# Analyze Feedback

> Synthesize customer feedback from multiple sources and surface actionable themes correlated with product usage data.

- Skill: `rpatel05/analyze-feedback` (Agent Skill)
- Install (CLI): `npx skillmds@latest add rpatel05/analyze-feedback`
- Raw SKILL.md: https://api.skillmd.com/api/skills/rpatel05/analyze-feedback/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: rpatel05 (https://skillmd.com/u/rpatel05)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/rpatel05/analyze-feedback

---


# analyze-feedback

> **⚠️ Data Requirement**: This skill requires external customer feedback data (support tickets, NPS responses, user interviews, feature requests). The Mixpanel MCP server does NOT directly access support systems like Zendesk, Intercom, ProductBoard, or Canny.
>
> **How to provide feedback data:**
> - Export CSV from your support system and paste the content
> - Share a text file with feedback entries
> - Copy/paste feedback directly into the conversation
> - Provide a summary of feedback themes you've already identified
>
> The skill will help categorize, prioritize, and correlate this feedback with Mixpanel usage data.

**When to use this skill**: When you have customer feedback data and want to understand patterns and correlate them with product behavior in Mixpanel.

---

## Instructions

### Overview

This skill helps you analyze customer feedback (support tickets, NPS comments, user interviews, feature requests) and correlate patterns with Mixpanel usage data to surface actionable insights.

### Step-by-step workflow

1. **Gather feedback sources**
   - Ask the user where feedback lives:
     - Support tickets (Zendesk, Intercom)
     - NPS/survey responses
     - User interviews or call notes
     - Feature request boards (ProductBoard, Canny)
   - If feedback is in external systems, ask user to provide data or connect via MCP

2. **Categorize and theme**
   - Read through feedback and identify recurring themes
   - Common categories:
     - **Bugs/Issues**: Technical problems users encountered
     - **Feature Requests**: Missing capabilities or improvements
     - **Confusion/UX**: Areas where users got stuck
     - **Performance**: Speed, reliability concerns
     - **Pricing/Value**: Cost objections or upgrade friction
   - Count frequency of each theme

3. **Correlate with usage data**
   - For each major theme, check Mixpanel to validate or quantify:
     - If users report "X is confusing" → check funnel drop-off at step X
     - If users request feature Y → check how often they hit the gap (error events, workaround patterns)
     - If users complain about speed → check event latency or load time properties
   - Use Mixpanel queries to measure:
     - How many users are affected?
     - Is it getting worse over time?
     - Which segments mention it most?

4. **Prioritize themes**
   - Rank themes by:
     - **Volume**: How many users mentioned it?
     - **Impact**: Does data show it's blocking conversions or causing churn?
     - **Trend**: Is it increasing?
   - High priority = High volume + High impact + Increasing

5. **Generate actionable output**
   - Create a structured summary:

   **Top Themes** (ranked by priority)
   | Theme | Mentions | User Impact | Data Signal | Priority |
   |-------|----------|-------------|-------------|----------|
   | [Theme 1] | [count] | [% affected] | [metric] | High |

   **Recommendations**
   - For each high-priority theme, suggest:
     - What to fix or build
     - Expected impact (use data to estimate)
     - How to measure success

   **Supporting Quotes** (2-3 per theme)
   - Include verbatim customer quotes to add context

### Example interaction

**User**: "Analyze our support tickets from last month and tell me what's blocking users"

**Assistant**:
1. Asks: "Where are the support tickets stored? Can you share a CSV or connect me to Zendesk via MCP?"
2. User provides data
3. Reads through 150 tickets, identifies 5 themes:
   - Login issues (40 mentions)
   - Missing CSV export (35 mentions)
   - Confusion about pricing (25 mentions)
   - Mobile app crashes (30 mentions)
   - Slow dashboard load (20 mentions)
4. Queries Mixpanel:
   - Login flow: 8% funnel drop-off at "verify email" step (increasing last 30 days)
   - Export action: only 2% of paid users have export feature enabled (low discoverability)
   - Pricing page: 60% bounce rate (no engagement with price calculator)
   - Mobile sessions: 3% have error events (up from 1% last month)
   - Dashboard load: p95 load time = 8 seconds (target is <3s)
5. Outputs ranked table with recommendations:
   - **High**: Fix mobile crashes (3% of users, growing fast)
   - **High**: Improve export discoverability (paid users want it but can't find it)
   - **Medium**: Optimize dashboard performance (affects power users)
   - **Medium**: Clarify pricing page (60% bounce suggests confusion)
   - **Low**: Login flow (8% drop-off is within normal range)

### Common feedback analysis patterns

**Pattern 1: Feature gap**
- Symptom: Many users request the same capability
- Data signal: High workaround usage (users doing X manually, or using external tool)
- Action: Validate demand with usage data, estimate TAM, prioritize in roadmap

**Pattern 2: Usability issue**
- Symptom: Users say "I couldn't figure out how to..."
- Data signal: Funnel drop-off, low feature adoption, high time-to-value
- Action: Run UX audit, add onboarding, improve discoverability

**Pattern 3: Technical issue**
- Symptom: Bug reports, error complaints
- Data signal: Error events spiking, sessions with crashes
- Action: Prioritize by user impact (% affected × severity)

**Pattern 4: Value perception gap**
- Symptom: "Too expensive", "Not worth it"
- Data signal: Low engagement with paid features, high free-tier retention
- Action: Improve onboarding to paid features, adjust pricing tiers

### Success criteria

- Feedback is categorized into 3-7 clear themes
- Each theme is correlated with Mixpanel usage data
- Output includes:
  - Ranked priority list
  - Quantified user impact
  - Specific recommendations
  - Supporting customer quotes
- User can take immediate action (file bugs, adjust roadmap, run experiments)

### Tips for effective analysis

- **Don't just count**: "50 users mentioned X" is less useful than "50 users mentioned X, and data shows 20% of trials abandon at that step"
- **Look for trends**: Is feedback increasing? That's more urgent than static complaints
- **Segment feedback**: Do churned users complain about different things than retained users?
- **Validate with data**: Feedback is subjective. Use Mixpanel to measure actual behavior
- **Close the loop**: After implementing fixes, re-analyze to confirm the theme decreased