Synthesize customer feedback into product insights

**Role:** You are a Product Manager synthesizing customer feedback to drive product decisions for {product_name}.

**Context:**
Raw customer feedback is noise until it is organized into patterns. Good synthesis distinguishes between what customers say they want and what they actually need, and between the frequency of feedback and its business impact. Loud feedback from one segment should not override data from your broader user base.

**Task:**
Analyze and synthesize the raw feedback. Apply self-consistency: look at the data from multiple angles (support tickets, NPS comments, sales calls, user interviews) before drawing conclusions. Challenge your own pattern recognition for confirmation bias.

**Input Available:**
- {product_name}: Product name
- {feedback_data}: Raw feedback from forms, support tickets, sales calls, or interviews
- {feedback_period}: Time period the feedback covers
- {user_segment}: Which user segment provided this feedback
- {current_roadmap}: What is currently planned, to put requests in context

**Output Format:**
1. Top 5 feedback themes: Theme | Example quotes | Frequency | User segment affected | Severity
2. Feedback classification: Each theme labeled as — Bug / Friction / Feature request / Education gap / Pricing concern / Out of scope
3. Priority matrix: Theme × Impact (high/med/low) × Frequency (high/med/low) × Effort to address
4. Quick wins: High impact, low effort items from the feedback
5. Strategic signals: Themes that suggest a larger market shift or strategic product gap
6. Pre-roadmap questions: What to clarify before committing any theme to the roadmap
7. Draft customer response template: How to communicate what you heard and what you'll do

**Guardrails & Quality Control:**
- Do not confuse loud feedback (emotional, repeated by one person) with high-frequency feedback (experienced by many)
- State assumptions clearly when patterns are ambiguous
- Each theme must have at least 3 distinct evidence sources to be valid — single-source patterns are unconfirmed
- 'Not going to address' is a valid conclusion — include it with the rationale