot: Same solution, different delivery

1. Before building anything:

- Get 3 paying pilots or LOIs - "I'll build this if you prepay $X"

Checklists & Templates

Customer Discovery Readiness Checklist

## Before Starting Customer Discovery

### Preparation
- [ ] Business Model Canvas hypotheses documented
- [ ] P1 hypotheses identified (highest risk)
- [ ] Target customer segment defined
- [ ] Interview questions drafted
- [ ] 15+ potential interviews scheduled

### Mindset
- [ ] Committed to learning, not selling
- [ ] Ready to hear "your idea is wrong"
- [ ] Willing to pivot based on evidence
- [ ] No code written yet (or willing to throw away)

### Resources
- [ ] 2-4 weeks allocated for discovery
- [ ] Note-taking system ready
- [ ] Team aligned on validation criteria

Interview Tracking Template

## Customer Discovery Interview Log

| # | Date | Name | Company | Segment | Problem Confirmed? | Solution Fit? | Commitment? | Key Quote |
|---|------|------|---------|---------|-------------------|---------------|-------------|-----------|
| 1 | | | | | Y/N | Y/N/NA | | |
| 2 | | | | | Y/N | Y/N/NA | | |
| 3 | | | | | Y/N | Y/N/NA | | |

### Running Totals
- Interviews completed: __/15
- Problem confirmed: __%
- Solution fit (if applicable): __%
- Commitments received: __

Go/No-Go Decision Template

## Customer Discovery Decision Point

**Date:** _______________
**Total Interviews:** _______________
**Hypothesis Being Tested:** _________________________________

### Evidence Summary

| Metric | Target | Actual | Pass? |
|--------|--------|--------|-------|
| Problem confirmation rate | 70%+ | | |
| Average severity score | 7+/10 | | |
| Currently paying for alternatives | 50%+ | | |
| Solution fit (if tested) | 70%+ | | |
| Concrete commitments | 3+ | | |

### Strongest Evidence FOR:
1.
2.
3.

### Strongest Evidence AGAINST:
1.
2.
3.

### Decision

- [ ] **PROCEED:** Build MVP / Start selling
- [ ] **ITERATE:** More interviews needed, adjust approach
- [ ] **PIVOT:** Change [customer/problem/solution]
- [ ] **STOP:** No viable market here

### If Proceeding, Next Steps:
1.
2.
3.

### If Pivoting, New Hypothesis:

Skill Boundaries

What This Skill Does Well

• Structuring persuasive content
• Applying copywriting frameworks
• Creating draft variations
• Analyzing competitor approaches

What This Skill Cannot Do

• Guarantee conversion rates
• Replace brand voice development
• Know your specific audience
• Make final approval decisions

References

• Blank, Steve. "The Four Steps to the Epiphany" (2005) - Original methodology
• Blank, Steve & Dorf, Bob. "The Startup Owner's Manual" (2012) - Expanded playbook
• Ries, Eric. "The Lean Startup" (2011) - Build-Measure-Learn context
• Fitzpatrick, Rob. "The Mom Test" (2013) - Interview techniques
• Osterwalder, Alex. "Business Model Generation" (2010) - BMC framework
• Maurya, Ash. "Running Lean" (2012) - Lean Canvas approach

Related Skills

• mom-test - Interview techniques for honest feedback
• lean-canvas - One-page business model documentation
• jobs-to-be-done - Understanding customer progress
• problem-interview - Deep dive on problem discovery
• solution-interview - Testing solution fit

Skill Metadata (Internal Use)

name: customer-discovery
category: validation
subcategory: methodology
version: 1.0
author: MKTG Skills
source_expert: Steve Blank
source_work: The Four Steps to the Epiphany
difficulty: intermediate
estimated_value: $5,000 startup consulting engagement
tags: [customer-development, validation, startups, YC, lean-startup, interviews]
created: 2026-01-25
updated: 2026-01-25