Feedback-Driven Development

Overview

Build what users actually need by systematically collecting, analyzing, and acting on feedback. Transform user insights into product improvements through structured processes.

Core principle: Users tell you what's broken. Your job is to find the pattern and fix the root cause.

When to Use

Use when:

• Planning product roadmap
• Deciding what to build next
• Users reporting issues or requests
• Post-launch evaluation
• Validating assumptions
• Prioritizing features

The Feedback Loop

Collect → Synthesize → Prioritize → Build → Measure → Repeat

1. Collect Feedback (Multi-Channel)

Sources:

• In-app feedback widgets
• App store reviews
• Support tickets
• User interviews
• Social media mentions
• Analytics (behavior is feedback)
• Surveys (targeted questions)

Collection agents:

• feedback-synthesizer (analyze collected feedback)
• support-responder (support ticket patterns)
• analytics-reporter (behavioral data)

2. Synthesize Patterns

Use feedback-synthesizer agent:

@feedback-synthesizer analyze feedback from:
- App store reviews (last 30 days)
- Support tickets (last 30 days)
- In-app feedback submissions

Identify:
- Top 5 pain points
- Feature requests by frequency
- Bug reports by severity
- Sentiment trends

Look for:

• Repeated complaints (patterns)
• Surprising requests (blind spots)
• Emotional language (strong feelings)
• Churned users' reasons

3. Categorize Feedback

Categories:

• Bugs: Something broken
• Feature requests: New functionality
• UX issues: Confusing or frustrating
• Performance: Slow or laggy
• Content: Missing info or unclear
• Noise: Not actionable

Priority levels:

• P0 - Critical: Blocking users, losing revenue
• P1 - High: Frequent pain point, affects many users
• P2 - Medium: Nice to have, affects some users
• P3 - Low: Edge cases, minor improvements

4. Prioritize Actions

Use sprint-prioritizer agent:

@sprint-prioritizer help prioritize these items:

Feedback themes from feedback-synthesizer:
1. [Theme 1]: Affects X% of users, severity [level]
2. [Theme 2]: Affects Y% of users, severity [level]

Constraints:
- 6-day sprint cycle
- Team size: [X developers]
- Current priorities: [list]

Recommend: Top 3 items for next sprint

Prioritization framework (RICE):

Score = (Reach × Impact × Confidence) / Effort

Reach: How many users affected (per quarter)
Impact: How much it helps (0.25 = minimal, 3 = massive)
Confidence: How sure are you (0-100%)
Effort: Person-weeks required

5. Validate Before Building

Don't build every request blindly:

Validation questions:

• Is this the real problem or a symptom?
• How many users actually need this?
• What's the underlying job-to-be-done?
• Can we solve it differently?
• What's the minimum we can build to test?

Example:

User request: "Add calendar integration"

Underlying need: "I forget to use the app"

Better solution: Push notifications
Faster to build, solves actual problem

6. Build & Ship

Build with rapid-prototyping:

• Start with MVP of feedback-driven feature
• Ship to subset of users first
• Gather feedback on the solution
• Iterate quickly

7. Close the Loop

Follow up with users:

• Announce fixes/features to those who requested
• "Thanks for your feedback on [X]. We shipped [solution]!"
• Track if it actually solved their problem

User Interview Techniques

When to Interview

Interview for:

• Understanding "why" behind feedback
• Discovering unspoken needs
• Validating new feature ideas
• Understanding churned users

Don't interview for:

• Validation of what you want to build (confirmation bias)
• Getting feature ideas (users aren't product managers)
• Detailed UX feedback (use usability testing)

Interview Script Pattern

Warm-up (2 min):

• "Thanks for your time"
• "Tell me about yourself and how you use [product]"

Problem exploration (10 min):

• "What's frustrating about [current solution]?"
• "Walk me through the last time you [did task]"
• "What workarounds have you tried?"
• "If you had a magic wand, what would you change?"

Feature validation (5 min):

• "We're considering [feature]. What do you think?"
• "How would this fit into your workflow?"
• "What concerns do you have?"

Wrap-up (3 min):

• "Anything else you'd like us to know?"
• "Can we follow up if we have questions?"

Total: 20 minutes

Interview Insights

Listen for:

• Emotional reactions (strong feelings = important)
• Workarounds (signals missing functionality)
• Frequency words ("every time", "always")
• Jobs-to-be-done (underlying goals)

Red flags:

• "It would be cool if..." (nice-to-have)
• "You should add..." (feature suggestion without pain)
• "I don't know, just better" (vague)

Feedback Metrics to Track

Track systematically:

• Feedback volume (trending up/down?)
• Sentiment (positive/negative/neutral %)
• Top categories (bugs, features, UX)
• Response time (how fast you address)
• Resolution rate (% of feedback acted on)

Dashboard view:

This month:
- 234 feedback items
- Sentiment: 65% positive, 25% neutral, 10% negative
- Top category: Performance (34%)
- Avg response: 2.3 days
- Resolution rate: 78%

Trends:
- Negative feedback down 15% (performance fixes working)
- Feature requests up 40% (users more engaged)

Integration with Development Cycle

Sprint Planning Integration

Every sprint:

1. @feedback-synthesizer analyze last sprint's feedback
2. @sprint-prioritizer integrate feedback into sprint planning
3. Select 1-2 feedback-driven items for sprint
4. Build and ship
5. Monitor feedback on changes

Feature Validation Pattern

Before building large feature:

1. Collect requests for feature (how many users?)
2. Interview 5-10 users about the need
3. Build lightweight MVP or fake door test
4. Ship to small group
5. Measure usage and feedback
6. Decide: expand, pivot, or kill

Feedback Response Patterns

Responding to Feature Requests

Template:

"Thanks for the suggestion!

We're tracking this idea - [brief acknowledge their point].

Quick question to help us prioritize: [clarifying question about their use case]

We'll keep you posted if this makes it into development."

For popular requests:

"Great timing - this is something we're actively working on!

Expected timeline: [timeframe]
We'll email you when it ships.

Want to beta test it first?"

Responding to Bug Reports

Template:

"Thanks for reporting this!

Reproduced the issue: [brief description]
Priority: [P0/P1/P2]
Fix ETA: [timeline]

We'll follow up when this is resolved.

Temporary workaround: [if available]"

Responding to Negative Feedback

Template:

"Sorry you're having a frustrating experience.

I understand [restate their issue] is impacting your [workflow/experience].

We're [what you're doing about it]:
- [Immediate action]
- [Longer-term fix]

Can we hop on a quick call to understand better? [email/calendly]"

Prioritization Frameworks

Impact × Effort Matrix

High Impact, Low Effort → DO NOW (quick wins)
High Impact, High Effort → PLAN (strategic bets)
Low Impact, Low Effort → LATER (when time permits)
Low Impact, High Effort → NEVER (waste of resources)

Kano Model

Feature categories:

• Must-have: Expected, absence causes dissatisfaction
• Performance: More is better (speed, reliability)
• Delighters: Unexpected, cause satisfaction

Apply:

• Must-haves: Fix immediately
• Performance: Continuous improvement
• Delighters: Differentiation opportunities

Feedback Quality Indicators

High-quality feedback:

• Specific problem description
• Context (when, where, how)
• Impact statement
• Frequency indication

Low-quality feedback:

• Vague complaints
• Feature lists without reasoning
• "Just make it better"
• One-off edge cases

Seek clarification for low-quality feedback

Resources

• feedback-synthesizer agent
• sprint-prioritizer agent
• analytics-reporter agent
• User interviewing: https://www.nngroup.com/articles/user-interviews/

Feedback is a gift. Collect it systematically, analyze it rigorously, act on it strategically.