# Summarize Interview > - Skill: `frank-luongt/summarize-interview` (Agent Skill) - Install (CLI): `npx skillmds@latest add frank-luongt/summarize-interview` - Raw SKILL.md: https://api.skillmd.com/api/skills/frank-luongt/summarize-interview/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Research & Search - Author: frank-luongt (https://skillmd.com/u/frank-luongt) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/frank-luongt/summarize-interview --- --- name: summarize-interview description: Transform customer interview transcripts or notes into structured summaries with JTBD insights, pain points, and action items. Use when processing interview data after customer discovery sessions. tags: [discovery, customer-research, interviews, synthesis] --- # Summarize Interview Transform raw interview transcripts or notes into structured, actionable summaries that feed directly into Opportunity Solution Trees, PRDs, and product decisions. --- ## Purpose Interview transcripts are long, unstructured, and hard to compare across participants. This skill extracts the signal — jobs, pains, outcomes, and surprises — into a consistent format that enables pattern recognition across interviews. --- ## When to Use - After conducting a customer discovery interview - When processing a batch of interview transcripts - When synthesizing research for a product brief or PRD - When building or updating an Opportunity Solution Tree --- ## Required Input Provide one of: - Full interview transcript (text or audio-to-text) - Detailed interview notes - Recording summary from a note-taking tool --- ## Output Template For each interview, produce this structured summary: ```markdown # Interview Summary ## Metadata | Field | Value | | --- | --- | | Date | [interview date] | | Participant | [name or ID] | | Role / Title | [their role] | | Company / Context | [company, size, industry] | | Interviewer | [your name] | | Duration | [minutes] | --- ## Current Situation **What they do today:** [1–3 sentences describing how they currently handle the relevant activity] **Tools / processes used:** - [tool or process 1] - [tool or process 2] **How long they've done it this way:** [duration] --- ## Jobs to Be Done For each job identified: | Job | Importance (1–10) | Current Satisfaction (1–10) | Opportunity Score | | --- | --- | --- | --- | | [Job description — what they're trying to accomplish] | [rating] | [rating] | [Importance × (1 − Satisfaction/10)] | --- ## Pain Points | Pain Point | Severity | Frequency | Current Workaround | | --- | --- | --- | --- | | [pain description] | High / Medium / Low | Daily / Weekly / Occasionally | [how they cope] | --- ## Desired Outcomes - [What "success" looks like in their words] - [Measurable outcome they care about] --- ## What They Like About Current Solution - [positive aspect] — why it matters to them - [positive aspect] — why it matters to them --- ## Willingness to Change / Pay - **Have they searched for alternatives?** [Yes/No — details] - **Have they spent money on this?** [Yes/No — how much, on what] - **Priority vs. other problems:** [High / Medium / Low] - **Switching cost concern:** [what would make switching hard] --- ## Key Insights - [unexpected finding 1] - [unexpected finding 2] --- ## Notable Quotes > "[exact verbatim quote — the most revealing thing they said]" > "[another important quote]" --- ## Action Items | Action | Owner | Due | | --- | --- | --- | | [follow-up action from this interview] | [who] | [when] | ``` --- ## Processing Rules 1. **Use their words, not yours** — preserve the participant's language, especially in quotes and job descriptions 2. **Fill in "—" for unavailable information** — never guess or infer missing data 3. **Flag contradictions** — if they say one thing but describe doing another, note both 4. **Separate observation from interpretation** — state what they said, then (separately) what you think it means 5. **Rate conservatively** — when scoring importance/satisfaction, err toward the middle unless evidence is strong 6. **One summary per interview** — don't combine multiple participants into one summary --- ## Cross-Interview Synthesis When processing **3+ interviews**, also produce a synthesis table: ```markdown ## Cross-Interview Patterns (N=[number] interviews) ### Recurring Jobs | Job | Mentioned By | Avg Importance | Avg Satisfaction | Opportunity Score | | --- | --- | --- | --- | --- | | [job] | P1, P3, P5 | 8.3 | 3.2 | 5.6 | ### Recurring Pain Points | Pain Point | Mentioned By | Severity Consensus | | --- | --- | --- | | [pain] | P1, P2, P4, P5 | High | ### Emerging Themes 1. [theme] — supported by [N] participants 2. [theme] — supported by [N] participants ### Contradictions / Segments - [Group A thinks X, Group B thinks Y — possible segmentation signal] ### Recommended Next Steps 1. [action based on patterns] 2. [action based on patterns] ``` --- ## Anti-Patterns | Avoid | Why | Instead | | --- | --- | --- | | Editorializing quotes | Loses the participant's voice | Use exact words, even if imperfect | | Confirmation bias in synthesis | Only highlighting data that supports your hypothesis | Include disconfirming evidence prominently | | Skipping "Willingness to Pay" | Leads to building things nobody will buy | Always assess priority and switching signals | | Combining interviews too early | Loses individual nuance | Summarize each individually first, then synthesize | | Ignoring what they like | Understanding satisfaction prevents breaking what works | Always capture positives alongside pains | --- ## References - Fitzpatrick, Rob. *The Mom Test* (2013) - Torres, Teresa. *Continuous Discovery Habits* (2021) - [Opportunity Score (Ulwick)](https://jobs-to-be-done.com/opportunity-score)