# Jtbd Interview > Generate a customized Switch interview script based on your product context and existing evidence gaps. Outputs a ready-to-use markdown guide with Moesta's timeline reconstruction questions and force-probing techniques. Use when: "interview script", "jtbd interview", "interview guide", "how to interview". - Skill: `savvides/jtbd-interview` (Agent Skill) - Install (CLI): `npx skillmds@latest add savvides/jtbd-interview` - Raw SKILL.md: https://api.skillmd.com/api/skills/savvides/jtbd-interview/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Docs & Writing - Author: savvides (https://skillmd.com/u/savvides) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/savvides/jtbd-interview --- ## Preamble ```bash # Detect .jtbd/ directory _JTBD_DIR=".jtbd" _HAS_JTBD="no" [ -d "$_JTBD_DIR" ] && _HAS_JTBD="yes" echo "JTBD_DIR: $_HAS_JTBD" # Detect gstack (optional integration) _HAS_GSTACK="no" [ -d "$HOME/.claude/skills/gstack" ] && _HAS_GSTACK="yes" _ROOT=$(git rev-parse --show-toplevel 2>/dev/null) [ -n "$_ROOT" ] && [ -d "$_ROOT/.claude/skills/gstack" ] && _HAS_GSTACK="yes" echo "GSTACK: $_HAS_GSTACK" # Detect git _HAS_GIT="no" git rev-parse --is-inside-work-tree 2>/dev/null && _HAS_GIT="yes" echo "GIT: $_HAS_GIT" # Read manifest if .jtbd/ exists if [ "$_HAS_JTBD" = "yes" ] && [ -f "$_JTBD_DIR/manifest.yml" ]; then echo "--- MANIFEST ---" cat "$_JTBD_DIR/manifest.yml" echo "--- END MANIFEST ---" fi # Count existing switch analyses if [ "$_HAS_JTBD" = "yes" ] && [ -d "$_JTBD_DIR/switches" ]; then _SWITCH_COUNT=$(ls "$_JTBD_DIR/switches/"*.yml 2>/dev/null | wc -l | tr -d ' ') echo "SWITCH_COUNT: $_SWITCH_COUNT" if [ "$_SWITCH_COUNT" -gt 0 ]; then echo "--- EXISTING SWITCHES ---" ls "$_JTBD_DIR/switches/"*.yml echo "--- END SWITCHES ---" fi else echo "SWITCH_COUNT: 0" fi ``` ## Initialize .jtbd/ (if needed) If `JTBD_DIR` is `no`, create the `.jtbd/` directory structure before proceeding. 1. Use AskUserQuestion to gather product info: > "No .jtbd/ directory found. I'll create one. What's your product called, and what stage are you at?" > > Options: > - A) Set up now (I'll provide product name and stage) > - B) Use defaults (product: "Unknown", stage: "pre-product") If A: Ask for product name, one-line description, stage (pre-product / has-users / has-paying-customers), target user role, and target user context. If B: Use defaults. 2. Create the directory structure: ``` .jtbd/ ├── manifest.yml ├── .gitignore ├── raw/ └── switches/ ``` 3. Write `manifest.yml` with the user's answers (or defaults): ```yaml schema_version: 1 product: name: "{product_name}" description: "{description}" stage: "{stage}" target_user: role: "{role}" context: "{context}" settings: auto_commit: true gitignore_raw: true evidence_threshold: 6 ``` 4. Write `.jtbd/.gitignore`: ``` # Raw interview transcripts contain PII - keep out of git by default raw/ ``` ## Gather Context ### Read existing evidence If `SWITCH_COUNT` > 0, read ALL existing switch analysis files from `.jtbd/switches/`. Analyze them for: 1. **Force gaps**: Which forces are thin across interviews? (e.g., habit forces often underprobed) 2. **Timeline gaps**: Which timeline stages are frequently `not_discussed`? 3. **Low-confidence areas**: Fields marked `confidence: low` or `inferred: true` 4. **Missing perspectives**: What roles/personas have been interviewed vs. not? 5. **Evidence strength patterns**: Which `evidence_strength` dimensions are consistently low? Store these gaps — they will be used to customize the interview script. ### Determine interview context Use AskUserQuestion: > "Who are you interviewing, and what's the context?" Options: - A) Someone who recently switched TO our product - B) Someone who recently switched AWAY from our product (churned) - C) Someone who recently switched between competitors (not our product) - D) Someone who is currently evaluating / hasn't decided yet Then ask: > "What do you already know about this person?" Options: - A) I have their name and role (I'll provide details) - B) I know their general persona but not specific details - C) Nothing yet — generate a general script If A: Collect name, role, company (optional), and any known context about their situation. If B: Collect the persona description. ## Generate the Interview Script Generate a markdown interview script customized to the context gathered above. The script must follow Moesta's Switch methodology precisely. ### Script Structure The script has 6 sections. Each section has: - A purpose statement (why you're asking these questions) - The questions themselves - Interviewer notes in blockquotes (tips, red flags, follow-up cues) #### Section 1: Opening & Rapport (2-3 minutes) Purpose: Make the interviewee comfortable. Establish that there are no wrong answers. You're interested in their story, not evaluating them. Generate 2-3 opening questions. These should be warm, easy, and related to their role. Example: ```markdown ## 1. Opening **Purpose:** Build rapport. No wrong answers — you want their story. - "Tell me a bit about your role at {company}. What does a typical day look like?" - "How long have you been in this role?" > **Interviewer note:** Keep this under 3 minutes. You're warming up, not interrogating. Match their energy. If they're terse, be terse. If they're chatty, let them talk. ``` #### Section 2: Set the Scene (3-5 minutes) Purpose: Establish what they were using before (the "old way") and get them to describe their world before the switch. This is critical context. Generate 3-4 questions. Key technique: ask about the **situation**, not opinions. ```markdown ## 2. The Old Way **Purpose:** Understand their world before the switch. What were they using? How did it work? - "Before you started using {new solution}, what were you using to {do the job}?" - "Walk me through how that worked on a typical {day/week/month}." - "How long had you been doing it that way?" > **Interviewer note:** Listen for workarounds. People compensate for bad tools without realizing it. "Oh, I had a spreadsheet for that" = habit force. Get specific: which spreadsheet? How often? Who maintained it? ``` #### Section 3: Timeline Reconstruction (10-15 minutes) Purpose: This is the core of the Switch interview. Reconstruct the timeline by working **backward** from the decision. Generate questions for each timeline stage. Start with the decision and work back to the first thought. **Deciding (start here):** ```markdown ## 3. The Timeline ### 3a. The Decision **Purpose:** Anchor on the specific moment they decided. Work backward from here. - "Take me back to the moment you decided to go with {new solution}. Where were you? What was happening?" - "What specifically tipped you over the edge? Was there a final straw?" - "Were you considering anything else at that point? What were the other options?" - "Who else was involved in the decision?" > **Interviewer note:** Push for specificity. "When was that exactly?" "Was it before or after [event]?" You want dates, locations, people. If they say "we just decided," probe: "Was there a meeting? An email? A conversation?" The decision always has a moment. ``` **Active Looking:** ```markdown ### 3b. Active Looking - "Before you decided, what did the search look like? Were you actively comparing options?" - "What triggered you to go from 'this is annoying' to 'I need to find something else'?" - "What did you Google? Who did you ask? What demos did you sit through?" - "How long did this phase last?" > **Interviewer note:** The trigger from passive to active is gold. It's usually a specific event — a bad meeting, a missed deadline, a public embarrassment. Probe until you get the story. ``` **Passive Looking:** ```markdown ### 3c. Passive Looking - "Before you were actively searching, were you aware that alternatives existed?" - "How did you first hear about {new solution} or tools like it?" - "How long were you in this 'I know there's better stuff out there but I'm not doing anything about it' phase?" > **Interviewer note:** This phase can last months or years. People tolerate a lot. The duration tells you how strong the habit forces are. ``` **First Thought:** ```markdown ### 3d. First Thought - "Can you remember the very first time you thought 'this isn't working'? What happened?" - "Was there a specific event, or was it a slow build?" > **Interviewer note:** The first thought is often surprisingly specific — a crash, a bad report, an embarrassing moment in front of a boss. If they say "I don't remember," try: "Was it before or after [major event]?" to jog memory. ``` #### Section 4: Force Probing (10-15 minutes) Purpose: Now that the timeline is established, probe each force explicitly. The timeline questions will have surfaced some forces naturally — this section fills gaps. **Push questions:** ```markdown ## 4. The Four Forces ### 4a. Push — What was broken? - "What was the most frustrating thing about {old solution}?" - "Tell me about a specific time it let you down." - "What did that cost you — time, money, reputation, stress?" > **Interviewer note:** Push forces must be specific events, not vague feelings. "It was slow" → "Tell me about a time the slowness actually caused a problem." Get the story, the stakes, the people involved. ``` **Pull questions:** ```markdown ### 4b. Pull — What attracted you? - "What was the first thing about {new solution} that caught your attention?" - "When you imagined using it, what did you picture being different?" - "Was there a specific feature or moment in the demo/trial that made you think 'this is it'?" > **Interviewer note:** Pull is about the *imagined future*, not features. "Real-time dashboards" is a feature. "My CEO can get his own numbers without asking me" is pull. Keep asking "why does that matter?" until you hit the emotional payoff. ``` **Anxiety questions:** ```markdown ### 4c. Anxiety — What scared you? - "What almost stopped you from switching?" - "What was the scariest part of the transition?" - "Was there a moment where you almost backed out?" - "What could have gone wrong?" > **Interviewer note:** People don't volunteer anxiety. You have to ask directly. Silence after the question is fine — let them think. Common anxieties: data loss, team resistance, looking foolish, wasted money, learning curve. If they say "nothing," try: "If you had to name one risk, even a small one..." ``` **Habit questions:** ```markdown ### 4d. Habit — What kept you on the old way? - "Was there anything about {old solution} that you actually liked or would miss?" - "Did anyone on your team resist the change? Why?" - "Were there workarounds you'd built that you had to give up?" - "What would have happened if you'd just... kept doing what you were doing?" > **Interviewer note:** Habit is the most underprobed force. People build elaborate workarounds and forget they're workarounds. "My analyst had a whole process" = someone's entire job was compensating for a broken tool. That's a habit force AND a push force. ``` #### Section 5: The Job Story (2-3 minutes) Purpose: Validate your understanding by reflecting back what you heard. ```markdown ## 5. Reflect Back **Purpose:** Confirm you understood the story correctly. - "Let me play this back to you. It sounds like [summarize the situation], and you wanted [motivation], so you could [outcome]. Does that sound right?" - "Is there anything I'm missing?" > **Interviewer note:** This is your job story draft. If they correct you, update it. Their correction is often more precise than your summary. Write it down verbatim. ``` #### Section 6: Wrap-up (2-3 minutes) ```markdown ## 6. Wrap-up - "If you could go back and give yourself advice at the start of this process, what would you say?" - "Is there anything I should have asked that I didn't?" - "Thanks — this was incredibly helpful." > **Interviewer note:** The "advice to yourself" question often surfaces the strongest anxiety or habit force. The "anything I missed" question catches things you didn't know to ask about. ``` ### Customization Rules Apply these customizations based on the context gathered: **If interview context is "switched AWAY" (churned):** - Flip the framing: the "old way" is YOUR product, the "new solution" is whatever they switched to - Add questions: "What would have kept you?" and "When did you first think about leaving?" - The push forces are about YOUR product's failures — this is painful but valuable **If interview context is "evaluating / hasn't decided":** - Focus on Sections 2 (old way), 3d (first thought), and 4a-4b (push and pull) - Skip 3a (deciding) — it hasn't happened yet - Add: "What would need to be true for you to make a change?" and "What's stopping you right now?" **If existing switch analyses have evidence gaps:** - Add targeted questions for the weak areas. For example: - If habit forces are thin across interviews: add 2 extra habit questions and a bold interviewer note: "⚠️ Habit forces are thin in your existing data. Spend extra time here." - If timeline stages are frequently `not_discussed`: add extra probes for those specific stages - If `evidence_strength.direct_quotes` is consistently low: add a note about getting verbatim quotes **If manifest has product/target_user info:** - Replace generic placeholders ({old solution}, {new solution}, {do the job}) with product-specific language - Tailor the "Old Way" questions to the target user's role and context ### Red Flags Reference Always include this section at the end of the script: ```markdown ## Red Flags to Watch For These patterns indicate you're getting low-quality data. Redirect when you spot them. | Red Flag | What It Sounds Like | Redirect | |----------|---------------------|----------| | **Hypothetical language** | "I would...", "I think I'd..." | "I want to hear what actually happened. Take me back to that moment." | | **Vague emotions** | "It was frustrating", "I was happy" | "Tell me about a specific time you felt that way. What happened?" | | **Category answers** | "We needed better analytics" | "What specifically couldn't you do? Walk me through the last time that was a problem." | | **No dates or people** | "At some point we decided..." | "Was that before or after [event]? Who was in the room?" | | **Feature lists** | "It has real-time dashboards and API access and..." | "Which of those mattered most to YOU? Why that one?" | | **Pleasing the interviewer** | "Your product is great because..." | "I appreciate that, but I'm more interested in the messy middle. What almost stopped you?" | ``` ### Interview Tips Always include this section: ```markdown ## Tips for a Great Switch Interview 1. **Record everything.** Use Fireflies, Otter, or your phone. You cannot take notes and listen at the same time. 2. **Shut up.** The best interviewers talk 20% of the time. Ask the question, then wait. Silence is productive. 3. **Follow the energy.** When they lean in or speed up, you've hit something. Probe deeper. 4. **No leading questions.** Not "Was it frustrating?" but "How did that go?" Let them name the emotion. 5. **Get the story, not the opinion.** "What happened?" beats "What do you think about...?" every time. 6. **30-45 minutes is the sweet spot.** Under 20 = too shallow. Over 60 = diminishing returns. 7. **The first answer is never the real answer.** Ask "why?" at least 3 times. The good stuff is 2-3 layers deep. ``` ## Human Review Present the generated script for review using AskUserQuestion: > **Interview Script Generated** > > **Target:** {interviewee context} > **Sections:** 6 (Opening → Old Way → Timeline → Forces → Reflect → Wrap-up) > **Estimated duration:** 30-45 minutes > **Customizations applied:** {list customizations, e.g., "Extra habit probes based on evidence gaps", "Churn-focused framing"} > > Options: > - A) Looks good, save it > - B) I want to edit something (tell me what to change) > - C) Start over If the user chooses B, make the requested changes and re-present. ## Write and Commit 1. Write the markdown file to `.jtbd/interviews/interview-script-{context-slug}-{YYYYMMDD}.md` using the Write tool. The `context-slug` should describe the interview target: - If a specific person: `{first-name}-{role-slug}` (e.g., `sarah-ops-manager`) - If a persona: `{persona-slug}` (e.g., `churned-enterprise-user`) - If general: `general` Create the `.jtbd/interviews/` directory if it doesn't exist. 2. If `GIT` is `yes` and the manifest has `auto_commit: true`: Substitute the real filename and context for `` and `` before running this. They sit in a bash block and are copied verbatim unless you replace them: an unsubstituted `git add` fails the pathspec and stages nothing. ```bash git add .jtbd/interviews/.md git commit -m "jtbd: add interview script for " ``` If the commit fails, write the file but skip the commit. Tell the user: "File written to .jtbd/interviews/{filename}.md but not committed. Run `git add` and `git commit` manually." 3. If `GIT` is `no`: write the file only. 4. Tell the user what was created and where: > "Interview script saved to `.jtbd/interviews/{filename}.md`" > > **Next steps:** > - Print it or open it on your phone during the interview > - Record the interview (Fireflies, Otter, or voice memo) > - After the interview, run `/jtbd-switch` with the transcript > - After 3+ switch analyses, run `/jtbd-patterns` to find cross-interview patterns