gr-user-interview

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

⚠️ 2C 产品的渠道调整

⚠️ Channel Adjustment for 2C Products

本 skill 默认 dev / B2B 渠道。2C 消费品 / 教育 / 应用：获客主战场是 垂类社区 + 短视频 + 垂直 KOL，按地区公开数据选第一平台（如印尼/泰国短视频已反超 Facebook）。KOL 优先 nano / micro 垂类——粉丝越多互动率越低，micro > mega 性价比更高。完整 2C 渠道数据库 + 公开来源见 →

gingiris-seo-geo/references/2c-adaptation.md

。

This skill defaults to dev / B2B channels. 2C consumer goods / education / apps: The main customer acquisition channels are vertical communities + short videos + vertical KOLs. Select the top platform based on regional public data (e.g., short videos have surpassed Facebook in Indonesia/Thailand). Prioritize nano / micro vertical KOLs — the more followers, the lower the engagement rate, and micro KOLs offer higher cost-effectiveness than mega ones. For the complete 2C channel database + public sources →

gingiris-seo-geo/references/2c-adaptation.md

gr-user-interview — 用户访谈

gr-user-interview — User Interview

什么时候用

When to Use

"我要找 PMF，怎么做访谈"
"用户反馈很杂，不知道抓哪个"
"我采访了 5 个人，都没 insight"
"怎么避免用户说假话"
"不知道该访谈谁 / 约不到人"
"激活率、付费率该怎么定义和对标"

"I want to find PMF, how do I conduct interviews?"
"User feedback is messy, I don't know which to focus on"
"I interviewed 5 people but got no insights"
"How to avoid users lying?"
"I don't know who to interview / can't get people to agree"
"How to define and benchmark activation rate and payment rate?"

核心原则（5 条）

Core Principles (5 Rules)

问 behavior，不问 opinion ❌ "你会用这个功能吗？" → 会得到礼貌性"yes" ✅ "上次遇到 X 问题是什么时候？你当时做了什么？" → 得到真实行为
20 分钟访谈 >> 5 个 feature 问题 深度一个场景，比问 5 个浅问题更有价值
录音 / 整稿 > 当场记笔记 当场记会错过微表情和语气
5 人后暂停，复盘 不要连访 20 人才总结，5 人后大概率已看到模式
用词镜像 用户用什么词，你也用什么词 —— 不要替换成产品语言

Ask about behavior, not opinion ❌ "Would you use this feature?" → You'll get a polite "yes" ✅ "When was the last time you encountered problem X? What did you do then?" → Get real behaviors
20-minute interview >> 5 feature questions Deep diving into one scenario is more valuable than asking 5 shallow questions
Recording / full transcript > taking notes on the spot Taking notes on the spot will cause you to miss microexpressions and tone
Pause and review after 5 interviews Don't wait until you've interviewed 20 people to summarize; patterns will likely emerge after 5 interviews
Mirror User's Vocabulary Use the same words the user uses — don't replace them with product jargon

访谈结构模板（20-30 分钟）

Interview Structure Template (20-30 minutes)

阶段	时长	问什么
Warm-up	2 min	"你今天做了什么？"（开放式，松弛下来）
Discover	5 min	"上次做 X 是什么时候？能讲讲吗？"
Deep-dive	10 min	"当时你用了什么工具？怎么用的？哪里卡住？"
Alternative	5 min	"有没有试过其他方法？为什么没继续？"
Wrap-up	3 min	"如果有一个魔法能解决这个问题，它会怎么工作？"

Stage	Duration	Questions
Warm-up	2 min	"What did you do today?" (Open-ended question to help the user relax)
Discover	5 min	"When was the last time you did X? Can you tell me about it?"
Deep-dive	10 min	"What tools did you use then? How did you use them? Where did you get stuck?"
Alternative	5 min	"Have you tried other methods? Why didn't you continue using them?"
Wrap-up	3 min	"If there was a magic solution to this problem, how would it work?"

Token Top10 访谈法（招牌打法）

Token Top10 Interview Method (Signature Approach)

后台用量 Top10 拉 list → 共享屏幕访谈（必录屏）→ transcript 喂 AI 提 feature/bug 进 Linear → 次日回访。每天 ≥2 场。

按 token 消耗 / 用量排 Top10，每周刷新——重度用户的反馈密度最高
必须共享屏幕 + 录屏：看真实操作 > 听口述
Transcript 交给 AI 自动提取 feature request / bug，直接进 issue 工具
次日回访"你提的 X 已修/已排期"——反馈闭环就是留存杠杆
访谈后截 5-10 分钟关键片段给产研

邀约节奏：

1/3/7/14 天四次触达（Day 1 → 3 → 7 → 14 各跟进一次）
双通道：邮件模板 + 邮箱反查 LinkedIn 私信，邮件没回转 LinkedIn 补一轮
话术用宽泛理由（"早期团队想做访谈"），不暴露真实筛选原因——避免用户表演

组合技：

KOL 访谈二合一：合作 YouTuber 顺便做用户访谈 + 可用性测试，一次合作两份价值（20-30 分钟视频访谈 ROI 最高）
Wisprflow 打法：launch 前人工 onboard 前 500 用户逐个访谈，用势能带 launch

Pull the list of Top10 users by backend usage → Conduct screen-sharing interviews (must record screen) → Feed transcript to AI to extract feature/bug requests into Linear → Follow up the next day. Conduct ≥2 interviews per day.

Rank users by token consumption / usage Top10, refresh weekly — power users provide the highest density of feedback
Must do screen-sharing + screen recording: Watching real operations is better than listening to verbal descriptions
Feed transcripts to AI to automatically extract feature requests / bugs, directly into issue tracking tools
Follow up the next day with "The X you mentioned has been fixed/scheduled" — closing the feedback loop is a retention lever
After the interview, clip 5-10 minutes of key segments to share with product and R&D teams

Invitation Rhythm:

Four touchpoints on Day 1/3/7/14 (Follow up once on Day 1 → 3 → 7 → 14)
Dual-channel: Email template + LinkedIn message via email reverse lookup; if no reply to email, follow up on LinkedIn
Use a broad reason in the script ("The early-stage team wants to conduct interviews") and don't reveal the real screening criteria — avoid user performance

Combination Techniques:

Dual-purpose KOL interviews: Collaborate with YouTubers to conduct user interviews + usability tests at the same time, getting two values from one collaboration (20-30 minute video interviews offer the highest ROI)
Wisprflow Approach: Manually onboard and interview the first 500 users one by one before launch, using momentum to drive launch

激活基准速查

Quick Check for Activation Benchmarks

项	基准
激活定义	必须是行为阈值：完成 ≥2 任务 / 连续 3 天 ≥20 轮对话 / 使用超 5 分钟——不是"登录过"
注册 → 付费合格线	>5%（Pro-C 5-8%、SMB 10-15%）
流失定义	两次对话后 24h 不回访 = 流失 → 自动触发挽回邮件
付费归因	四时间戳交叉：注册 × 激活 × 支付 × 最近登录

Item	Benchmark
Activation Definition	Must be a behavior threshold: Complete ≥2 tasks / ≥20 consecutive daily dialogues for 3 days / Usage over 5 minutes — not "logged in"
Qualified rate from registration to payment	>5% (Pro-C 5-8%, SMB 10-15%)
Churn Definition	No follow-up within 24h after two dialogues = churn → Automatically trigger win-back emails
Payment Attribution	Four timestamp cross-check: Registration × Activation × Payment × Last login

深度参考

In-depth References

📂 https://github.com/Gingiris-1031/gingiris-skills/tree/main/skills/gingiris-user-interview

```
references/templates.md
```
— 不同阶段（Discovery / Validation / Retention）的问题模板
HeyGen 937 访谈案例拆解

📂 https://github.com/Gingiris-1031/gingiris-skills/tree/main/skills/gingiris-user-interview

```
references/templates.md
```
— Question templates for different stages (Discovery / Validation / Retention)
Case analysis of HeyGen's 937 interviews

合成流程

Synthesis Process

每次访谈后 24h 内整理 3 张卡片：
- Pain（用户的真实痛）
- Workaround（他们目前怎么凑合）
- Quote（可引用的原话）
每 5 个访谈做一次复盘：
- 重复出现 ≥ 3 次的 Pain → 候选 PMF 方向
- 重复出现 ≥ 2 次的 Workaround → 竞品信号
每 20 个访谈做一次决策：
- 继续这个假设 / 换方向 / 需要更多数据

Within 24h after each interview, organize 3 cards:
- Pain (User's real pain points)
- Workaround (How they currently cope)
- Quote (Quotable original statements)
Conduct a review after every 5 interviews:
- Pain points mentioned ≥3 times → Candidate PMF direction
- Workarounds mentioned ≥2 times → Competitor signals
Make a decision after every 20 interviews:
- Continue with this hypothesis / Change direction / Need more data

级联推荐

Cascading Recommendations

访谈发现新 PMF 方向 →
```
gr-blog-post
```
写一篇（验证市场反应）
访谈发现对手名字被提 ≥ 3 次 →
```
gr-competitor
```
深度分析
PMF 明确 →
```
gr-b2b-growth
```
或
```
gr-oss-marketing
```
启动增长
需要招募访谈对象 → 从
```
gr-seo-patrol
```
的 GSC 流量挑读者，或直接上「Token Top10 访谈法」拉后台用量榜
在跑红人合作 → 用「KOL 访谈二合一」顺路拿访谈（红人打法见
```
gingiris-kol-outreach
```
）
访谈后要定激活/付费目标 → 对照上方「激活基准速查」先把行为阈值写死

If a new PMF direction is found through interviews → Write an article with
```
gr-blog-post
```
(Validate market response)
If a competitor's name is mentioned ≥3 times in interviews → Conduct in-depth analysis with
```
gr-competitor
```
Once PMF is confirmed → Launch growth with
```
gr-b2b-growth
```
or
```
gr-oss-marketing
```
If you need to recruit interview subjects → Select readers from
```
gr-seo-patrol
```
's GSC traffic, or directly use the "Token Top10 Interview Method" to pull the backend usage list
If running influencer collaborations → Use the "Dual-purpose KOL interviews" to get interviews along the way (For influencer strategies, see
```
gingiris-kol-outreach
```
)
If you need to set activation/payment targets after interviews → First lock in the behavior threshold by referring to the "Quick Check for Activation Benchmarks" above

反模式

Anti-patterns

❌ 问 "你愿意付费吗" → 100% 是假答案
❌ 发问卷替代访谈（信息密度差 10x）
❌ 只访现有用户不访流失用户（流失才是真金）
❌ 一次访谈超过 45 分钟（疲劳 → 敷衍）
❌ 访谈中推销自己产品（污染答案）
❌ 邀约只发一次就放弃（1/3/7/14 四次触达才是基线）
❌ 邀约话术暴露筛选原因（"因为你用量高所以找你" → 用户开始表演）
❌ 激活定义写成"注册/登录过"（必须是行为阈值）
❌ 访谈完不回访（次日回访"已修/已排期"是最便宜的留存动作）

❌ Asking "Would you pay for this?" → 100% of answers are fake
❌ Using questionnaires instead of interviews (Information density is 10x lower)
❌ Only interviewing existing users, not churned users (Churn users hold real value)
❌ Interviews longer than 45 minutes (Fatigue leads to perfunctory answers)
❌ Promoting your product during interviews (Pollutes answers)
❌ Giving up after sending only one invitation (Four touchpoints on Day 1/3/7/14 is the baseline)
❌ Revealing screening criteria in invitation scripts ("We're reaching out because you're a heavy user" → Users start performing)
❌ Defining activation as "registered/logged in" (Must be a behavior threshold)
❌ Not following up after interviews (Following up the next day with "Fixed/scheduled" is the cheapest retention action)