A complete, battle-tested workflow for product teams: conduct 10 generative interviews, synthesize insights, create a PRD, break it into tasks, and load a prioritized backlog into your project management tool.
Visualize the entire workflow as a Value Stream Map. Identify bottlenecks, handoffs, and optimization opportunities.
Recruit participants, run discovery interviews, and capture rich qualitative data
What do you want to learn? What decisions will this research inform?
5-7 open-ended questions focused on understanding problems, workflows, and pain points
Plan for 10 completed interviews (some may cancel). Use User Interviews or Respondent for recruitment.
45-60 min sessions. Record with Dovetail, Great Question, or Lookback. Take notes during the session.
Most tools auto-transcribe. Review transcripts for accuracy.
Don't wait until all interviews are done to start analysis. Review transcripts and tag insights after every 2-3 interviews to identify emerging themes early.
Extract themes, pain points, opportunities, and user quotes using AI-assisted analysis
From Dovetail or Great Question, export all transcripts and tagged highlights as text or CSV
Feed transcripts to an LLM with a synthesis prompt (see below)
Validate AI-generated themes against original data. Add context and nuance.
Identify 3-5 key opportunity areas based on pain points and user needs
I conducted 10 user interviews to understand [RESEARCH OBJECTIVE].
Below are the transcripts from all sessions.
Please analyze these interviews and provide:
1. **Key Themes** (5-7 major themes that emerged)
- For each theme: description, frequency across interviews,
supporting quotes
2. **Pain Points** (Top 5 pain points mentioned)
- For each: severity, frequency, user quotes
3. **Opportunities** (3-5 actionable opportunity areas)
- Based on unmet needs, workarounds, and expressed desires
4. **User Segments** (If applicable)
- Different user types or personas that emerged
5. **Workflow Patterns**
- How users currently accomplish [TASK/GOAL]
- Tools they use and handoffs between tools
Format the output in markdown with clear sections and
include specific quotes to support each finding.
[PASTE TRANSCRIPTS HERE]How to use: Replace [RESEARCH OBJECTIVE] and [TASK/GOAL] with your specifics. Paste all transcripts. Review AI output and validate against source data.
Transform insights into a structured PRD with clear requirements and success criteria
What problem are we solving? For whom? Why now?
Key findings from research (pull from synthesis doc)
What defines success? How will we measure it?
High-level approach to solving the problem
Detailed features and capabilities (must-have vs nice-to-have)
As a [user], I want [capability] so that [benefit]
Dependencies, constraints, integration points
What we're explicitly NOT doing (prevents scope creep)
Based on the research synthesis below, help me draft a Product Requirements Document (PRD) for [FEATURE/PRODUCT NAME]. Research Synthesis: [PASTE YOUR SYNTHESIS FROM PHASE 2] Please create a PRD with the following sections: 1. Problem Statement (2-3 paragraphs) 2. User Insights (key findings from research) 3. Goals & Success Metrics (3-5 measurable goals) 4. Solution Overview (high-level approach) 5. Functional Requirements (organized as Must-Have, Should-Have, Nice-to-Have) 6. User Stories (8-12 stories in "As a...I want...so that" format) 7. Technical Considerations 8. Out of Scope (what we're NOT doing) For functional requirements, be specific and actionable. For user stories, reference actual pain points from the research. Include relevant user quotes where appropriate.
How to use: Paste your research synthesis. Review AI draft and refine with team input. Validate requirements against research findings.
Decompose user stories and requirements into granular, estimatable tasks
Start with must-have requirements first
Break each story into engineering tasks (see prompt below)
Refine tasks, identify dependencies, add missing tasks
What does "done" look like for this task?
Use story points, t-shirt sizes, or hours based on your team's practice
Break down this user story into granular engineering tasks: User Story: [PASTE USER STORY] Context: - Tech stack: [YOUR TECH STACK] - Architecture: [BRIEF ARCHITECTURE DESCRIPTION] Please provide: 1. Frontend tasks (if applicable) 2. Backend tasks (if applicable) 3. Database/infrastructure tasks 4. Testing tasks 5. Documentation tasks For each task: - Make it specific and actionable - Identify dependencies on other tasks - Suggest acceptance criteria - Flag any technical unknowns or risks Format as a numbered list with dependencies clearly marked.
How to use: Run this prompt for each user story. Review with engineers to validate and refine. This gives you a comprehensive task list.
Task Title: [Action verb] + [specific work] Example: "Build API endpoint for user profile updates" Description: [1-2 sentences explaining what and why] Acceptance Criteria: - [ ] Criterion 1 - [ ] Criterion 2 - [ ] Criterion 3 Dependencies: [Task IDs this depends on] Estimate: [Points/Hours/Size] Labels: [frontend, backend, testing, etc.]
Use frameworks to determine what to build first based on value, effort, and dependencies
Score each feature: (Reach × Impact × Confidence) / Effort
Plot tasks on 2×2 grid: High Value/Low Effort = do first
Categorize every task into one of four buckets:
RICE for data-driven teams, Value/Effort for speed, MoSCoW for clear scoping
Involve PM, engineering, and design in the exercise
Some low-priority tasks must be done first (e.g., infrastructure)
Number tasks 1, 2, 3... in priority order
Help me prioritize these tasks using the RICE framework. Context: - Total addressable users: [NUMBER] - Our key success metric: [METRIC] - Sprint capacity: [POINTS/HOURS] Tasks to prioritize: [PASTE YOUR TASK LIST] For each task, provide RICE scores: - Reach: How many users impacted (0-10) - Impact: How much it moves the metric (0.25/0.5/1/2/3) - Confidence: How certain we are (50%/80%/100%) - Effort: Story points or hours Calculate RICE score: (R × I × C) / E Output as a table sorted by RICE score descending.
Import prioritized tasks into Linear, Jira, or your PM tool of choice
Use this format for bulk import into most PM tools:
Title,Description,Priority,Estimate,Labels,Parent,Dependencies "Build user profile API endpoint","Create REST API endpoint for updating user profile data. Should accept JSON payload with user fields.",High,5,"backend,api",, "Design profile edit UI","Create Figma designs for profile editing screen with form validation",High,3,"design,frontend",, "Implement profile edit form","Build React form component based on designs",Medium,5,"frontend","Design profile edit UI","Build user profile API endpoint" "Write API integration tests","Unit and integration tests for profile API",Medium,3,"testing,backend","Build user profile API endpoint", "Update user documentation","Add profile editing to user guide",Low,2,"docs","Implement profile edit form",
Use our Value Stream Mapper to visualize this entire flow. See the handoffs, identify bottlenecks, and track cycle times for continuous improvement.
Open in Value Stream MapperYou've gone from raw user insights to a prioritized, ready-to-build backlog.
Every task traces back to user insights and validated pain points
Used AI to synthesize research, draft PRD, and break down tasks
Prioritized using frameworks, not opinions or HiPPO
1. Share PRD and backlog with your team
2. Schedule sprint planning to pull in first tasks
3. Start building! 🚀