Compare Periods
npx machina-cli add skill jackhendon/ecom-feedback-intelligence/compare-periods --openclawSkill: compare-periods
Reads memory/history.json and compares trends across stored snapshots. No new classification — pure analysis of historical data.
Usage
/compare-periods
/compare-periods 4 # compare last 4 snapshots
/compare-periods all # compare all snapshots in history
Default: compare last 4 snapshots (or all if fewer than 4 exist).
Steps
Step 1: Load history
Read memory/history.json. Extract the snapshots array.
If fewer than 2 snapshots exist:
Not enough history to compare periods.
Snapshots found: N
Run /analyze-reviews at least twice to build history.
And stop.
Step 2: Select snapshots
Apply the count argument (default 4). Use the most recent N snapshots, ordered chronologically.
Step 3: Compute trends
For each metric, show direction of change across the selected window:
Sentiment trend:
- Positive %: is it rising, falling, or flat (< 2pp change = flat)?
- Negative %: same
- Net sentiment score: positive% - negative% per period
Theme trends:
- Rank each theme by frequency per period
- Identify: themes rising in rank, themes falling in rank, themes newly appearing, themes disappearing
- Flag any theme that has increased by > 5 mentions period-over-period as "accelerating"
Priority trends:
- Average priority score per period
- Count of high-priority issues (≥ 6.0) per period
- Any themes consistently generating high-priority issues
Review volume:
- Review count per period (context: is volume increasing/decreasing?)
Step 4: Output trend report
Compute all values directly from the JSON data — no estimation.
TREND ANALYSIS — [earliest period] to [latest period]
N snapshots | N total reviews
SENTIMENT TREND
Period | Positive | Neutral | Negative | Net
──────────────────────────────────────────────
YYYY-WNN | N% | N% | N% | ±N
THEME MOVEMENT
Rising: [theme] (↑ N), ...
Falling: [theme] (↓ N), ...
Stable: [themes with < 2 mention change]
Accelerating: [any theme up > 5 mentions period-over-period]
PRIORITY TRENDS
Avg score: N → N → N
High-priority count: N → N → N
KEY SIGNALS
• [signal 1]
• [signal 2]
• [signal 3]
Step 5: PM recommendation
RECOMMENDED ACTION
[One specific, evidence-based recommendation derived from the trend data]
Source
git clone https://github.com/jackhendon/ecom-feedback-intelligence/blob/main/.claude/skills/compare-periods/SKILL.mdView on GitHub Overview
Compares stored snapshots from memory/history.json to reveal sentiment shifts, theme movement, and priority changes over time. It performs a pure historical analysis without introducing new classifications. By default, it analyzes the most recent four snapshots, or all if fewer exist.
How This Skill Works
Loads history.json, selects the last N snapshots (default 4 unless you pass all), and computes trend metrics for sentiment, themes, priorities, and review volume. It then outputs a TREND ANALYSIS report with exact values from the data and a PM recommendation. No estimation or reclassification is performed.
When to Use It
- Spot sentiment shifts across time in customer reviews.
- Compare how themes rise or fall in frequency across periods.
- Detect accelerating themes with more than five mentions between periods.
- Track changes in high-priority issues per period.
- Obtain a data-driven PM recommendation based on historical trends.
Quick Start
- Step 1: Run /compare-periods with the desired window (default 4) or /compare-periods all.
- Step 2: Review the TREND ANALYSIS and note sentiment, theme movement, and priority trends.
- Step 3: Read the PM recommendation and plan data-driven actions.
Best Practices
- Ensure memory/history.json exists and contains a valid snapshots array.
- Run with the default 4 snapshots or specify all to include every period.
- Interpret the TREND ANALYSIS output using the raw data, not estimates.
- Watch for accelerating themes (more than 5 mentions) as early signals.
- Cross-check the PM recommendation with current business context before acting.
Example Use Cases
- Sentiment improves from 52% positive to 64% positive over four snapshots.
- Shipping/theme climbs in rank and shows accelerating mentions across periods.
- High-priority issues rise from 2 to 7 per period, centering on damaged items.
- Review volume trends upward, suggesting growing customer engagement.
- New theme eco-friendly packaging appears in the latest snapshot.
