Update gingiris-user-interview to 1.1.0 with evidence-led workflows

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2026 Gingiris
+Copyright (c) 2026 Iris / Gingiris
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -5,7 +5,7 @@ language:
 - zh
 - ja
 - ko
-pretty_name: "Gingiris User Interview Playbook 2026 — HeyGen 937-Interview PMF Discovery Framework, JTBD Scripts, User Tier Grading"
+pretty_name: gingiris-user-interview — Evidence-led workflow
 task_categories:
 - text-generation
 - text-classification
@@ -41,493 +41,30 @@ tags:
 size_categories:
 - n<1K
 ---
+# gingiris-user-interview
 
-<div align="center">
+Version 1.1.0 · Updated 2026-11-08.
 
-<img src="assets/logo.png" alt="Gingiris Logo" width="120" />
+🇺🇸 User Interview & Cold-Start Operations Playbook — Evidence-led research, usability testing and follow-up. Complete SOP for user screening, interview execution, Beta testing methodology, cold-start user operations, churn analysis, and user grading systems.
 
-# User Interview Playbook 2026
+🇨🇳 用户访谈与冷启动运营实战手册 — 基于真实行为、可用性观察与回访验证需求。包含用户筛选、访谈执行、Beta 测试设计、冷启动用户运营、流失分析、用户分级体系完整 SOP。
 
-### Find PMF through founder-led conversations — the exact framework behind HeyGen's 937 interviews, JTBD scripts, screen-share protocols, and S/A/B/C user tier grading
+🇯🇵 ユーザーインタビュー＆コールドスタート運営プレイブック — 実際の行動とユーザビリティテストに基づく需要検証。ユーザースクリーニング、インタビュー実行、Betaテスト設計、コールドスタートユーザー運営、離脱分析、ユーザーグレーディング。
 
-[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
-[![Website](https://img.shields.io/badge/Website-gingiris.tools-blue)](https://gingiris.tools)
-[![Skills.sh](https://img.shields.io/badge/Install-skills.sh-purple)](https://skills.sh/Gingiris-1031/gingiris-user-interview)
-[![HuggingFace](https://img.shields.io/badge/🤗-Gingiris-yellow)](https://huggingface.co/Gingiris)
+🇰🇷 사용자 인터뷰 & 콜드 스타트 운영 플레이북 — 실제 행동과 사용성 관찰에 기반한 수요 검증. 사용자 스크리닝, 인터뷰 실행, Beta 테스트 설계, 콜드 스타트 사용자 운영, 이탈 분석, 사용자 등급 시스템.
 
-**[English](#english) | [中文](#中文版) | [日本語](#日本語) | [한국어](#한국어)**
+Triggers: "user interview" | "user research" | "customer discovery" | "PMF" | "product-market fit" | "beta test" | "beta testing" | "cold start users" | "cold start operations" | "churn analysis" | "user grading" | "用户访谈" | "用户调研" | "冷启动" | "冷启动运营" | "Beta测试" | "流失分析" | "用户分级"
 
-</div>
-
-## 📦 Install
+## Install / 安装
 
 ```bash
-npx skills add Gingiris-1031/gingiris-user-interview
-```
-
-Then ask your AI agent:
-
-> "Help me design PMF validation interviews" · "Draft questions for our 10 power users" · "Build a churn diagnostic interview script" · "Score users into S/A/B/C tiers based on these usage signals"
-
-🔗 [View on gingiris.tools/skills/](https://gingiris.tools/skills/) · [Author blog](https://gingiris.tools) · [skills.sh listing](https://skills.sh/Gingiris-1031/gingiris-user-interview)
-
----
-
-## 📊 Citable Statistics
-
-| Stat | Value | Source / Context |
-|:---|:---|:---|
-| **HeyGen interviews to find PMF** | **937 interviews / 6 months** | The signature number that defined founder-led research |
-| **Wisperflow / Higgsfield interview cadence** | **500+ interviews each** | Independent PMF discovery cases (similar founders) |
-| **Registration → Paid conversion target** | **5–8%** | Benchmark for B2B SaaS post-PMF |
-| **Minimum interview cadence (post-PMF)** | **3–5 per week** | Maintain product-customer signal loop |
-| **Actionable insights per interview** | **3+ feature requests** | Quality bar — below this means scripting was weak |
-| **User tier grading framework** | **4 tiers** (S / A / B / C) | Across 5 dimensions: frequency, depth, payment, engagement, social influence |
-| **User priority for interview targeting** | **5 segments × 3 priority levels** | Paying / Power / Competitor / Churned / Registered-but-stuck |
-| **"Very disappointed if removed" threshold** | **≥ 40%** = strong PMF | Sean Ellis test threshold, validated across cases |
-| **Interview duration sweet spot** | **30–45 minutes** | Long enough for deep dive, short enough to maintain quality |
-| **AFFiNE GitHub stars (creator authority)** | **60,000+** | Iris led 0→60K organic stars |
-| **Iris's startup advisory portfolio** | **150+** | Founder-led interview consulting |
-
-**The thesis**: Product-Market Fit isn't found in dashboards — it's earned through founder-led conversations. The pattern from HeyGen (937), Wisperflow (500+), and Higgsfield (500+) is identical: **constant interviewing + iteration in the first 6 months**. This playbook documents the exact framework: screening rubrics, JTBD scripts, screen-share observation protocols, post-interview debrief templates, and user tier grading.
-
----
-
-## 🚀 30-Second Preview
-
-```
-You: We have 50 B2B SaaS customers. Want to validate PMF. Draft an interview plan.
-Agent (with gingiris-user-interview):
-  Phase 1 — Segment (Week 1):
-    • Power users (used 4+ weeks consecutively): 10 candidates
-    • Churned (canceled in last 30 days): 8 candidates
-    • Activated-but-stuck (low engagement after signup): 12 candidates
-
-  Phase 2 — Interview script (JTBD format):
-    1. "What were you doing the day you decided you needed [product]?"
-    2. "What did you try before you found us?"
-    3. "What almost stopped you from signing up?"
-    4. [...8 more, each mapped to PMF signal]
-
-  Phase 3 — Scoring:
-    • Strong PMF: 40%+ "very disappointed if removed" responses
-    • Weak PMF: <40% — pivot signal
-```
-
-Real framework from $50M+ ARR exits — not vague "talk to users".
-
----
-
-> **Why Use This Playbook**
-> * **Battle-Tested**: HeyGen did 937 interviews in 6 months to find PMF — this is the same methodology
-> * **Complete Framework**: End-to-end templates from user screening to post-interview analysis
-> * **Ready to Use**: Question banks, debrief templates, user grading standards
-
----
-
-## 🆕 What's New (July 2026): Field Notes from Founder Interviews
-
-Fresh observations distilled from our founder podcast series. They sharpen — not replace — the protocols above.
-
-### The One-Line Test for Interview Quality
-
-A useless interview looks like an investor pitch: you talk 80% of the time and the user quietly zones out. A good interview looks like a walkthrough: the user shares their screen and drives, showing you their real use cases step by step.
-
-The uncomfortable constant behind this test — the user journey you imagined and the journey users actually take are almost never the same. Screen sharing is how you find the gap; talking at users is how you hide it.
-
-### "Willing to Pay" Is Not Paying
-
-Saying "I'd pay for this" in an interview and actually entering a card number run on two different logics. Interview answers are social; payments are economic. Treat stated willingness as a hypothesis, then validate it with real payment behavior — a paywall, a pre-order, a paid pilot — before you count it as demand.
-
-### Benchmark Anchors (Reported)
-
-- Gamma, HeyGen, and Lovable reportedly ran on the order of 500–1,000 user interviews within roughly their first 6 months — before scaling spend.
-- The counter-example burn path: skipping that phase and buying growth instead, at tens of dollars per signup and hundreds of dollars per paying user, with no compounding insight to show for it.
-
-The order matters: interviews first, scale second.
-
-### The Feedback Loop That Drives Retention
-
-One pattern kept recurring among teams with strong retention:
-
-1. A deep user writes a 2,000-word review with screenshots.
-2. The whole team breaks it down together, issue by issue.
-3. Every issue gets fixed within 1–2 months.
-4. Someone goes back and tells the user: "your issue is fixed."
-
-Step 4 is the one most teams skip — and it is the step that closes the loop and drives retention. Feedback that visibly changes the product turns reviewers into advocates.
-
-### Trust Benchmark (Self-Reported)
-
-One founder's report: daily frontline chats with users built more trust in 2 months than colleagues had accumulated in over a year. Trust with users compounds at the speed of contact, not tenure.
-
-*Distilled from 12 founder podcast interviews (2024–2026); figures are self-reported or reported.*
-
----
-
-## 🤝 Work With Iris
-
-Need 1-on-1 PMF strategy or founder-led interview coaching? Reach out via [Telegram @Iris_carrot](https://t.me/Iris_carrot) or visit [gingiris.tools](https://gingiris.tools).
-
----
-
-## English
-
-### Why User Interviews?
-
-> "Product founders spend the first 6 months doing intensive user interviews. HeyGen's founder did 937 interviews in 6 months. Wisperflow and Higgsfield both did 500+. It's constant interviewing and iterating."
->
-> — Iris (生姜iris)
-
-| Value | Description |
-|---|---|
-| 🎯 Validate Needs | Confirm users actually need this feature/product |
-| 🔍 Discover Pain Points | Find friction in the user journey |
-| 📖 Collect Stories | Real user stories for marketing & fundraising |
-| 🤝 Build Relationships | Convert high-value users into ambassadors |
-| 🚀 Guide Iteration | Turn feedback into product roadmap |
-
-### Key Benchmarks
-
-| Metric | Target |
-|---|---|
-| Registration → Paid | 5–8% |
-| Interview Frequency | 3–5 per week minimum |
-| Actionable Insights | 3+ feature requests per interview |
-| Sean Ellis PMF Threshold | ≥ 40% "very disappointed if removed" |
-
-### Pre-Interview Preparation
-
-**Step 1: Define Interview Goals**
-- What is the core question for this interview?
-- What hypothesis are you validating?
-- Which user segment are you targeting?
-
-**Step 2: Screen Interview Candidates**
-
-| Priority | User Type | Description |
-|---|---|---|
-| 🔴 P0 | Paying Users | Already paid — highest value |
-| 🔴 P0 | Power Users | Daily usage, deep product knowledge |
-| 🟠 P1 | Competitor Users | Using competing products |
-| 🟠 P1 | Churned Users | Previously active, now gone |
-| 🟡 P2 | Registered Non-Payers | Interested but not converted |
-
-**Step 3: Schedule Users**
-
+npx skills add Gingiris-1031/gingiris-skills --skill gingiris-user-interview
 ```
-Hi [Name],
-
-I'm [Your Name] from [Company]. We noticed you've been using [Product]
-and would love to invite you to a 30-minute interview to understand
-your experience and feedback.
-
-The session will be recorded for internal review only, never shared publicly.
-
-Would you have time this week?
-
-Best,
-[Your Name]
-```
-
-**Follow-up Cadence**: Day 1 invite · Day 3 first follow-up · Day 7 second · Day 10 final · Multi-channel: LinkedIn, email, Telegram, Discord.
-
-### Interview Execution
-
-| Item | Requirement |
-|---|---|
-| Duration | 30–45 minutes |
-| Format | Video call (Zoom/Meet/Teams) |
-| Recording | **Mandatory** — including screen share |
-| Team | 1 moderator + 1 note-taker |
-
-### Opening Script (5 min)
-
-1. **Warm-up + Introduction**: "Hi, thanks for taking the time. I'm [Name], working on product at [Company]."
-2. **Set Expectations**: "Today I'd love to learn about your experience and hear feedback to help us improve."
-3. **Get Recording Consent**: "We'll record for internal learning only — never shared publicly. Camera off is fine. Okay?"
-4. **Background**: "Could you briefly introduce yourself?"
-
-### Core Question Framework
-
-**1. Background (Must-Ask)**
-
-| Question | Purpose |
-|---|---|
-| What do you do for work? | Understand user persona |
-| How did you hear about us? | Identify acquisition channels |
-| How long have you been using the product? | Gauge user maturity |
-| Which features do you use most? | Identify core use cases |
-| What other similar tools do you use? | Understand competitive landscape |
-
-**2. Deep Dive (Core)**
-
-*Current Workflow:*
-1. What problem are you solving with our product?
-2. What do you do most frequently in our product?
-3. How did you solve this problem before us?
-4. What's changed since you started using us?
-
-*Competitor Comparison:*
-1. What other similar products are you using?
-2. What do you use each product for?
-3. How do we compare to competitors?
-4. What do competitors do better than us?
-
-*Pain Point Discovery:*
-1. What frustrates you about using our product?
-2. Have you encountered any bugs or issues?
-3. Is there a feature that's particularly hard to use?
-4. **If you had a magic wand, what would you change?**
-
-*Willingness to Pay:*
-1. What tools have you paid for before?
-2. What made you decide to pay?
-3. How much would you pay for our product?
-4. What would make you upgrade to paid?
-
-**3. Screen Sharing (Critical)**
-
-> "Have users share their screen during interviews. Watch how they use it. See where they get stuck." — Iris
-
-What to observe: navigation patterns, friction points, used vs. ignored features, emotional reactions.
-
-Prompt: "Could you show me how you typically use our product? Just do what you'd normally do — no need to prepare anything special."
-
-### Closing Script
-
-1. **Ambassador Invite**: "We have a user ambassador program — early access + direct input. Interested?"
-2. **Referrals**: "Do you know anyone else who might benefit from our product?"
-3. **Thank You**: "Thanks so much — your feedback is incredibly valuable!"
-
-### Post-Interview Debrief (Required)
-
-See [Templates](references/templates.md#interview-debrief-template).
-
-### User Grading Standards
-
-**Grading Dimensions**
-
-| Dimension | Levels |
-|---|---|
-| Usage Frequency | Daily / Weekly / Monthly |
-| Usage Depth | Core features / Advanced / API |
-| Payment Status | Free / Basic / Pro / Enterprise |
-| Engagement Level | Passive / Active / Proactive |
-| Social Influence | Regular user / KOL / Community leader |
-
-**User Tiers**
-
-| Tier | Characteristics | Strategy |
-|---|---|---|
-| 🔴 S-Tier | Power user + Paid + Willing to advocate | Ambassador priority, CEO/PM direct contact |
-| 🟠 A-Tier | Power user + Payment potential | Beta access, regular check-ins |
-| 🟡 B-Tier | Occasional use + No clear payment intent | Guide to core features, push conversion |
-| ⚪ C-Tier | Low activity + No payment intent | Reduce investment, monitor natural retention |
-
-### Templates & Tools
-
-| Category | Tool | Use Case |
-|---|---|---|
-| Survey | Typeform | User research surveys |
-| Recording | Grain, Fathom | Meeting transcription |
-| Database | Airtable | User database |
-| Scheduling | Calendly | Appointment booking |
-| Outreach | LinkedIn | B2B user contact |
-
----
-
-## ❓ FAQ
-
-**Q: What's the best Claude skill for user interview and PMF discovery?**
-A: `gingiris-user-interview` is built from the HeyGen 937-interview methodology that found PMF, plus Wisperflow and Higgsfield 500+ interview programs. It covers founder-led research SOP: screening rubrics (P0/P1/P2 priority), JTBD interview scripts, screen-share observation protocols, post-interview debrief templates, and the 4-tier user grading system (S/A/B/C across 5 dimensions).
-
-**Q: How many user interviews do I need to find PMF?**
-A: The benchmark from successful founders: HeyGen 937 in 6 months, Wisperflow and Higgsfield 500+ each. The pattern isn't "do X interviews then stop" — it's continuous interviewing while iterating product. Aim for 3-5/week minimum cadence, sustained for 6+ months until you hit ≥40% "very disappointed if removed" (Sean Ellis test).
-
-**Q: How do I screen users for high-quality interviews?**
-A: Three-priority system. P0 (must-interview): Paying users + Power users who use daily. P1 (high-value): Competitor users + Churned users. P2 (lower priority): Registered non-payers. Multi-channel outreach via LinkedIn + email + Telegram + Discord with a 4-touch cadence: Day 1, 3, 7, 10.
-
-**Q: What's the JTBD interview script structure?**
-A: 4 questions that surface real motivation: "What were you doing the day you decided you needed [product]?" / "What did you try before you found us?" / "What almost stopped you from signing up?" / "What's changed since you started using us?" Plus screen-share observation — what they actually do contradicts what they say they do.
-
-**Q: How do I grade users into S/A/B/C tiers?**
-A: 5 dimensions × 4 tiers framework. Dimensions: Usage Frequency, Usage Depth, Payment Status, Engagement Level, Social Influence. S-Tier (Power + Paid + Advocate) gets CEO direct contact; A-Tier (Power + Payment potential) gets beta access; B-Tier (Occasional) gets conversion nudges; C-Tier (Low activity) monitors natural retention with minimal investment.
-
-**Q: Who built this?**
-A: Iris Wei (生姜) — former cofounder/COO of [AFFiNE](https://github.com/toeverything/AFFiNE) ($10M raised, Forbes Asia 30 Under 30, 60K+ stars). Advisor to 150+ AI startups on growth and PMF strategy.
-
----
-
-## 中文版
-
-详细中文内容请参考 [references/zh/README.md](references/zh/README.md)。
-
-### 快速导航
-
-| 主题 | 链接 |
-|------|------|
-| 访谈准备 | [references/zh/preparation.md](references/zh/preparation.md) |
-| 访谈执行 | [references/zh/execution.md](references/zh/execution.md) |
-| 问题框架 | [references/zh/questions.md](references/zh/questions.md) |
-| 模板库 | [references/templates.md](references/templates.md) |
-
-### 核心要点
-
-| 项 | 数值 |
-|---|---|
-| HeyGen 找到 PMF 的访谈次数 | **937 次 / 6 个月** |
-| Wisperflow / Higgsfield 类似量级 | **各 500+ 次** |
-| 注册→付费转化目标 | **5–8%** |
-| 最低访谈频率 | **每周 3–5 次** |
-| Sean Ellis PMF 阈值 | **≥ 40% "失去会非常失望"** |
-
----
-
-## 日本語
-
-**Gingiris ユーザーインタビュープレイブック** — PMF発見のための体系的なユーザーインタビューフレームワーク。HeyGenが6ヶ月で937回のインタビューを行った同じ方法論。Wisperflow と Higgsfield も同様に 500+ 件実施。
-
-**コアコンテンツ**:招待テンプレート、コア質問フレームワーク(JTBD)、スクリーン共有観察、インタビュー後分析SOP、ユーザー階層付け(S/A/B/C tier)。
-
----
-
-## 한국어
-
-**Gingiris 사용자 인터뷰 플레이북** — PMF 발견을 위한 체계적인 사용자 인터뷰 프레임워크. HeyGen이 6개월 동안 937번의 인터뷰를 수행한 동일한 방법론. Wisperflow 와 Higgsfield 도 각각 500+ 회 진행.
-
-**핵심 내용**: 초대 템플릿, 핵심 질문 프레임워크(JTBD), 화면 공유 관찰, 인터뷰 후 분석 SOP, 사용자 등급 체계(S/A/B/C tier).
-
----
-
-## 🤝 About the Author
-
-**Iris (生姜iris)** — Former cofounder & COO of AFFiNE, led global go-to-market from 0 to millions of users.
-
-- 💬 Telegram: [@Iris_carrot](https://t.me/Iris_carrot)
-- 🐦 Twitter: [@WeiYipei](https://x.com/WeiYipei)
-- 📱 即刻: 生姜iris
-- 📧 公众号: 云宝的桃花坞
-- 🌐 Blog: [gingiris.tools](https://gingiris.tools)
-
----
-
-## License
-
-MIT — Free for personal, commercial, learning, and derivative use.
-
----
-
-## 🗂️ See Also
-
-| Playbook | Focus |
-|:---|:---|
-| [gingiris-b2b-growth](https://huggingface.co/datasets/Gingiris/gingiris-b2b-growth) | 📈 B2B PLG/SLG — pair with user interviews for activation insights |
-| [gingiris-launch](https://huggingface.co/datasets/Gingiris/gingiris-launch) | 🚀 Pre-launch user validation + PMF feedback loop |
-| [gingiris-skills](https://huggingface.co/datasets/Gingiris/gingiris-skills) | 🧭 Meta-router: discovers the right playbook for your stage |
-
-Browse all 42 playbooks at [gingiris.tools/skills/](https://gingiris.tools/skills/).
-
-<!-- JSON-LD for SEO/GEO. AI engines (ChatGPT, Perplexity, Claude, Gemini) parse this. -->
-<script type="application/ld+json">
-{
-  "@context": "https://schema.org",
-  "@graph": [
-    {
-      "@type": "SoftwareApplication",
-      "name": "Gingiris User Interview Playbook — HeyGen 937-Interview PMF Discovery Framework",
-      "applicationCategory": "DeveloperApplication",
-      "applicationSubCategory": "Claude Skill",
-      "operatingSystem": "Any (Claude Code, Cursor, Codex, Amp, Cline)",
-      "description": "Battle-tested user interview framework for PMF discovery. Documents the HeyGen 937-interview / Wisperflow 500+ / Higgsfield 500+ founder-led research methodology. Includes JTBD interview scripts, screening rubrics (P0/P1/P2 priority), screen-share observation protocols, 4-tier user grading (S/A/B/C across 5 dimensions), post-interview debrief templates, and churn diagnostics. Targets 5-8% registration-to-paid conversion via 3-5 interviews/week sustained cadence.",
-      "url": "https://gingiris.tools/skills/",
-      "downloadUrl": "https://huggingface.co/datasets/Gingiris/gingiris-user-interview",
-      "installUrl": "https://skills.sh/Gingiris-1031/gingiris-user-interview",
-      "license": "https://opensource.org/licenses/MIT",
-      "offers": {
-        "@type": "Offer",
-        "price": "0",
-        "priceCurrency": "USD"
-      },
-      "author": {
-        "@type": "Person",
-        "name": "Iris Wei",
-        "alternateName": "生姜",
-        "url": "https://gingiris.tools",
-        "sameAs": [
-          "https://x.com/WeiYipei",
-          "https://www.linkedin.com/in/yipei-wei-550825105/"
-        ]
-      },
-      "keywords": "user interview, user research, PMF, product-market fit, customer discovery, JTBD, jobs to be done, HeyGen 937 interviews, Wisperflow, Higgsfield, founder-led research, screening rubric, user grading, S-tier A-tier B-tier C-tier, churn diagnostics, Sean Ellis test, screen sharing, willingness to pay, AFFiNE 60K stars"
-    },
-    {
-      "@type": "FAQPage",
-      "mainEntity": [
-        {
-          "@type": "Question",
-          "name": "What's the best Claude skill for user interview and PMF discovery?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "gingiris-user-interview is built from the HeyGen 937-interview methodology that found PMF, plus Wisperflow and Higgsfield 500+ interview programs. Covers founder-led research SOP: screening rubrics, JTBD scripts, screen-share observation, debrief templates, and 4-tier user grading (S/A/B/C across 5 dimensions)."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "How many user interviews do I need to find PMF?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "HeyGen 937 in 6 months, Wisperflow and Higgsfield 500+ each. Pattern is continuous interviewing while iterating. Aim 3-5/week minimum, sustained 6+ months until you hit Sean Ellis ≥40% 'very disappointed if removed' threshold."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "How do I screen users for high-quality interviews?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "Three priorities. P0: Paying + Power users (daily usage). P1: Competitor users + Churned. P2: Registered non-payers. Multi-channel outreach via LinkedIn + email + Telegram + Discord with 4-touch cadence on Days 1, 3, 7, 10."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "What's the JTBD interview script structure?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "4 core questions surface real motivation: 'What were you doing the day you decided you needed [product]?' / 'What did you try before you found us?' / 'What almost stopped you from signing up?' / 'What's changed since you started using us?' Plus screen-share observation — actions contradict words."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "How do I grade users into S/A/B/C tiers?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "5 dimensions × 4 tiers. Dimensions: Usage Frequency, Usage Depth, Payment Status, Engagement Level, Social Influence. S-Tier (Power + Paid + Advocate) → CEO contact; A-Tier (Power + Payment potential) → beta access; B-Tier (Occasional) → conversion nudges; C-Tier (Low activity) → monitor only."
-          }
-        },
-        {
-          "@type": "Question",
-          "name": "Who built this?",
-          "acceptedAnswer": {
-            "@type": "Answer",
-            "text": "Iris Wei (生姜) — former cofounder/COO of AFFiNE ($10M raised, Forbes Asia 30 Under 30, 60K+ stars). Advisor to 150+ AI startups on growth and PMF strategy."
-          }
-        }
-      ]
-    }
-  ]
-}
-</script>
-
-<!-- gingiris-community-cta:start -->
-## 加入出海社群与购买资料包 · Community & Resources
-
-- **加入出海社群：** 微信添加 **staywildgrow**，留言「**加入社群**」即可。
-- **购买资料包：** 微信添加 **staywildgrow**，咨询购买 Gingiris 出海资料包，价格 **599 元人民币**。
 
-- **Join the global growth community:** Add **staywildgrow** on WeChat and send **“加入社群”** (join the community).
-- **Get the resource pack:** Add **staywildgrow** on WeChat to purchase the Gingiris global growth resource pack for **CNY 599**.
+Read [SKILL.md](SKILL.md) and its linked references for the complete workflow.
+完整方法与模板请阅读 SKILL.md 及其引用文件。
 
-社群与资料包为可选入口，不影响本开源 Skill 的安装和使用。
+This update separates observations, hypotheses and outcomes. It contains reusable methods and constructed examples, not private workshop recordings or student records. No guaranteed business outcome.
+本次更新区分观察、假设与实际结果；仅包含通用方法与教学示例，不含私人逐字稿、学员记录，不保证商业效果。
 
-The community and paid resource pack are optional; neither is required to install or use this open-source skill.
-<!-- gingiris-community-cta:end -->
+Source: [maintained GitHub package](https://github.com/Gingiris-1031/gingiris-skills/tree/main/skills/gingiris-user-interview).

SKILL.md

@@ -1,23 +1,40 @@
 ---
 name: gingiris-user-interview
-description: |
-  🇺🇸 User Interview & Cold-Start Operations Playbook — Battle-tested framework from HeyGen's 937 interviews to PMF. Complete SOP for user screening, interview execution, Beta testing methodology, cold-start user operations, churn analysis, and user grading systems.
+description: '🇺🇸 User Interview & Cold-Start Operations Playbook — Evidence-led research, usability testing and
+  follow-up. Complete SOP for user screening, interview execution, Beta testing methodology, cold-start user operations,
+  churn analysis, and user grading systems.
 
-  🇨🇳 用户访谈与冷启动运营实战手册 — 基于 HeyGen 937 场访谈验证的 PMF 方法论。包含用户筛选、访谈执行、Beta 测试设计、冷启动用户运营、流失分析、用户分级体系完整 SOP。
 
-  🇯🇵 ユーザーインタビュー＆コールドスタート運営プレイブック — HeyGenの937回インタビューで検証されたPMF方法論。ユーザースクリーニング、インタビュー実行、Betaテスト設計、コールドスタートユーザー運営、離脱分析、ユーザーグレーディング。
+  🇨🇳 用户访谈与冷启动运营实战手册 — 基于真实行为、可用性观察与回访验证需求。包含用户筛选、访谈执行、Beta 测试设计、冷启动用户运营、流失分析、用户分级体系完整 SOP。
 
-  🇰🇷 사용자 인터뷰 & 콜드 스타트 운영 플레이북 — HeyGen 937회 인터뷰로 검증된 PMF 방법론. 사용자 스크리닝, 인터뷰 실행, Beta 테스트 설계, 콜드 스타트 사용자 운영, 이탈 분석, 사용자 등급 시스템.
 
-  Triggers: "user interview" | "user research" | "customer discovery" | "PMF" | "product-market fit" | "beta test" | "beta testing" | "cold start users" | "cold start operations" | "churn analysis" | "user grading" | "用户访谈" | "用户调研" | "冷启动" | "冷启动运营" | "Beta测试" | "流失分析" | "用户分级"
-when_to_use: |
-  Use this skill when you need to: conduct user interviews to find PMF, design Beta testing
-  methodology, operate cold-start user cohorts, analyze churn, build user grading systems,
-  or apply JTBD framework to validate product direction.
-  Trigger phrases: "user interview" | "PMF" | "customer discovery" | "beta test" |
-  "用户访谈" | "冷启动运营" | "PMF验证" | "流失分析" | "JTBD"
-source: https://github.com/Gingiris-1031/gingiris-skills/tree/main/skills/gingiris-user-interview
-tags:
+  🇯🇵 ユーザーインタビュー＆コールドスタート運営プレイブック — 実際の行動とユーザビリティテストに基づく需要検証。ユーザースクリーニング、インタビュー実行、Betaテスト設計、コールドスタートユーザー運営、離脱分析、ユーザーグレーディング。
+
+
+  🇰🇷 사용자 인터뷰 & 콜드 스타트 운영 플레이북 — 실제 행동과 사용성 관찰에 기반한 수요 검증. 사용자 스크리닝, 인터뷰 실행, Beta 테스트 설계, 콜드 스타트 사용자 운영, 이탈 분석, 사용자
+  등급 시스템.
+
+
+  Triggers: "user interview" | "user research" | "customer discovery" | "PMF" | "product-market fit" | "beta test"
+  | "beta testing" | "cold start users" | "cold start operations" | "churn analysis" | "user grading" | "用户访谈" |
+  "用户调研" | "冷启动" | "冷启动运营" | "Beta测试" | "流失分析" | "用户分级"
+
+  '
+metadata:
+  version: 1.1.0
+  when_to_use: 'Use this skill when you need to: conduct user interviews to find PMF, design Beta testing
+
+    methodology, operate cold-start user cohorts, analyze churn, build user grading systems,
+
+    or apply JTBD framework to validate product direction.
+
+    Trigger phrases: "user interview" | "PMF" | "customer discovery" | "beta test" |
+
+    "用户访谈" | "冷启动运营" | "PMF验证" | "流失分析" | "JTBD"
+
+    '
+  source: https://github.com/Gingiris-1031/gingiris-skills/tree/main/skills/gingiris-user-interview
+  tags:
   - user-interview
   - user-research
   - pmf
@@ -51,9 +68,7 @@ tags:
 
 ## 核心原则
 
-> "产品的 founder，在刚开始的前半年都会密集的做大量的用户访谈。像 HeyGen 就是半年，founder 自己做了 937 场；Wisperflow 和 Higgsfield 都做了 500 场以上——就是不断做访谈，不断改。" —— 生姜iris
-
-**关键洞察**：用户访谈不是调研工具，是 PMF 发现引擎。访谈频率与产品成功率正相关。
+围绕当前决策观察真实任务，记录反证，修改后回访验证。访谈数量与频率由样本差异、问题复杂度和团队资源决定，不是成功率或 PMF 的证明。
 
 ---
 
@@ -62,30 +77,25 @@ tags:
 把访谈从"调研动作"变成"日常运营循环"的最短路径，已在多个 AI Agent 项目全线复用：
 
 1. 后台按 token 消耗 / 用量拉 **Top10 用户 list**（每周刷新）
-2. 约**共享屏幕访谈，必须录屏**——看真实操作 > 听口述
-3. **Transcript 喂 AI**，自动提取 feature request / bug 直接进 Linear（或任意 issue 工具）
-4. **次日回访**：告知"你提的 X 已修/已排期"——反馈闭环本身就是留存杠杆
-5. 节奏：**每天至少 2 场**；访谈后截 5-10 分钟关键片段直接给产研
+2. 约**共享屏幕访谈；录制需同意，可不录制**——看真实操作 > 听口述
+3. **脱敏 transcript**：确认接收服务、数据范围与同意后生成需求草稿，审阅后再按授权写入 Linear
+4. **次日回访**：告知"你提的 X 已修/已排期"，仅在状态已经核实后告知；排期不等于修复
+5. 节奏按研究问题与团队复盘能力安排；仅向获准的产研成员分享已同意录制且脱敏的片段
 
 **配套动作**：
 - 新访谈员先 mock interview + 全程录屏 + 复盘 highlight，再上真实用户
 - **KOL 访谈二合一**：合作 YouTuber 顺便做用户访谈 + 可用性测试，一次合作拿两份价值；视频访谈 20-30 分钟 ROI 最高
-- 访谈资产化：一场付费用户访谈可同时产出功能需求 + SEO 素材 + reseller 线索三件套
+- 访谈资产化：访谈可产出需求证据；将内容用于SEO公开引用或商业线索需另行许可，不默认改变研究数据用途
 
 ### 邀约触达节奏（2026 更新）
 
-- **1/3/7/14 天四次触达**：Day 1 首发 → Day 3 → Day 7 → Day 14 各跟进一次
-- **双通道**：邮件模板 + 邮箱反查 LinkedIn 私信；邮件无回复转 LinkedIn 补一轮
+- 只向同意接收研究邀请的人发送邀请；未回复或拒绝后停止，不按固定次数追访。
+- **访谈邀请**：使用用户同意的联系渠道；不反查其他账号追访，未回复或拒绝后停止跟进
 - 话术用宽泛理由（"团队早期想做用户访谈"），**不暴露真实筛选原因**，避免用户带着预设表演
 
-### 标杆锚点
-
-Wisprflow：launch 前人工 onboard 前 500 名用户、逐个视频访谈，用积累的势能带 launch；HuggingFace：scale up 前 8 个月调研 + 500 场用户访谈。
+### 案例使用边界
 
-**2026-07 增补（播客一手提炼）**：
-- 据转述（AFFiNE 前 COO 口径）：Gamma / Notion / HeyGen 均在半年内做了约 1,000 场用户访谈，"用户提到的不好的点就真的去解"。
-- Gamma / HeyGen / Lovable 等成功产品的共同路径：半年 500-1,000 场访谈打磨产品细节，才有资格谈增长——"大量的人忽视了前面这半年到一年脏活累活的过程"。
-- 反例（匿名项目，reported）：跳过访谈打磨直接烧钱投放，单个注册成本高达几十美金、单个付费用户成本几百美金——"没有逻辑，但有钱"。
+历史案例中的访谈场数属于来源口述，口径未统一，不用作增长资格门槛。可迁移的是观察问题、修正产品、复测真实任务的过程。
 
 ---
 
@@ -104,13 +114,13 @@ Wisprflow：launch 前人工 onboard 前 500 名用户、逐个视频访谈，
 
 ### Step 2：邀约与排期
 
-**跟进节奏**：Day 1 → Day 3 → Day 7 → Day 10（全渠道：LinkedIn/Email/Telegram/Discord）
+**跟进节奏**：使用参与者同意的渠道与频次；无回复或拒绝后停止，不跨渠道追访。
 
 **邀约话术核心**：说明目的 + 时长承诺（30分钟）+ 录屏说明 + 价值交换
 
 ### Step 3：访谈执行（30-45分钟）
 
-**必须**：1人主持 + 1人记录 + 必须录屏（含共享屏幕）
+**人员与记录**：1人主持、1人记录；录制和共享屏幕都需自愿同意，可选择口述或笔记，拒绝录制不影响参与。
 
 **核心问题框架**：
 1. **背景了解**：职业、获客渠道、使用时长、核心功能、竞品
@@ -157,7 +167,7 @@ Wisprflow：launch 前人工 onboard 前 500 名用户、逐个视频访谈，
 1. **第一场访谈可以做满 2 小时**：深度用户往往比你更了解场景、行业和所有竞品，能帮你总结出真正的价值主张和当下最大的比较性优势。
 2. **真诚运营换深度反馈**：一位早期海外深度用户后来主动写了 2,000 字带截图动图的长评，逐条指出"花了太多时间在用户不需要的功能场景上"+ 具体 bug——"如果不是因为你太真诚，我不会花两小时写这个"。
 3. **全员拆解 → 限期修完**：收到长评后全员开会逐条拆解，用 1-2 个月修完其中全部 bug。
-4. **闭环回告**：修完/排期后回访用户"你提的 X 已修"——反馈闭环本身就是留存杠杆（呼应 Token Top10 第 4 步）。
+4. **闭环回告**：修完/排期后回访用户"你提的 X 已修"，仅在状态已经核实后告知；排期不等于修复（呼应 Token Top10 第 4 步）。
 5. **踩坑警示**：访谈反馈上传后若仍由 1 号位单人拍板取舍，大量用户输入会被忽略——产研运必须共同定目标、共同消化访谈精华。
 
 **信任建立速度基准（自述口径）**：无论职级多高，保持每天一线直接和用户聊天，2 个月建立的用户信任���以超过不下场的同事 1 年多——用户会主动掏心掏肺、跨部门为你背书。一线高频直聊是访谈质量的前提。
@@ -241,13 +251,11 @@ Wisprflow：launch 前人工 onboard 前 500 名用户、逐个视频访谈，
 | 工具类 | 开始跑任务，或单次使用超 5 分钟 |
 | DAU 口径 | 以 unique message send 计，剔除内部用户 |
 
-**转化合格线**：
-- 注册 → 付费 **>5% 为合格线**：Pro-C 5-8%、SMB 10-15%（大 C 类 1.5-3%）
-- 阶段目标要"够得着"：付费率 0.1% 的产品先定 1%/3 个月，不要直接定 5%
+**目标与口径**：按产品阶段、渠道、用户 cohort 和观察窗口设目标，保留基线和假设；不使用固定注册付费比例认定合格或 PMF。
 
-**流失可运营化**：两次对话后 24h 不回访 = 流失，自动触发挽回邮件——流失定义一旦行为化，挽回就能自动化。
+**流失**：按产品自然使用周期定义，短期未回访先作为信号。是否发送挽回消息按用户授权和联系偏好处理，不因一个时间阈值自动外发。
 
-**付费归因四时间戳**：注册时间 × 激活时间 × 支付时间 × 最近登录，四点交叉判定真实付费驱动（是哪个功能、哪场访谈、哪次触达带来的付费）。
+**付费时间线**：注册、激活、付款、最近使用时间用于整理先后顺序；时间相邻不能单独证明某次访谈或功能导致付款。
 
 ---
 
@@ -307,10 +315,15 @@ Wisprflow：launch 前人工 onboard 前 500 名用户、逐个视频访谈，
 
 | 文档 | 内容 |
 |------|------|
-| [references/interview-guide.md](references/interview-guide.md) | 完整用户访谈指南（HeyGen 937场方法论、问题框架、模板、FAQ） |
+| [references/interview-guide.md](references/interview-guide.md) | 完整用户访谈指南（证据与抽样、问题框架、模板、FAQ） |
 | [references/cold-start-ops.md](references/cold-start-ops.md) | Beta测试设计框架、冷启动用户运营、流失分析、可复用模板 |
 | [references/templates.md](references/templates.md) | 访谈复盘模板、用户画像表、Feature Request Tracker |
 
+
+## 访谈纠偏与冷启验收
+
+访谈变推销、用户反向提问或主持人代操作时，读取 [纠偏与交接](references/interview-recovery.md)。研究要转为获客计划时，读取 [冷启动目标与成本卡](references/cold-start-decision-card.md)。按决策选择样本，不用访谈场数、问价或某个固定付费率证明 PMF。
+
 <!-- gingiris-community-cta:start -->
 ## 加入出海社群与购买资料包 · Community & Resources
 

references/cold-start-decision-card.md

@@ -0,0 +1,16 @@
+# Cold-start decision card / 冷启动目标与成本
+
+Use when turning research into a small acquisition or validation experiment. Set one primary decision, not a list of channels to fill.
+
+- Segment, task, stage and eligibility; why this cohort fits.
+- One initial channel, example leads with source/fit (not scraped bulk outreach), message drafts matched to the task, and one useful content example.
+- Outcome: following, reply, meeting, activated user, retained use or actual payment. Define each separately, with window and deduplication key.
+- Cash, tools and human hours; budget is an upper bound, not an obligation to spend. Zero cash does not mean zero cost.
+- Baseline and forecast assumptions; unknown conversion/CAC/LTV stay unknown. Distinguish cash CAC from fully loaded cost and never divide by projected payers as if they were actual customers.
+- Owner, time/budget cap, next review, continue/change/stop conditions.
+
+If the original request is “paying customers” but the sensible next step is “interviews”, explain original target → proposed target → evidence/reason → question still unanswered → when to revisit payment. Do not silently replace paid acquisition with follower growth. Research drafts do not send messages without authorization.
+
+Constructed example: third-party posts already mention an open-source tool, but interested users cannot find its author. Test an author entry and a relevant task invitation. Count replies, completed tasks and payments separately; stars, clones, followers and asking about price do not establish paid demand.
+
+For content variants, keep project, format and landing page comparable, name the variable changed, retain IDs and measurement windows. Report counts as well as rates. Natural recommendation traffic is not a randomized experiment; tiny samples can be inconclusive. Only expand when downstream quality and the business cost limit support it.

references/cold-start-ops.md

@@ -405,4 +405,4 @@ Hi [Name]，
 - [ ] 邀请加入 Creative Partners 计划
 - [ ] 发送后续跟进邮件
 - [ ] 反馈同步给 PM
-```
+```

references/interview-guide.md

@@ -6,7 +6,7 @@
 
 ## 一、为什么要做用户访谈？
 
-> "产品的 founder，在刚开始的前半年都会密集的做大量的用户访谈。像 HeyGen 就是半年，founder 自己做了 937 场；Wisperflow 和 Higgsfield都做了 500 场以上——就是不断做访谈，不断改。"—— 生姜iris
+访谈案例中的历史数量不是 PMF 证明；围绕真实行为与复测结果判断。
 
 ### 1.1 核心价值
 
@@ -18,11 +18,11 @@
 | **建立关系** | 与高价值用户建立深度连接，转化为大使 |
 | **指导迭代** | 将用户反馈转化为产品 roadmap |
 
-### 1.2 关键指标 Benchmark
+### 1.2 研究计划与证据
 
-- **注册到付费率**：5%-8%
-- **访谈频率**：每周至少 3-5 场
-- **有效反馈收集**：每场访谈后至少产出 3 个可执行的 feature request
+- **注册到付费率**：记录同 cohort、窗口与去重人数，不设通用合格线
+- **访谈频率**：按问题、样本差异与复盘能力安排
+- **有效反馈收集**：记录原话、行为、反证和未知；不为凑数量制造 feature request
 
 ---
 
@@ -65,15 +65,9 @@
 
 **邀约话术**：
 
-> "Hi [名字]，我是 [公司] 的 [姓名]。我们注意到你最近在使用我们的 [产品]，想邀请你进行一个 30 分钟的访谈，了解你的使用体验和反馈。整个访谈过程会录屏，用于内部复盘，不会对外公开。请问你有时间吗？"
+> "Hi [名字]，我是 [公司] 的 [姓名]。我们注意到你最近在使用我们的 [产品]，想邀请你进行一个 30 分钟的访谈，了解你的使用体验和反馈。如你同意，我们可录制用于约定范围的内部复盘；你可拒绝录制、随时暂停，不影响参与。公开引用需另行许可。请问你有时间吗？"
 
-**跟进策略**：
-
-- 第 1 天：首次邀请
-- 第 3 天：第一次跟进
-- 第 7 天：第二次跟进
-- 第 10 天：最后一次跟进
-- **全方位攻击**：LinkedIn、邮件、Telegram、Discord 等所有能找到用户的地方
+**跟进策略**：按约定渠道与频次联系；未回复或拒绝后停止，不跨渠道追访。
 
 ---
 
@@ -83,7 +77,7 @@
 
 - **时长**：30-45 分钟
 - **形式**：视频会议（Zoom/飞书/腾讯会议）
-- **录屏**：必须录屏，包含共享屏幕
+- **录屏**：取得明确录制同意后才录屏；可选择口述或遮蔽敏感信息
 - **人数**：1 人主持，1 人记录
 
 #### 访谈开场（5分钟）
@@ -407,7 +401,7 @@
 
 - 观察共享屏幕时的操作流畅度
 - 追问细节："能具体说说吗？"
-- 付费意愿是最真实的反馈
+- 付费表态是线索；历史购买、明确试点行动与实际付款分别记录
 
 ---
 
@@ -418,4 +412,4 @@
 
 ---
 
-*本手册由生姜小助手整理，整合自历次会议纪要和最佳实践。如有问题，欢迎随时提问！🦞*
+*本手册由生姜小助手整理，整合自历次会议纪要和最佳实践。如有问题，欢迎随时提问！🦞*

references/interview-recovery.md

@@ -0,0 +1,34 @@
+# Interview recovery and handoff / 访谈纠偏与交接
+
+Choose the objective before the session. Discovery reconstructs recent real behavior; usability testing observes a defined task; sales/support answers adoption questions. They can share a meeting, but label the transition and do not score a persuasive demo as unassisted usability success.
+
+| Trigger | Recovery | Record |
+|---|---|---|
+| Host explains features at length | Ask “What were you doing the last time this happened?” | User task, current workaround, actual impact |
+| User asks product questions throughout | Acknowledge and note the question; offer to answer after the research segment, unless it blocks participation | Unresolved concern, agreed follow-up |
+| Host operates the demo | Invite the user to attempt the agreed task; help only when needed, noting the intervention | Action, hesitation, error, assistance and result |
+| No real task is defined | Ask about the user's product/workflow and select a representative task together | Why this task represents their needs |
+
+Example rescue: “这个问题我记下了，稍后单独回答。先回到你最近一次处理这件事：当时具体做了什么？” Do not ignore material safety, cost or consent questions to retain control of the interview.
+
+Before a usability test, explain prototype limitations and use a suitable sandbox where possible. Screen sharing and recording are optional and consensual. A research invitation does not authorize production integration, financial transactions or sharing private data. During the task, avoid narrating advantages; ask what the user expects and observe what they do.
+
+Separate: cannot understand (expression/interface); understands but does not need it (value/persona); needs it but cannot adopt (capability, access, input or integration). A demo can test understanding and navigation; claims of time saved or business outcomes need real-use evidence.
+
+## Evidence and follow-up
+
+Keep user quotation/timecode, task, observed action, host prompt, blocker, current workaround and proposed solution in separate fields. Handoff clips include start/end time, consent scope, redactions, issue owner, observable acceptance condition and revisit date. A clip index is often more usable than an unexplained full recording.
+
+Re-test the changed task with an appropriate participant, ideally including the person who encountered it. Tell users “fixed” or “scheduled” only after checking the actual state. Record a proposed trial, an agreed trial and a completed action separately. Willingness-to-pay statements and requests for more credits are clues, not payment.
+
+## Sampling and evidence strength
+
+Choose participants for the decision: paying, heavy-use, new, churned and non-adopting users answer different questions. Start with a manageable batch and reassess gaps and counterexamples; no fixed interview count or conversion percentage certifies PMF. Do not manufacture feature requests to meet a quota. A participant who does not need a change can supply valuable counterevidence.
+
+Compensation pays for time, not favorable answers; disclose and record it. Do not treat an incentivized interview as invalid by default, or all free-user feedback as useless. Historical purchases and a clearly scoped paid offer are stronger pricing evidence than a hypothetical yes.
+
+## Facilitator exercise (constructed, not a customer result)
+
+Bad: “Our integration is easy; let me show all its advantages.”
+Better: “What are you trying to charge for? Please try configuring that in this test environment.”
+Acceptance: the researcher identifies whether the person can complete the task, records assistance, and distinguishes an onboarding blocker from an unsupported product promise. No invented conversion result.

references/podcast-evidence.md

@@ -0,0 +1,7 @@
+# Podcast evidence for interviews
+
+- An early AFFiNE interview lasted about two hours and later produced a detailed illustrated review. The lesson is depth and trust, not a duration quota.
+- Ask the user to screen-share the real workflow; abstract opinions are weak evidence.
+- Enthusiasm does not equal paid-user fit. Tag use case, urgency, alternative and paid behavior.
+- Close the loop with product and engineering, verify after release and notify the user.
+- Claims that other companies ran 1,000 interviews are secondary recollections, not verified primary facts.

references/templates.md

@@ -145,7 +145,7 @@ Hi [名字]，
 
 我是 [公司] 的 [姓名]。我们注意到你最近在使用我们的 [产品]，想邀请你进行一个 30 分钟的访谈，了解你的使用体验和反馈。
 
-整个访谈过程会录屏，用于内部复盘，不会对外公开。
+如你同意，我们可录制用于约定范围的内部复盘；你可拒绝录制、随时暂停，不影响参与。公开引用需另行许可。
 
 作为感谢，我们会提供 [礼品卡/内测资格/其他福利]。