Skip to content

Continuous Discovery

Continuous Discovery is Teresa Torres's framework for integrating customer research into the daily product development process through weekly touchpoints, opportunity solution trees, and assumption testing.

When to Use

Use Continuous Discovery when:

  • You want to stay connected to customers continuously, not just during research phases
  • You need to reduce risk before building by testing assumptions
  • You want a structured approach to discovery that integrates with delivery
  • You're working with a product trio (PM, Design, Engineering)

Core Concepts

Concept Description
Weekly Touchpoints Talk to customers every week
Opportunity Solution Tree Visualize path from outcome to solutions
Assumption Testing Test risky assumptions before building
Story-Based Interviews Collect stories (past behavior), not opinions
Compare and Contrast Test multiple solutions in parallel

The Continuous Discovery Flow

1. OUTCOME
   Define a clear, measurable outcome
2. OPPORTUNITIES
   Discover customer needs through weekly interviews
3. SOLUTIONS
   Generate multiple solution ideas per opportunity
4. ASSUMPTIONS
   Map assumptions by type, prioritize by risk
5. EXPERIMENTS
   Test riskiest assumptions with small experiments
6. LEARNINGS
   Update OST based on what you learn
   (repeat weekly)

Key Artifacts

Opportunity Solution Tree (OST)

Visual map from outcome to opportunities to solutions to experiments.

[Outcome: Increase activation to 60%]
         |
         +-- [Opportunity: Users don't understand value]
         |       |
         |       +-- [Solution: Interactive tutorial]
         |       |       +-- [Experiment: Prototype test]
         |       |
         |       +-- [Solution: Personalized onboarding]
         |               +-- [Experiment: A/B test]
         |
         +-- [Opportunity: Setup is too complex]
                 |
                 +-- [Solution: Wizard flow]

Discovery Snapshot

Weekly summary of discovery activities:

  • Interviews conducted
  • Stories collected
  • Opportunities discovered/updated
  • Assumptions tested
  • Key learnings
  • Decisions made

Assumption Map

Assumptions organized by type:

Type Question
Desirability Will customers want this?
Viability Will it work for the business?
Feasibility Can we build this?
Usability Can users figure it out?
Ethical Should we build this?

Prioritize: High importance + Low confidence = Test first

Using the Continuous Discovery Profile

Initialize a Project

multispec init my-feature --profile continuous-discovery

Create an Opportunity Solution Tree

multispec draft ost -p my-feature

Create Weekly Discovery Snapshot

multispec draft discovery-snapshot -p my-feature

Create Assumption Map

multispec draft assumption-map -p my-feature

Evaluate Discovery Snapshot

multispec eval discovery-snapshot -p my-feature

Story-Based Interviews

In Continuous Discovery, we collect stories, not opinions.

Bad (Opinion) Good (Story)
"What do you want?" "Tell me about the last time you..."
"Would you use this?" "What happened when you tried to..."
"Is this important to you?" "Walk me through how you..."

Story Structure

Element Question
Situation What was the context?
Behavior What did you do?
Outcome What happened?
Emotions How did you feel?
Pain Points What was frustrating?
Workaround How did you work around it?

Assumption Testing

Risk Matrix

                    CONFIDENCE
                Low         High
           ┌───────────┬───────────┐
     High  │  TEST     │  MONITOR  │
IMPORTANCE │  FIRST    │           │
           ├───────────┼───────────┤
     Low   │  TEST     │  SKIP     │
           │  LATER    │           │
           └───────────┴───────────┘

Test Methods

Method Best For Duration
Prototype test Usability 1 week
Fake door Desirability 1-2 weeks
Survey Quantifying 1 week
Interview Exploring Ongoing
Data analysis Validation 1 week
A/B test Comparing 2+ weeks

Example Workflow

# 1. Initialize project
multispec init user-activation --profile continuous-discovery

# 2. Create OST with outcome
multispec draft ost -p user-activation
# Set outcome: "Increase user activation rate to 60%"

# 3. Week 1: Start discovery
multispec draft discovery-snapshot -p user-activation
# Conduct interviews, capture stories, identify opportunities

# 4. Update OST with opportunities
# Add opportunities discovered from interviews

# 5. Generate solutions for top opportunity
# Add 3+ solution ideas to the OST

# 6. Map assumptions for top solution
multispec draft assumption-map -p user-activation

# 7. Test riskiest assumption
# Run small experiment, capture results

# 8. Week 2: Continue discovery
multispec draft discovery-snapshot -p user-activation
# More interviews, more learning, update OST

# 9. Evaluate discovery cadence
multispec eval discovery-snapshot -p user-activation
# Ensure weekly touchpoints are happening

# 10. Once validated, synthesize PRD
multispec synthesize prd -p user-activation

Rubric Categories

Discovery Snapshot Evaluation

Category Weight Description
Weekly Touchpoints 25% At least one interview per week
Story Capture 25% Stories with situation-behavior-outcome
Opportunity Tracking 20% Opportunities discovered and prioritized
Assumption Testing 20% Tests running with clear hypotheses
Decision Making 10% Evidence-driven decisions

Assumption Map Evaluation

Category Weight Description
Assumption Coverage 25% All types considered
Risk Assessment 25% Importance and confidence rated
Prioritization 20% Riskiest assumptions identified
Test Planning 20% Tests designed with success criteria
Document Quality 10% Clear and actionable

Principles

  1. Weekly Touchpoints - Talk to customers every week
  2. Outcome-Driven - Start with a clear, measurable outcome
  3. Opportunity Mapping - Map customer needs, pains, desires
  4. Solution Trees - Visualize the path from outcome to solutions
  5. Assumption Testing - Test assumptions before building
  6. Small Experiments - Run small, fast experiments
  7. Story-Based Interviews - Collect stories, not opinions
  8. Collaborative Discovery - Product trio works together
  9. Compare and Contrast - Test multiple solutions in parallel
  10. Iterate on Learning - Let learnings drive next experiment

References