---
name: product-discovery
description: Product discovery and market research expert. Use when validating product ideas, conducting market research, user interviews, competitive analysis, or opportunity assessment. Covers JTBD, Kano model, and Value Proposition Canvas.
---
# Product Discovery

## Core Principles

- **Continuous Discovery** — Weekly user conversations, not episodic research
- **Outcome-Driven** — Start with outcomes to achieve, not solutions to build
- **Assumption Testing** — Validate risky assumptions before committing resources
- **Co-Creation** — Build with customers, not just for them
- **Data-Driven** — Use evidence over intuition and stakeholder opinions
- **Problem-First** — Deeply understand the problem space before ideating solutions

---

## Hard Rules (Must Follow)

> These rules are mandatory. Violating them means the skill is not working correctly.

### No Solution-First Thinking

**Never start with a solution. Always define the problem and outcome first.**

```markdown
❌ FORBIDDEN:
"We should build a search bar for the product page"
"Let's add AI recommendations"
"Users need a mobile app"

✅ REQUIRED:
"Problem: Users can't find products (40% exit rate on catalog)
Outcome: Reduce exit rate to 20%
Possible solutions:
1. Search bar with filters
2. AI-powered recommendations
3. Better category navigation
4. Visual product browsing"
```

### Evidence-Based Decisions

**Never assume user needs without evidence from real user research.**

```markdown
❌ FORBIDDEN:
- "Users probably want X" (assumption without data)
- "Our competitor has X, so we need it too" (copycat without validation)
- "The CEO thinks we should build X" (HiPPO without evidence)
- "It's obvious users need X" (intuition without validation)

✅ REQUIRED:
- "5 out of 8 interviewed users mentioned X as a pain point"
- "Analytics show 60% of users abandon at step 3"
- "Prototype test: 7/10 users completed task successfully"
- "Survey (n=500): 45% rated feature as 'must have'"
```

### Minimum Interview Threshold

**Never validate a problem with fewer than 5 user interviews per segment.**

```markdown
❌ FORBIDDEN:
- "We talked to 2 users and they loved the idea"
- "One customer requested this feature"
- "Based on a quick chat with sales..."

✅ REQUIRED:
| Segment | Interviews | Key Finding |
|---------|------------|-------------|
| Power Users | 6 | 5/6 struggle with X |
| New Users | 5 | 4/5 drop off at onboarding |
| Churned | 5 | 3/5 cited missing feature Y |

Minimum per segment: 5 interviews
Confidence increases with more interviews
```

### Falsifiable Assumptions

**Every assumption must be testable and falsifiable with clear success criteria.**

```markdown
❌ FORBIDDEN:
- "Users will like the new design" (not falsifiable)
- "This will improve engagement" (no success criteria)
- "The feature will be useful" (vague)

✅ REQUIRED:
| Assumption | Test | Success Criteria | Result |
|------------|------|------------------|--------|
| Users will complete onboarding in new flow | Prototype test with 10 users | >70% completion | TBD |
| Users prefer visual search | A/B test | >10% lift in conversions | TBD |
| Price point is acceptable | Landing page test | >3% conversion | TBD |
```

---

## Quick Reference

### When to Use What

| Scenario | Framework/Tool | Output |
|----------|---------------|--------|
| Validate product idea | Product Opportunity Assessment | Go/no-go decision |
| Size market opportunity | TAM/SAM/SOM | Market size estimates |
| Understand user needs | User Research (interviews, surveys) | User insights, pain points |
| Analyze competition | Competitive Analysis | Competitive landscape map |
| Discover user motivations | Jobs-to-be-Done (JTBD) | Job stories, outcomes |
| Prioritize features | Kano Model | Feature categorization |
| Define value proposition | Value Proposition Canvas | Value prop statement |
| Test product concept | Lean Startup / MVP | Validated learnings |
| Map opportunities | Opportunity Solution Tree | Prioritized opportunities |

---

## Continuous Discovery Habits

### The Product Trio

Discovery is led by three roles working together weekly:

```
Product Manager → Defines outcomes, owns roadmap
Designer        → Explores solutions, tests usability
Engineer        → Assesses feasibility, proposes technical solutions
```

### Weekly Activities

```markdown
## 1. Customer Interviews (Weekly)
- Schedule 3-5 interviews per week minimum
- Mix of current users, churned users, prospects
- Focus on understanding problems, not pitching solutions
- Record and share insights with team

## 2. Assumption Testing (Weekly)
- Identify riskiest assumptions about solutions
- Design quick tests (prototypes, landing pages, fake doors)
- Run experiments with real users
- Measure results against success criteria

## 3. Opportunity Mapping (Ongoing)
- Build opportunity solution tree
- Map customer needs to potential solutions
- Prioritize based on impact and feasibility
- Update as you learn
```

### Discovery vs Delivery

```
Discovery (What to Build)          Delivery (How to Build It)
├─ Customer interviews             ├─ Sprint planning
├─ Prototype testing               ├─ Development
├─ Assumption validation           ├─ QA testing
├─ Market research                 ├─ Deployment
└─ Opportunity assessment          └─ Post-launch monitoring

Key difference: Discovery reduces risk BEFORE committing to build
```

---

## Product Opportunity Assessment

### Marty Cagan's 10 Questions

Before starting any product initiative, answer these questions:

```markdown
## 1. Problem Definition
**What problem are we solving?**
- Be specific and measurable
- Validate it's a real problem (not assumed)

## 2. Target Market
**For whom are we solving this problem?**
- Define specific user segments
- Size the addressable market (TAM/SAM/SOM)

## 3. Opportunity Size
**How big is the opportunity?**
- Revenue potential
- User growth potential
- Strategic value

## 4. Success Metrics
**How will we measure success?**
- Leading indicators (usage, engagement)
- Lagging indicators (revenue, retention)
- Define targets upfront

## 5. Alternative Solutions
**What alternatives exist today?**
- Direct competitors
- Indirect solutions
- Current user workarounds

## 6. Our Advantage
**Why are we best suited to solve this?**
- Unique capabilities
- Market position
- Technical advantages

## 7. Strategic Fit
**Why now? Why us?**
- Market timing
- Strategic alignment
- Resource availability

## 8. Dependencies
**What do we need to succeed?**
- Technical dependencies
- Partnership requirements
- Regulatory considerations

## 9. Risks
**What could go wrong?**
- Market risk (will anyone want it?)
- Execution risk (can we build it?)
- Monetization risk (will they pay?)

## 10. Cost of Delay
**What happens if we don't build this?**
- Competitive disadvantage
- Lost revenue
- Market opportunity window
```

### Value vs Effort Framework

Quick prioritization of opportunities:

```
High Value, Low Effort  → Do First (Quick Wins)
High Value, High Effort → Plan Strategically (Big Bets)
Low Value, Low Effort   → Do Later (Fill Gaps)
Low Value, High Effort  → Don't Do (Money Pit)
```

---

## Discovery Methods

### When to Use What Method

```markdown
## Generative Research (What problems exist?)
Use when: Starting new product area, exploring unknown space
Methods:
- Ethnographic field studies
- Contextual inquiry
- Diary studies
- Open-ended interviews

## Evaluative Research (Does our solution work?)
Use when: Testing specific solutions, validating designs
Methods:
- Usability testing
- Prototype testing
- A/B testing
- Concept testing

## Quantitative Research (How much? How many?)
Use when: Need statistical validation, measuring impact
Methods:
- Surveys
- Analytics analysis
- A/B experiments
- Market sizing

## Qualitative Research (Why? How?)
Use when: Understanding motivations, uncovering insights
Methods:
- User interviews
- Focus groups
- Customer advisory boards
- User observation
```

### Interview Best Practices

```markdown
## Preparation
- Define research goals and hypotheses
- Create interview guide (but stay flexible)
- Recruit right participants (6-8 per segment)
- Schedule 45-60 min sessions

## During Interview
✓ Ask open-ended questions ("Tell me about...")
✓ Follow up with "Why?" 5 times to get to root cause
✓ Listen more than talk (80/20 rule)
✓ Ask about past behavior, not future hypotheticals
✓ Look for workarounds and pain points
✓ Record and take notes

✗ Don't ask leading questions
✗ Don't pitch your solution
✗ Don't ask "Would you use X?" (people lie)
✗ Don't multi-task while interviewing

## Example Questions
- "Walk me through the last time you [did task]"
- "What's most frustrating about [current solution]?"
- "How are you solving this problem today?"
- "What would make [task] easier for you?"
- "Tell me more about that..."
```

### Survey Best Practices

```markdown
## When to Survey
✓ Validate findings from qualitative research
✓ Measure satisfaction or sentiment at scale
✓ Prioritize features (Kano surveys)
✓ Segment users by behavior/needs

## Survey Design
- Keep it short (<10 min to complete)
- One question per screen on mobile
- Mix question types (multiple choice, scale, open-ended)
- Avoid leading or biased questions
- Test survey with 5 people before sending

## Question Types
- Multiple choice → Segmentation, categorization
- Likert scale (1-5) → Satisfaction, importance
- Open-ended → Qualitative insights
- Ranking → Prioritization
- NPS (0-10) → Loyalty measurement

## Distribution
- In-app surveys (high response, biased to engaged users)
- Email surveys (broader reach, lower response)
- Incentivize thoughtful responses ($10 gift card, early access)
- Follow up with interviews for interesting responses
```

---

## 2025 Trends in Product Discovery

### AI-Powered Research

```markdown
## AI Tools for Discovery
- **Insight synthesis** — AI analyzes interview transcripts, identifies patterns
- **Synthetic personas** — AI-generated user proxies for rapid testing
- **Market intelligence** — AI tracks competitor moves, pricing changes
- **Survey analysis** — Automated sentiment analysis, theme extraction
- **Trend detection** — AI identifies emerging market trends early

## Examples
- Crayon → Competitive intelligence automation
- Glimpse → Trend detection from web data
- Delve AI → Automated persona creation
- Attest → AI-powered survey insights
- Quantilope → Machine learning research automation

## Best Practices
✓ Use AI to scale research, not replace human insight
✓ Validate AI findings with real user conversations
✓ Combine AI analysis with qualitative depth
✗ Don't rely solely on synthetic users
✗ Don't skip talking to real customers
```

### Continuous Discovery at Scale

```markdown
## Modern Approach
- Discovery is embedded in every sprint, not a phase
- Weekly user touchpoints (interviews, tests, feedback)
- Rapid experimentation (dozens of tests running)
- Fast pivots based on evidence (days, not months)

## Team Structure
- Product trios own discovery for their area
- Centralized research team supports (tools, methods)
- Customer success shares feedback loop
- Data analysts provide quantitative insights

## Cadence
- Weekly: Customer interviews, prototype tests
- Bi-weekly: Opportunity review, assumption validation
- Monthly: Market analysis, competitive review
- Quarterly: Strategic discovery (new markets, big bets)
```

---

## Opportunity Solution Tree

### What It Is

Visual framework for mapping the path from outcome to solution:

```
        OUTCOME (Business goal)
             |
    ┌────────┴────────┐
    │                 │
OPPORTUNITY 1    OPPORTUNITY 2
    │                 │
    ├─ Solution A     ├─ Solution C
    ├─ Solution B     └─ Solution D
    └─ Solution C
```

### How to Build One

```markdown
## Step 1: Define Outcome
Start with measurable business outcome
Example: "Increase Day 30 retention from 20% to 30%"

## Step 2: Map Opportunities
Discover customer needs/pain points through research
Example: "Users don't understand core features"

## Step 3: Generate Solutions
For each opportunity, brainstorm multiple solutions
Example:
- Better onboarding tutorial
- In-app tooltips
- Interactive product tour

## Step 4: Test Assumptions
For each solution, identify riskiest assumption and test
Example: "Users will complete a 5-step tutorial"
Test: Build simple prototype, test with 10 users

## Step 5: Compare Solutions
Use evidence to choose best path forward
Build what tests validate, discard what fails
```

### Benefits

```
✓ Visualizes multiple paths to outcome
✓ Prevents jumping to first solution
✓ Encourages broad exploration before narrowing
✓ Documents why decisions were made
✓ Keeps team aligned on priorities
```

---

## Integrating Discovery with Delivery

### Discovery Kanban

```markdown
## Discovery Board Columns

┌─────────────┬──────────────┬──────────────┬─────────────┐
│ OPPORTUNITIES│ ASSUMPTIONS  │  EXPERIMENTS │  VALIDATED  │
│             │              │              │             │
│ Customer    │ Riskiest     │ Running      │ Ready to    │
│ needs we've │ assumptions  │ tests        │ build       │
│ identified  │ to validate  │              │             │
└─────────────┴──────────────┴──────────────┴─────────────┘

## Flow
1. Opportunities flow from research
2. Solutions generate assumptions to test
3. Experiments validate/invalidate assumptions
4. Validated solutions enter delivery backlog
```

### Definition of Ready

Before moving from discovery to delivery:

```markdown
## Discovery Checklist
- [ ] Customer problem validated (5+ interviews)
- [ ] Solution tested with prototype (10+ users)
- [ ] Success metrics defined and measurable
- [ ] Technical feasibility confirmed by engineering
- [ ] Business case approved (revenue/retention impact)
- [ ] Design mocks completed and tested
- [ ] Open questions resolved or explicitly acknowledged
- [ ] Story broken into shippable increments
```

---

## Common Anti-Patterns

### What NOT to Do

```markdown
## ✗ Solution-First Discovery
Starting with "We should build X" then finding evidence to support it
→ Instead: Start with outcome and problem, explore multiple solutions

## ✗ Episodic Research
Doing discovery as a phase, then stopping when development starts
→ Instead: Continuous weekly discovery throughout product lifecycle

## ✗ Confirmation Bias
Only talking to users who will validate your ideas
→ Instead: Seek disconfirming evidence, talk to churned users

## ✗ Fake Validation
Asking "Would you use this?" and trusting the answer
→ Instead: Test with realistic prototypes, measure actual behavior

## ✗ Analysis Paralysis
Endless research without ever shipping
→ Instead: Define upfront what evidence is "enough" to move forward

## ✗ Building for Everyone
Trying to solve for all users at once
→ Instead: Focus on specific segment, nail it, then expand

## ✗ Ignoring Weak Signals
Dismissing early negative feedback as "just a few users"
→ Instead: Treat complaints as early warning signs, investigate
```

---

## See Also

- [reference/market-research.md](reference/market-research.md) — TAM/SAM/SOM, Porter's Five Forces
- [reference/user-research.md](reference/user-research.md) — Interview guides, survey methods, ethnography
- [reference/competitive-analysis.md](reference/competitive-analysis.md) — Competitive frameworks and analysis
- [reference/opportunity-frameworks.md](reference/opportunity-frameworks.md) — JTBD, Kano, Value Proposition Canvas
- [templates/discovery-template.md](templates/discovery-template.md) — Product discovery document template