Truman: "When running a startup,business model is not our top priority — demand analysis is.. "
Yitang / Complete Five-Step Method System (Key Assumptions + Business Milestones + 5 Advanced Levels of the Five-Step Method + Implementation) The in-depth operating system for the first step "Demand".The four-character formula "Decompose-Deduce-Evaluate-Calculate" plus six proficiency levels.
1. Overview of the Four-Character Formula
| Character | Session | Core Action |
|---|---|---|
| Decompose | Practical 1: Peel-and-Decompose Chapter | Break down vague demands into specific scenarios + user groups + pain points |
| Deduce | Practical 2: Scenario Deduction Chapter | Use JTBD to deduce the task map and gain insight into real demands |
| Evaluate | Practical 3: Qualitative Evaluation Chapter | Judging good demands across 3 levels: beginner → intermediate → advanced |
| Calculate | Practical 4: Calculate the Ceiling Chapter | Commercial space sizing: how big the business can grow |
3 Values of Demand Exploration
- Multiple-choice question — Validate/falsify as early as possible, low-cost trial and error (decides whether to do it)
- Design question — Comprehensively improve the quality of all decisions (decides how to do it)
- Math question — Build conviction and understand the business space (decides the scale)
"It's so uncomfortable to feel like you're carrying a time bomb all the time when starting a business. Spend some time in advance researching the market and demand,build conviction about user problems and the ceiling space of the business,and your entire management mindset will be much more at ease."
Six Proficiency Levels of Demand Analysis (From Beginner to Advanced)
| Level | State | Training Before the Leap |
|---|---|---|
| L1 | Does not distinguish between demand / product / feature | — |
| L2 | Knows demand is important, but does not know how to decompose it | Decompose |
| L3 | Able to decompose and deduce | Decompose + Deduce |
| L4 | Able to evaluate good demands | Evaluate |
| L5 | Able to calculate the ceiling | Calculate |
| L6 | Holistic strategy, demand drives product / growth / moats | Comprehensive |
3 Capability Leaps:
- 1st Leap(Vague → Decomposed): Core training Decompose
- 2nd Leap(Simple → Professional): Core training Deduce + Evaluate
- 3rd Leap(Local → Holistic): Core training Calculate + Holistic Strategy
2. Stage 1: Peel-and-Decompose (Practical 1)
The core foundational skill for the 1st leap.
2 Core Suggestions
- What you want to provide does not matter; what matters is what specific problems users want to solve
- Users are never a monolith — you can try to segment them
2 Decomposition Principles
| Principle | Meaning |
|---|---|
| Cut each cell 2–5 times | Cut until you reach specific, clear, real demands |
| User group × Scenario × Pain point × Frequency Decompose | 4-dimensional cross-analysis |
6 commonly used decomposition cheat sheets
(For the specific 6 tables, see the supporting candy: One-Page Template "From Task Insight to In-Depth Demand Deduction")
| # | Dimension |
|---|---|
| 1 | By User group(age / gender / income / occupation / region) |
| 2 | By Scenario(time / location / trigger condition) |
| 3 | By Pain point severity |
| 4 | By Frequency(daily / weekly / monthly / yearly) |
| 5 | By Amount(average order value + LTV) |
| 6 | By Existing alternative solutions(direct competitors + indirect competitors) |
III. Stage 2: Scenario Deduction (Hands-on Practice 2) — JTBD Framework
Truman: "Excellent business leaders and product managers must develop top-tier 'micro-level perception'. "
Micro-level perception
Definition: having extremely sharp intuition about the entire process of users using the product,.
How to develop it(3–6 months focusing on one specific demand scenario):
- Use it yourself 100 times
- Use it together with users 10 times
- Watch users use it 100 times
- Review the details of every session
- Have in-depth conversations with the most extreme users (those who love it most + those who hate it most)
JTBD (Jobs To Be Done, task-driven) — Yitang localized version
Core idea: Users do not 'buy a product'; they 'hire a product to get a job done.'
Classic example: Users do not buy 'a quarter-inch drill bit'; they hire the drill bit to make a hole in the wall to hang a photo frame; while the real job might be'making family members feel the home is warm'.
L1–L4 Job Insight Canvas (supporting candy one-page template)
L1+L2 初步粗拆解
↓
L3 圈定具体任务
↓
L4 分解任务地图
L1–L2 Initial rough decomposition
| Segment | My answer (hypothesis to be verified) |
|---|---|
| Demand | List 3–5 items |
| Scenario | List 3–5 items |
| Problem | List 3–5 items |
Cut each cell 2–5 times to find the most authentic and clear demands.
L3 Define the specific job
| Sort out | Suggestions |
|---|---|
| Core executor | The person who feels the most anxious, the greatest pain, and the strongest need + other participants / supporters / payers / beneficiaries |
| Core job | Aspiration and vision → Big job → Small job → Micro job |
| Alternative products | Direct competitors + indirect competitors |
Final one-sentence job description format:[动词] + [对象] + [情境约束]
Example:
- ❌ "Purchase an AI coach app"
- ✅ "Eliminate + the gap in parenting concepts + when relying on the elderly to help take care of children"
Verb List: eliminate / ensure / balance / optimize / improve / reduce / prevent / accelerate
L4 Task Map Decomposition
Map = The complete execution process from task start to finish, with 4–8 steps.
Experience each step's:
- Action state
- Anxiety and pain points
- Potential improvement opportunities
AI Collaboration Version
Newly added by Truman in this course iteration: Feed the L1–L4 canvas to AI (Claude / ChatGPT), and let AI help you decompose + deduce + find blind spots.
Prompt Mode: Give the task definition and scenario to AI → AI lists 30 hypotheses to be verified → You filter → Second-round deduction.
IV. Stage 3: Qualitative Evaluation (Hands-on Practice 3)
3 tiers: Beginner / Intermediate / Advanced
| Tier | Capability | Tool |
|---|---|---|
| Beginner | Judge "Is this demand valid?" | 4-dimension basic evaluation (rigidity / frequency / scale / timing) |
| Intermediate | Judge "Among multiple demand options, which is the best to choose?" | Cross-comparison matrix for multiple requirements |
| Advanced | Judge "Should we bet on high-risk, high-return projects?" | Hypothesis ranking + key evidence collection |
4-Dimension Basic Evaluation (Beginner)
| Dimension | Key Question | Score |
|---|---|---|
| Rigidity | What happens if it is not solved? | 1-10 |
| Frequency | How often does it occur? | Daily / Weekly / Monthly / Yearly |
| Scale | How many people have this problem? | Ten thousand / Hundred thousand / Million / Ten million |
| Timing | Is there a difference between solving it now and solving it 5 years later? | Must do now / Not urgent |
Counterexamples of "looks good on the surface": Everyone thinks it is a good demand, but decomposition reveals weak rigidity / low frequency / small scale — kill it in the bud.
Cross-Comparison Matrix for Multiple Requirements (Intermediate)
| 需求 A | 需求 B | 需求 C |
刚性 | 8 | 6 | 9 |
频率 | 日 | 月 | 日 |
规模 | 100w | 10w | 50w |
时机 | 现在 | 不急 | 现在 |
能力匹配| 高 | 高 | 中 |
团队意愿| 中 | 高 | 高 |
Truman's decision habit:First filter with the 4-dimension evaluation, then compare horizontally with the 6-dimension matrix.
Advanced: Requirement Decisions for High-Risk Projects
Key questions: Is this demand a 60% certainty or 80%? How to obtain key evidence at the lowest cost?
3 major strategies:
- Extreme user interviews — 10 users who love it most + 10 users who hate it most
- Behavioral data observation — Don't listen to what users say, watch what users do
- Low-cost MVP validation — Produce a demo within 1 week, and observe users' willingness to pay
V. Stage 4: Calculate the Ceiling (Hands-on Practice 4)
Truman: "In business,the scariest thing is not lack of effort. The scariest thing is that after working hard for half a year, you realize the space of this business simply cannot support your ambition. "
3 Typical Scenarios
| Scenario | Question |
|---|---|
| Small-and-beautiful business | Can this revenue scale support the investment? Can it break even? |
| Project of a certain scale | After years of huge investment, how big can it get at maximum? |
| Pursuing market leadership + financing | Is being number one big enough? Can it support the business story (BP) and an excellent team? |
Truman's 1v1 data: Having discussed thirty to fifty projects,most were found to fall short of expectations after on-site decomposition and estimation
Eventually had to:Scale down / Scale up / Abandon financing.
Systematic Ceiling Estimation Method (New model added in this course)
Core 3 steps:
Step 1: Top-Down
全行业规模 × 你的细分市场比例 × 你能拿到的份额(集中度判断)
= 你的理论天花板
Data sources: iResearch / Analysys / 36Kr / Financial reports of leading listed companies.
Step 2: Bottom-Up
单店 / 单单 / 单客户 收入 × 复制次数(店数 / 用户数) × LTV
= 你的实际天花板
Data sources: Your own unit model + actual validation data.
Step 3: Two-way comparison
If Top-Down >> Bottom-Up:
- Either the market is not that big(Top-Down overestimation)
- Or you can grow bigger(Bottom-Up underestimation)
If Top-Down ≈ Bottom-Up:The estimate is credible.
Anticipate in advance whether the project is a big win or small win
A master card player can see the final hand's big win or small win early on, and will play cards much more smartly and calmly.
Predictable at the startup stage:
- Industry ceiling (scale)
- Concentration endgame (one of 4 types)
- The upper limit of the market share you can capture
- Unit model LTV / CAC
- Whether the team can hold out until that scale
VI. Re-understanding "Demand" (L6 Big Picture)
3 levels of sublimation
Level 1: User perception
Don't listen to what users say,watch what users do + Use it yourself 100 times.
Level 2: Treat demands as hypotheses
Every demand is a hypothesis to be verified:
- Is the key executor X? (To be verified)
- Is the core job Y? (To be verified)
- Is the task map 4 steps? (To be verified)
Level 3: Demand drives the whole picture
需求 (好/坏)
↓
解决方案 (产品内核)
↓
商业模式 (单元模型)
↓
增长 (引擎效率)
↓
壁垒 (终局护城河)
For a bad demand, even excellent execution in the following 4 steps cannot save it.
VII. Supporting candy: One-page template "From Task Insight to In-Depth Demand Deduction"
Feishu document template(Can be copied and used directly):
Complete L1–L4 canvas,condensed essence of the 4 hands-on sessions, suitable for personal use or team discussion, supports AI collaboration.
How to use:
- Click "Use this template" in the upper right corner of the Feishu document
- Create a blank template
- Fill in according to L1–L4 order
- Mark each cell with "hypothesis / verification status"
- Team joint review + AI-assisted deduction
VIII. Connection to other steps of the Five-Step Method
1. 需求 (本 wiki 深度)
↓
2. 解决方案 → 产品内核(在 [[yitang-five-step-method-complete]])
↓
3. 商业模式 → 单元模型(在 [[yitang-five-step-method-complete]])
↓
4. 增长 → 增长引擎
↓
5. 壁垒 → 长期护城河
See also
- Yitang / Complete Five-Step Method System (Key Hypotheses + Business Milestones + Five-Step Method 5 Advancements + Implementation) — Complete Five-Step Method System (this wiki is the in-depth content of Step 1)
- Yitang / Advanced Business Prediction (Industry Canvas + Concentration + Cases) — Industry canvas + concentration (endgame judgment)
- Yitang Opportunity Prediction Methodology — Cognitive framework for opportunity prediction
- Yitang / Complete 66-page Notes on Wang Huiwen's Tsinghua Product Course — Wang Huiwen's 4 major traits of demand + STP
- Yitang / Complete Inspiration Flash System (Cognition + Cases + Tools + Notes) — Inspiration flash (creative production during decomposition)
- Yitang / First Lesson on In-depth Review — In-depth review (post-hoc review of demand misjudgments)
- Personal memory user_role — Product manager perspective
Relationship with existing Skills
This wiki is an in-depth expansion of the demand analysis submodule of yitang-master-business-model (Skill)in-depth expansion. The Skill provides a 4-week operational framework; this wiki presents the complete methodology of 4 lessons + Truman's practical experience from 30+ projects.