pm-feedback — User feedback analysis
Part of the Personal Corp framework — running a one-person business through AI agents. Structure raw feedback into a decision-driving insight report. Built-in classification, sentiment, theme clustering, NPS, trend analysis, source triangulation, and persona extraction.
Inputs
| Field | Required | Notes | |---|---|---| | Feedback data | yes | Excel / CSV / pasted text / review screenshots | | Purpose | no | Product improvement / satisfaction / topic-specific (e.g. post-launch reaction); default product improvement | | Time range | no | For freshness tagging and trend analysis | | Source channels | no | Multiple channels enable triangulation |
Mode: ≤ 20 items → close-read mode (item-by-item with detailed reading); > 20 → statistical mode (auto-classify + aggregated report).
Step 1 — Pre-process data
- Drop exact duplicates
- Merge near-duplicates (similarity > 90%), record merge count
- Ultra-short items (< 5 chars, no substance like "good"/"bad") → counted separately, not in deep analysis
- If a rating column exists (1-10 or 1-5 stars) → extract for NPS
- Identify source channel (in-app feedback, app store, support ticket, social media, etc.)
Step 2 — Classification
Six-category taxonomy:
| Category | Criterion | Example | |---|---|---| | Feature request | User wants something not yet built | "I'd like batch export" | | Bug report | Existing feature behaves incorrectly | "Save button loses my data" | | Usage question | User can't find or doesn't know how | "How do I change my password?" | | UX complaint | Feature exists but experience is poor | "Loading is too slow" / "UI too cluttered" | | Positive review | Satisfaction, praise, recommendation | "Love this feature!" | | Other | Unclassifiable or off-topic | Spam, ads, noise |
When ambiguous (one item spans multiple), tag primary + secondary.
Step 3 — Sentiment analysis
| Sentiment | Signals | Calibration | |---|---|---| | Positive | Likes, praise, recommends, thanks | Pure factual praise ("works") = neutral, not positive | | Neutral | Statement of fact, question, calm suggestion | Feature requests = neutral by default unless angry | | Negative | Complaint, anger, disappointment, threats | "I wish you supported X" = neutral; "Why don't you support X yet?" = negative |
Negative-intensity grading:
- Mild: calm dissatisfaction ("not very convenient")
- Medium: explicit disappointment ("very disappointed", "bad experience")
- Severe: threats ("I'll uninstall if not fixed", "I'll file a complaint") → high-priority handling
Step 4 — Theme clustering
Apply two methods to extract core themes.
Method A — Affinity mapping:
- Split observations: decompose each feedback item into independent observation cards
- Natural cluster: group by similarity without preset labels — let themes emerge
- Name themes: label each cluster ("payment flow friction", "search results irrelevant")
- Identify hierarchy: group small clusters under larger themes (e.g. "payment friction" + "long refund cycle" → "transaction experience")
- Flag outliers: items that fit no cluster — possible early signals
Method B — Thematic coding:
- Open coding: tag each item with descriptive labels ("slow load", "crash", "hidden entry point")
- Axial coding: group descriptive labels into abstract themes ("slow load" + "crash" → "performance issues")
- Selective coding: identify core themes and their relationships
- Quantify frequency: count mentions and share per theme
Cluster output:
| Theme | Sub-theme | Mentions | Share | Representative quote | |---|---|---|---|---| | {theme 1} | {sub-a} | {N} | {X%} | "verbatim quote" |
Step 5 — NPS analysis (if rating data exists)
- NPS = % Promoters (9-10) − % Detractors (0-6)
- Industry benchmarks: SaaS avg 30-40, consumer apps avg 20-30
- 5-star → 10-pt mapping: 5★=10, 4★=8, 3★=6, 2★=4, 1★=2
Step 6 — Trend analysis (if time data exists)
MoM (or WoW) change calculation:
- Aggregate by week or month per category
- Growth rate = (current − previous) / previous × 100%
- Watch for > 30% changes — flag as "needs attention"
Inflection-point detection:
- 3+ consecutive periods in one direction → established trend
- Sudden direction reversal → trigger investigation
- Correlate with external events: releases, campaigns, competitor moves
Trend output:
- Time-series description per category
- Mark significant changes + likely cause
- Early-warning: which metrics are deteriorating, which improving
Step 7 — Triangulation
When data spans multiple channels, cross-validate to lift confidence.
Method triangulation: same problem confirmed by different methods
- e.g. theme cluster says "slow load = top pain" → check if NPS detractors' open-ended answers also concentrate on performance
Source triangulation: same finding across channels
- App-store complaints + support tickets + community chatter all cite "crash" → high confidence
- Single-channel finding → tag "single-source, needs validation"
Time triangulation: persistence of the same problem
-
3 weeks consistent → systemic
- One-off → likely transient or already fixed
Confidence tiers:
| Tier | Conditions | Tag | |---|---|---| | High | Multi-source + multi-method + persistent | Decision-ready | | Medium | 2 of the 3 dimensions support | Recommend more data before deciding | | Low | Single source or single method | Reference only, validate further |
Step 8 — Persona extraction
Identify typical user types from the feedback corpus.
Method:
- Behavior cluster: infer user types (newbie / veteran / power user / occasional)
- Need cluster: which users care about efficiency, which about experience, which about price
- Sentiment cluster: loyal advocates / silent users / vocal complainers / churn-edge
Persona template:
[Persona name]: {one-sentence description}
- Typical traits: {usage frequency, focus, behavior pattern}
- Core need: {primary concern}
- Main pain: {recurring problem}
- Feedback style: {how they express}
- Estimated share: {% of feedback corpus}
- Quote: "{verbatim}"
Cap at 3-5 personas — more loses actionability.
Step 9 — Pain-point ranking
Pain priority = Frequency × Severity × User weight × Confidence
| Dimension | Scoring | |---|---| | Frequency | High (> 10) = 3, Medium (3-10) = 2, Low (< 3) = 1 | | Severity | Critical (feature broken) = 3, Severe (blocks core flow) = 2, Mild (annoying but usable) = 1 | | User weight | Paying = 1.5, Free = 1.0 (or 1.0 if no segmentation data) | | Confidence | High (triangulated) = 1.2, Medium = 1.0, Low (single source) = 0.8 |
Sort descending; output Top 10.
Step 10 — Generate report
# User Feedback Analysis Report
**Period:** {date range}
**Total feedback:** {N} (after dedup: {M})
**Sources:** {channel list}
## 1. Classification
| Category | Count | Share | MoM change (if available) |
|---|---|---|---|
## 2. Sentiment
**Positive:** {X}% | **Neutral:** {Y}% | **Negative:** {Z}%
(Negative breakdown: mild {a} / medium {b} / severe {c})
## 3. Themes
| Theme | Sub-theme | Mentions | Share | Confidence |
|---|---|---|---|---|
## 4. NPS (if rating data)
**Score:** {n} (Promoters {X}% − Detractors {Y}%)
**Benchmark:** {above/below} industry by {Δ}
## 5. Trends (if time data)
- Significant rises: {category}, +{X}% MoM
- Significant drops: {category}, −{X}% MoM
- Inflection events: {description}
## 6. Top 10 Pain Points
| Rank | Pain | Freq | Severity | Confidence | Score | Quote | Recommendation |
|---|---|---|---|---|---|---|---|
## 7. Personas
<!-- 3-5 personas -->
## 8. Key Insights
<!-- Each insight: finding + data + confidence + meaning -->
1. {insight 1}
2. {insight 2}
3. {insight 3}
## 9. Improvement Recommendations
| Priority | Recommendation | Linked pain | Expected impact | Validation method |
|---|---|---|---|---|
## 10. Statistical Notes
- Classification confidence: {high/medium} (sample {N})
- Ambiguous classifications: {count}
- Triangulation coverage: {X%} of findings multi-source verified
- Validity: {sufficient sample / limited sample, results reference-only}
Quality bar
- Classifications grounded; ambiguous items tag confidence
- Insights backed by numbers; every insight cites a count
- Recommendations actionable to feature level
- Sample < 50 → tag "limited sample, results reference-only"
- Stats computed via code for accuracy
- Sentiment runs through calibration rules
- Theme clusters MECE (mutually exclusive, collectively exhaustive)
- Triangulation tier explicit per finding
Red lines
- No over-extrapolation — 3 of 20 items mention X ≠ "many users say X"
- Preserve verbatim — every pain point includes a representative quote for traceability
- No fabricated trends — no MoM analysis without history
- No invented personas — personas grounded in cluster results, not imagined
When input is incomplete
- < 10 items → close-read each; skip statistics (sample too small)
- No source/time info → analyze, but tag "missing source/time, recommend supplementing"; skip trend + triangulation
- Mixed languages → group by language, analyze separately
- Single-source → analyze, but tag "single source, recommend cross-channel validation"
Related skills
/pm-prioritize— feature requests from feedback → RICE-rank/pm-prd— high-frequency requests → PRDs/pm-competitive— competitor mentions in feedback → enrich competitor study/pm-metrics— cross-validate feedback trends with product metrics