# Search Playbook

## Contents

1. Search goals
2. Query patterns
3. Source mapping
4. Verification
5. Sampling
6. Capture habits

## 1. Search Goals

Do not search randomly. Search to answer a concrete research need:

- identify who the case is
- find first-party expressions
- map platforms and business entry points
- build a timeline
- collect representative content
- find visible monetization clues

## 2. Query Patterns

Do not search only the person's name. Combine name, role, platform, and business clues.

### 2.1 Base Queries

Use combinations such as:

- `name or brand + interview`
- `name or brand + podcast`
- `name or brand + Bilibili / Xiaohongshu / Weibo / YouTube / newsletter`
- `name or brand + founder / studio / company / workshop`
- `name or brand + course / membership / community / training camp`
- `name or brand + event / signup / ticket / workshop`
- `name or brand + collaboration / sponsor / advisor`

### 2.2 Deep-Dive Queries

Use combinations such as:

- `site:platform-domain name or brand`
- `name or brand + year`
- `name or brand + niche keyword`
- `name or brand + launch / update / review / recap`
- `name or brand + official site`

### 2.3 Business Model Queries

Use combinations such as:

- `name or brand + pricing`
- `name or brand + member`
- `name or brand + consulting`
- `name or brand + workshop`
- `name or brand + sponsor`
- `name or brand + service`
- `name or brand + application form`

### 2.4 Timeline Queries

Use combinations such as:

- `name or brand + joined`
- `name or brand + founded`
- `name or brand + launched`
- `name or brand + became`
- `name or brand + keynote`
- `name or brand + guest`

## 3. Source Mapping

Build the source map before deep reading.

### 3.1 A-Level Sources

Prioritize:

- public posts by the person or brand
- official pages
- original program pages
- original event pages

### 3.2 B-Level Sources

Use next:

- interviews by reputable media
- platform profile pages
- databases
- partner official pages

### 3.3 C-Level Sources

Use only for leads:

- reposts
- summaries
- forums
- comments

Rule:

- never let a C-level source be the only basis for an important conclusion

## 4. Verification

Cross-check the claims that are most likely to distort the case if wrong:

- education background
- job history
- project ownership
- launch date
- visible scale metrics
- revenue-related claims

Preferred verification pattern:

1. find the first-party expression
2. find a second source of a different type
3. record the date or capture time
4. note conflicts instead of smoothing them away

## 5. Sampling

Do not sample only recent viral content.

Minimum sampling dimensions:

- recent samples, usually the last 30 days
- stable-period samples, usually the last 6 months
- early or transition-period samples
- cross-platform samples on the same topic

Suggested sample sizes:

- podcasts: 5-10
- videos: 10-20
- articles or posts: 5-10
- event or signup pages: collect as completely as possible

Recommended fields:

- link
- publish date
- platform
- content form
- topic tag
- key point
- style feature
- conversion action visible or not

## 6. Capture Habits

While researching:

- save exact links, not only platform names
- save dates whenever possible
- note what is observed versus what is inferred
- keep short evidence notes instead of copying long passages
- record why a source matters