Truman: "If you only learn one capability that is 100% actionable,that can earn back 10 years of tuition fees,it is the systematic research behind today's course.. "
Truman: "*The Lean Startup* is one of only five books Wang Huiwen recommended at Tsinghua University.People who can apply lean methods will hold a huge advantage in competition."
Part A: Intelligence Research (5 lessons + 13 IPO prospectuses + Zhang Lei's nursing home case)
1. Why learn this
Core data from the first research lesson:
- In 2018, Truman launched the "Business Research" project → 2022 course series → 2026 third round of AI-powered upgrade(current round)
- Tens of thousands of assignments + business research cases across various industries
- The only systematic training course on "Competitive Intelligence" on the market
95% of operators lack research capabilities:
- Lack of awareness
- Lack of methodology
- Lack of research skills
Capabilities of students who do research well (senior + paid heavy tuition):
- Search for reports / stake out stores / use search engines / estimate revenue / visit under assumed identities / find job seekers / use AI for monitoring
Real cases after completing the course
- One student used research methods to directly falsify and abandon their startup project
- One student found after the course that their marketing director did not understand the market, and fired them directly
- One student paused heavily invested R&D, comprehensively researched competitors' technology roadmaps, and got the answer
2. OSCAR Research Model (taught by Zhang Lei)
Zhang Lei (Partner at Yitang / Head of MBA): "The value of systematic research — it helps you obtain high-value information and make high-quality decisions. "
Zhang Lei reviews 5-10 projects every day, mentors a 2-billion-yuan payment company to explore unit economics models, and mentors a daily chemical overseas expansion company that raised 25 million yuan.
OSCAR Model (5 steps)
| O | S | C | A | R |
|---|---|---|---|---|
| Objective(Objective) | Scope(Scope) | Channel(Channel) | Analyze(Analyze) | Report(Report) |
O - Objective: Research objective
Key question: Why conduct this research? What are the core 1-3 assumptions to be verified?
Avoid: Research for the sake of research.
S - Scope: Research scope
Define:
- Industry / sub-sector / time range
- Company level (top / mid-tier / long-tail)
- Geography (nationwide / first-tier / overseas)
Avoid: Scope too large, making the report look good but unusable; scope too small, missing key information.
C - Channel: Research channels
5 major channels (detailed in the exploratory arsenal below):
- Public information(reports / financial reports / news / Zhihu)
- Expert interviews(peers / upstream suppliers / downstream customers)
- User interviews(target users / churned users / power users)
- On-site research(staking out stores / counting foot traffic / assumed identities)
- AI monitoring(automation / large model collaboration)
A - Analyze: Analysis
Core point: Do not pile up "fancy" templates such as SWOT / PEST / Porter's Five Forces,answer whether your key assumptions hold true.
R - Report: Report
Key point: It must be able to guide business decisions for the next 6-12 months.
If it cannot guide decisions, this report is a waste.
3. Exploratory Research Arsenal (taught by Truman)
Contrast between business novices and veterans
| Novice | Veteran |
|---|---|
| Likes to analyze | Likes to gather intelligence |
| Discusses SWOT / strategy / flywheels | Stakes out stores / counts foot traffic / finds job seekers / AI monitoring |
| Stops at a superficial level | Gets core intelligence in one or two phone calls |
Truman's observation: When novices discuss option selection / pain points / pricing,veterans have already obtained reliable answers verified by peers through search + phone calls.
Research arsenal (new in this lesson)
5 major research tools:
| Tool | Applicable scenarios | Tools |
|---|---|---|
| Public information | Full industry landscape | 36Kr / iResearch / Analysys / Zhihu / prospectuses |
| Expert interviews | Industry know-how | LinkedIn / job platforms / school alumni |
| User interviews | Real user needs | Extreme users (love most + hate most) / churned users |
| On-site research | Real frontline situation | Staking out stores / counting foot traffic / pretending to be a job seeker / pretending to be a customer |
| AI monitoring | Continuous tracking | Large models + crawlers + dashboards |
Anything else? Anything else?
Truman's questioning technique: After every research session, ask "Anything else? Anything else?""
Veterans' research habits:
- First wave: 5 conventional channels
- Second wave: "Anything else?" list 5 new angles
- Third wave: "Anything else?" list 5 unconventional methods
- Until there is truly "nothing more to say"
4. Expert Interview SOP (2021 → 2024 upgraded version)
Hong Lin of Dawang Capital: "This is no longer the pure internet era; information no longer fully circulates in 2-3 months.Many offline/vertical industries now have high barriers; if you cannot quickly build professional cognition, it will be very hard to get opportunities."
3 professional cognition criteria investors use to evaluate founders
- Industry cognition — Deep understanding of the industry chain / abstracted strengths and weaknesses of competitors / new observations on trends
- Business cognition — Why it can succeed / small and beautiful vs scalable / changes after scaling
- Organization cognition — Whether values are useful / organizational capabilities at each stage / how to improve
4-step expert interview method
Step 1: Find the right people
- Direct competitor CEO (very hard)
- Direct competitor executive (moderate)
- Upstream suppliers / downstream customers (easy)
- Departed core employees / candidates on job platforms (easy)
- Industry veterans / investors / media
Step 2: Build trust
- Self-introduction + let the other party share "without burden"
- Break the ice with peer cases (demonstrate professionalism)
- Provide value to the other party (information exchange / introduce connections)
Step 3: High-quality questioning
- Do not ask "What do you think of XX industry" (too broad)
- Ask "What was the hardest decision you made recently" (specific)
- Ask "What would you do differently if you started over" (counterfactual)
- Ask "What will the industry look like in 5 years" (future)
- Ask "Who else should I talk to" (snowball)
Step 4: Distill + verify
- Ask the same question to 3-5 experts
- Find consistent points (reliable patterns)
- Find inconsistent points (deep disagreements → find the reasons)
5. User Interview SOP
Since starting his business, how many target users has Truman talked to 1v1 face-to-face?
Answer:At least 100-200(different numbers in each period).
User interviews vs user questionnaires
| Dimension | Interview | Questionnaire |
|---|---|---|
| Depth | Deep | Shallow |
| Volume | 5-50 people | 100-1000 people |
| Timing | Early-stage discovery | Mid-stage validation |
| Main value | Identify new opportunities | Quantify existing assumptions |
5-sentence interview technique
- Ask less "What do you want"(Users will say they want faster horses — famous quote from Ford)
- Ask more "What was the scenario the last time you did XX"(Behavioral anchoring)
- Ask more "Why don't you use XX alternative"(Competitor comparison)
- Ask more "How much would you pay for this solution"(Willingness to pay)
- Ask more "Who else / what group of people should I talk to"(Snowball)
Early-stage discovery vs mid-stage validation
| Stage | Method |
|---|---|
| Early-stage discovery | Extreme user interviews (love most + hate most) — identify non-obvious needs |
| Mid-stage validation | Multi-user questionnaires + A/B testing — quantify key conversions |
| Late-stage optimization | Data-driven — look at real behavior instead of listening to what users say |
6. Bonus 1: Zhang Lei's nursing home research case
Student: Zhang Lei (later joined Yitang, became Partner / Head of MBA)
Early assignment:Systematic research data of 5 nursing homes(Excel)
Research dimensions(typical complete structure):
- Pricing / room types / occupancy rate / service list / staff ratio / location / surrounding competitors
- Estimate revenue / profit for each facility
- Deduce the market ceiling of the entire regional market
Result: This assignment became Zhang Lei's "foot in the door" to join Yitang (later Zhou Zijing gave a Strong Hire rating, and Zhang Lei joined Yitang).
7. Bonus 2: Collection of excellent research hacker assignments
(See the original document for the specific list, covering high-quality research report examples across multiple industries)
8. Bonus 3: 13 real Hong Kong stock IPO prospectuses (understand listed company reports)
Truman: "A prospectus is one of the most information-dense business documents in the world, because of legal liability, all data must be accurate. Understanding a prospectus is equivalent to getting top companies to do industry research for you for free."
List of 13 real prospectuses
| # | File name | Company | Stock code | Date |
|---|---|---|---|---|
| 1 | ltn20180625034_c.pdf |
Xiaomi Group | 1810 | 2018-06 IPO |
| 2 | ltn20180907018_c.pdf |
Meituan Dianping(HKEX IPO announcement) | 3690 | 2018-09 |
| 3 | 2019122900028_c.pdf |
Jiumaojiu International Holdings(parent company of Tai Er Chinese Sauerkraut Fish) | 9922 | 2019-12 IPO |
| 4 | 2020060200010_c.pdf |
NetEase(Hong Kong secondary listing) | 9999 | 2020-06 |
| 5 | 2020102900114_c.pdf |
New Oriental(Hong Kong secondary listing) | 9901 | 2020-10 |
| 6 | 2020120100100_c.pdf |
Pop Mart | 9992 | 2020-12 IPO |
| 7 | 2021012600014_c.pdf |
Kuaishou Technology | 1024 | 2021-01 IPO |
| 8 | 2021030400014_c.pdf |
Autohome(Autohome secondary listing) | 2518 | 2021-03 |
| 9 | 2021031200022_c.pdf |
Baidu(Hong Kong secondary listing) | 9888 | 2021-03 |
| 10 | 2021031800174_c.pdf |
Bilibili(Hong Kong secondary listing) | 9626 | 2021-03 |
| 11 | 2021040800020_c.pdf |
Trip.com Group | 9961 | Secondary listing in April 2021 |
| 12 | 2021061800010_c.pdf |
Nayuki | 2150 | 2021-06 IPO |
| 13 | 2021062500040_c.pdf |
XPeng Motors | 9868 | Secondary listing in June 2021 |
Total information volume:13 documents × average 500 pages ≈ 6500 pages of full business landscape of top companies.
Core information in prospectuses (included in every document)
- Industry overview(full pages of reports from industry consulting firms included)
- Company business(core products / users / business model / unit economics)
- Historical financials(3-5 years of income statement / balance sheet / cash flow)
- Management / shareholder structure
- Risk factors(legal / compliance / competition / operations)
- Use of proceeds
3 major uses for reading prospectuses
- Industry research — Extract the full industry overview pages from prospectuses of the corresponding industry and use them as industry reports
- Peer BP comparison — See how top companies tell their stories → imitate their narrative structure
- Unit model benchmarks — Look at top companies' average order value / repurchase rate / gross margin / net margin → use as the base for your own unit model
Recommended reading order (corresponding to business type)
| Your business | Recommended to read first |
|---|---|
| Food & beverage / chains | Jiumaojiu (Tai Er) / Nayuki |
| Consumer goods / designer toys | Pop Mart / Xiaomi |
| Short video / content | Kuaishou / Bilibili |
| Education | New Oriental |
| E-commerce / services | Trip.com / Meituan / Autohome |
| Mobility / hardware | XPeng |
| Internet platforms | Baidu / NetEase |
Part B: Lean Experimentation (4 lessons)
1. Why learn this
Special status of this topic:
- This was the first formal business course Truman taught at the end of 2018
- First version delivered by Truman (practical experience)
- Second version taught by Cheng Hao(founder of Xunlei, author of *Lean Business Thinking*), with Truman assisting
- 2024 third version taught by Truman, advised by Cheng Hao, co-created by a thousand people
"This is truly where Yitang's dream began."
Relaunched after 5 years of being banned, the most requested topic for upgrade, and also the core capability Truman used to build Yitang from 0 to 1.
The six "worst" wastes
| # | Type | Description |
|---|---|---|
| 1 | Most tragic: Wrong demand | Working hard to solve an unimportant / non-existent demand |
| 2 | Most testing: Wrong solution | Demand is right but solution is wrong, product core does not hold |
| 3 | Most blind spot: Premature refinement | Making the product polished and detailed early on, but PMF is not validated |
| 4 | Most costly: Premature scaling | Expanding to 15 cities before the unit model is proven |
| 5 | Most regrettable: Building behind closed doors | Not engaging with users, building products based on intuition |
| 6 | Hardest to adjust: Too slow to pivot | Data is bad but unwilling to pivot, continuing to lose money |
Truman's 5-character summary:"Love to hold back big moves" — common bad habit of business leaders.
2. 4 key principles of lean startup
| Principle | Key point |
|---|---|
| 1. Release early | Release early, engage with real users |
| 2. Iterate quickly in small steps | Only validate 1-2 key assumptions each time |
| 3. Restrained design | Do not build a "great product", only build what is "sufficient for validation" |
| 4. Embrace change | Pivot decisively when data is bad, no sentimentality |
3. Lean practice 1: Strip assumptions
Key assumption prioritization
Every business has 10-30 assumptions,rank them by priority(high → low):
| Priority | Assumption type | Example |
|---|---|---|
| P0 | Does the demand hold? | Do these people really have this pain point? |
| P0 | Does the unit model hold? | Does a single order make money? |
| P1 | Are channels effective? | What is the customer acquisition cost on Douyin? |
| P1 | Is retention sufficient? | Is 30-day retention > 30%? |
| P2 | Can the team scale? | Can we open in 50 cities? |
Principles:Validate P0 first, then P1, and finally P2. Do not work on P2 before the P0 hypothesis has been validated.
IV. Lean Practice 2: MVP Design
MVP (Minimum Viable Product) = Minimum Viable Product.
4 Major Types of MVP
| Type | Description | Example |
|---|---|---|
| Smoke test | No product at all; check whether anyone will buy | Dropbox used a demo video |
| Wizard of Oz | Real front end, manual back end | Early Zappos took orders manually |
| Concept test | Landing page + traffic-driving ads → observe conversion | Test willingness to pay |
| Limited release | Real product, but limited to 100 people / 1 city | Internal beta |
MVP Design Principles
- Look at real payments, not stated intent
- Test core hypotheses, not additional features
- Data must be quantifiable(conversion rate / repurchase / NPS)
- Set clear Go / No-Go thresholds
V. Lean Practice 3: Systematic Testing
Systematic testing = a series of consecutive MVPs
Cadence: one complete MVP cycle every 2–4 weeks.
设计假设 → MVP 测试 → 收集数据 → 分析 → 调整 → 下一轮
Record for each round
- Hypothesis validated in this round
- MVP format for this round
- Expected results for this round (Go / No-Go threshold)
- Actual results
- Lessons learned (experience + takeaways)
- New hypotheses for the next round
Key disciplines
- Each round ≤ 4 weeks(otherwise it is not lean)
- Cost per round ≤ 50,000 RMB(otherwise it is not low-cost)
- Test only 1–3 hypotheses per round(otherwise you will not learn)
- Each round must end with a Go / No-Go decision(otherwise it is not validation)
VI. Relationship with the Five-Step Method
[[yitang-demand-analysis-complete]] 需求拆推评算
↓
[[yitang-product-core-unit-model]] 产品内核 + 单元模型
↓
情报调研(本 wiki)→ 拿到真实数据
↓
精益实验(本 wiki)→ 低成本验证关键假设
↓
单元模型 L5 找准基准值
↓
单元模型 L6 规模经济判断
↓
[[yitang-business-prediction-deep]] 集中度终局
↓
[[yitang-机会预判]] 完整 4 步预判
See also
- Yitang / Complete Five-Step Method System (Key Hypotheses + Business Milestones + 5 Progressions of the Five-Step Method + Implementation) — Complete Five-Step Method System
- Yitang / Complete Demand Analysis System (4 Stages: Decompose, Deduce, Evaluate, Calculate + JTBD) — Demand Analysis (user interviews are its implementation)
- Yitang / Complete Product Core + Unit Model System (Steps 2+3 of the Five-Step Method) — Product Core + Unit Model
- Yitang / Advanced Business Pre-Judgment (Industry Canvas + Concentration + Cases) — Industry Canvas + Concentration
- Yitang Opportunity Pre-Judgment Methodology — Cognitive Framework for Opportunity Pre-Judgment
- Yitang / Complete 66-Page Notes from Wang Huiwen's Tsinghua Product Course — Wang Huiwen's Tsinghua Course:People who can apply lean methods have a huge competitive advantage
- Yitang / First Lesson on In-Depth Review — In-Depth Review
- Yitang / Complete Inspiration Flash System (Cognition + Cases + Tools + Notes) — Inspiration Flash
- Personal memory user_role — Product Manager + Entrepreneur
Relationship with Existing Skills
This wiki is an in-depth expansion of the research + lean experimentation submodule of yitang-master-business-model (Skill). The Skill provides the workflow; this wiki covers:
- The complete 5 steps of the OSCAR research model
- Reference list of 13 Hong Kong stock IPO prospectuses
- Lean Startup "Six Greatest Wastes + 4 Principles" + 4 MVP Types + Systematic Testing Disciplines