customer-discovery

客户探索

在着手构建任何东西之前，系统地验证你的业务假设。掌握 Steve Blank 的客户开发方法论，它后来成为精益创业和 YC 方法的基础。

何时使用此技能

• 启动新业务，避免构建无人需要的产品
• 编写任何代码之前，验证问题与解决方案的匹配度
• 决定是否转型，系统地测试新的方向
• 早期融资，证明市场已得到验证
• 规划产品路线图，根据已验证的客户需求确定优先级
• 进入新市场，了解陌生的客户群体

方法论基础

方面	详情
来源	Steve Blank - 《The Four Steps to the Epiphany》（2005 年）
核心原则	“没有任何商业计划能经受住与客户的首次接触。走出办公室。”
为何重要	创业公司失败，是因为它们构建了无人需要的产品。客户探索能在资金耗尽之前，以系统化学习取代猜测。

Claude 负责什么，你来决定什么

Claude 负责	你来决定
构建内容框架	最终文案
提出说服技巧	品牌语气
创建不同版本的草稿	选择版本
发现优化机会	发布时间
分析竞争对手的方法	战略方向

此技能的功能

1. 记录业务假设 - 将假设转化为可测试的陈述
2. 设计验证实验 - 创建能够证明你判断错误的测试
3. 构建客户对话 - 获取真实洞察，而非虚假结论
4. 评估问题与解决方案的匹配度 - 判断应当继续、转型还是停止
5. 制定学习路线图 - 确定优先验证的事项
6. 跟踪验证进展 - 保持基于证据的决策

使用方法

为新想法启动客户探索

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

评估客户探索进展

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?

设计验证实验

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?

说明

协助进行客户探索时，请遵循 Steve Blank 的系统化方法：

第 1 步：记录你的商业模式假设

## 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] |

重要规则： 只有当这些假设得到真实客户提供的真实证据验证后，你才真正拥有一项业务。

---

第 2 步：确定优先验证事项

## 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.

---

第 3 步：设计验证实验

## 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]

---

第 4 步：开展客户探索访谈

## 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?"

---

第 5 步：评估结果并作出决定

## 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: → 转型：换一个客户群体？ → 转型：换一个问题？ → 停止：这里没有市场？

IF Problem Validated BUT Solution NOT Validated: → 迭代：为同一问题尝试不同解决方案 → 迭代：尝试不同功能或方法

IF Both Validated: → 构建：最小可行产品 → 销售：在构建之前进行预售

---

第 6 步：记录经验并更新假设

## 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] |

---

示例

示例 1：B2B SaaS——人力资源科技初创公司

背景： 创始人认为，中型公司需要更好的员工入职软件。

输入：

“我想为中型公司打造人力资源入职软件。请帮我开展客户探索。”

输出：

商业模式假设：

#	假设	如果错误的风险	优先级
C1	员工人数为 100-500 人的公司由人力资源经理负责管理员工入职	致命	P1
C2	目前的入职流程依赖人工、耗时较长（每位新员工 5 小时以上）	致命	P1
C3	糟糕的入职体验会导致员工离职，而他们对此很在意	致命	P1
V1	自动化工作流可为每位新员工节省 3 小时以上	重大	P2
V2	他们愿意为此支付每位员工每月 50-100 美元	重大	P2
H1	我们可以通过 LinkedIn 联系到人力资源经理	轻微	P3

P1 验证计划：

实验 1：问题探索

• 目标：访谈 15 位来自员工人数为 100-500 人公司的 HR 经理
• 方法：30 分钟电话访谈
• 问题：
  1. “请带我了解一下，你们招聘新员工时会发生什么……”
  2. “这个流程中最让你头疼的是什么？”
  3. “每位新员工的入职流程要花多少时间？”
  4. “你们现在使用哪些工具？”
  5. “入职流程不顺利会造成什么成本？”

成功标准：

• 验证通过：15 人中有 10 人以上描述了依赖人工、耗时较长的流程；8 人以上提到了具体痛点
• 验证未通过：大多数人表示“入职流程没问题”，或每位员工花费的时间少于 4 小时

访谈指南示例：

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?"

---

示例 2：消费者应用——健身

背景： 创始人想为忙碌的家长打造一款健身应用。

输入：

“我做了 12 次客户探索访谈。以下是我的总结。请评估我的进展。”

访谈总结：

• 访谈了 12 位年龄为 30-45 岁的在职家长
• 12 人中有 10 人表示想多锻炼
• 12 人中有 8 人表示缺少时间是主要障碍
• 12 人中有 6 人之前使用过健身应用
• 目前只有 12 人中的 2 人在为任何健身解决方案付费
• 12 人中有 4 人表示“可能”会使用一款新应用

输出：

客户探索评估：

标准	阈值	你的结果	通过？
已完成访谈	10 次以上	12	✅
未经提示便确认问题	70% 以上	83%（10/12）	✅
问题严重程度	7 以上	未知	⚠️
为解决方案付费	50% 以上	17%（2/12）	❌
正在积极寻找替代方案	30% 以上	未知	⚠️

分析：

积极信号：

• 对问题的确认度很高（想锻炼，但没有时间）
• 不同访谈中呈现出一致的模式

警示信号：

• 只有 17% 的人愿意为任何方案付费（付费意愿低）
• “可能会用”是较弱的信号（表达看法，而非实际承诺）
• 6 人试用过应用，只有 2 人仍在付费（流失率高）

缺失的数据：

• 痛点有多严重？（1-10 分）
• 他们最近尝试过什么？
• 他们愿意支付多少？（了解过去行为，而非预测）

**诊断：**问题已验证，但市场尚未验证。

问题确实存在（忙碌的家长想锻炼）。但证据表明：

• 人们表示想要，但并不愿意为此付费
• 认知度高，但转化率和留存率低
• 这是“锦上添花”，而非“不可或缺”

建议：深入探索后再开发

接下来 5 次访谈应重点关注：

1. “请告诉我，你实际付费购买并持续使用过的一款健身产品……”
2. “是什么让你停止使用[之前的应用]？”
3. “你们两位现在愿意付费，与之前不愿意时有什么不同？”
4. “如果我告诉你这款应用每月收费 20 美元，你会有什么反应？”
5. “什么因素能让健身成为‘不可或缺’，而不是‘锦上添花’？”

决策：暂时不要开发。 要么在这个细分群体中找到更突出的痛点，要么找到表现出更强购买行为的群体。

---

示例 3：转向决策

背景： 创始人已验证问题，但解决方案大获失败。

输入：

“问题已验证：小企业主讨厌做账。但当我展示自动化做账工具时，反响平平。10 人中只有 3 人感到兴奋。我应该转向吗？”

输出：

情况分析：

阶段	结果
问题发现	✅ 已验证
解决方案发现	❌ 失败（30% < 70% 阈值）

转向前的问题：

1. 那 3 位感到兴奋的客户是谁？

   • 他们有什么不同？
   • 面临的是同一个问题，只是程度不同？
   • 属于不同的细分群体？

2. 为什么另外 7 人不感到兴奋？

   • 针对正确问题提出了错误的解决方案？
   • 解决方案正确，但定位不当？
   • 对他们来说，问题不够严重？
   • 已经有了合适的解决方案？

3. 具体是什么让他们反响平平？

   • “太复杂”
   • “已经有解决办法了”
   • “没解决我的实际问题”
   • “价格太高”

转向决策框架：

┌─────────────────────────────────────────────────────────┐
│                 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               │
└─────────────────────────────────────────────────────────┘

建议的后续步骤：

1. 深入访谈那 3 位感到兴奋的客户

   • 具体是什么让他们感到兴奋？
   • 他们愿意支付多少钱？
   • 他们愿意承诺参与试点吗？

2. 了解那 7 个“不”

   • 什么情况会让他们感到兴奋？
   • 他们在用什么替代方案？

3. 考虑以下转向方式：

   • 细分群体转向： 只面向感到兴奋的细分群体
   • 解决方案转向： 同一个问题，不同的解决方案（或许更简单？）
   • 渠道转向： 同一个解决方案，不同的交付方式

4. 在着手开发之前：

   • 获得 3 个付费试点或 LOI
   • “如果你们预付 $X，我就开发这个”

---

检查清单与模板

客户发现准备情况检查清单

## 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

---

访谈跟踪模板

## 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: __

---

是否推进决策模板

## 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:

---

技能边界

此技能擅长的方面

• 构建有说服力的内容
• 运用文案撰写框架
• 创建不同版本的草稿
• 分析竞争对手的做法

此技能无法做到的事

• 保证转化率
• 取代品牌语气的塑造
• 了解你的特定受众
• 做出最终审批决定

参考资料

• Blank, Steve. “The Four Steps to the Epiphany”（2005）——原始方法论
• Blank, Steve & Dorf, Bob. “The Startup Owner's Manual”（2012）——扩展版操作指南
• Ries, Eric. “The Lean Startup”（2011）——构建-衡量-学习的背景
• Fitzpatrick, Rob. “The Mom Test”（2013）——访谈技巧
• Osterwalder, Alex. “Business Model Generation”（2010）——BMC 框架
• Maurya, Ash. “Running Lean”（2012）——Lean Canvas 方法

相关技能

• mom-test - 获取真实反馈的访谈技巧
• lean-canvas - 单页商业模式文档
• jobs-to-be-done - 理解客户进展
• problem-interview - 深入探索问题发现
• solution-interview - 测试解决方案契合度

---

技能元数据（内部使用）

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