Rate Message - Learn What Works

You track message outcomes and update the Feedback Insights page so future drafts improve over time.

Input Format

User will say something like:

• "/rate-message Sarah replied: 'thanks for reaching out'"
• "Sarah got back to me, she wants to chat"
• "No reply from Sam after a week"
• "Mark Jane's message as no reply"

Step 1: Load Configuration

Read .claude/config/workspace.json to get:

• Messages database collection URL
• Feedback Insights page ID

If file doesn't exist: Tell user to run /setup-crm first.

Step 2: Find the Message

Search Messages database for the lead using mcpnotionnotion-search:

query: [lead_name]
data_source_url: [messages collection URL from config]

Find the most recent message for this lead with Result = "Pending".

If not found: Ask if they logged the message with /update-crm first.

Step 3: Update Message Result

Use mcpnotionnotion-update-page to update the message:

{
  "page_id": "[message_page_id]",
  "command": "update_properties",
  "properties": {
    "Result": "[Reply / No Reply / Meeting]"
  }
}

Also fetch the full message content for analysis.

Step 4: Update Lead Status

Find and update the lead's status based on outcome:

Use mcpnotionnotion-update-page on the lead.

Step 5: Analyze the Message

If the message got a reply, analyze what worked:

Element Analysis

Review the message against the 5-element framework:

• Which elements were present?
• Which were strongest?
• Any patterns that stand out?

Pattern Detection

Look for:

• Hook type: What opened the message?
• Weakness style: How explicit was the vulnerability?
• Pedestal specificity: How personalized?
• Ask type: Binary vs open-ended?

If No Reply

Note potential issues:

• Missing elements?
• Generic language?
• Too long?
• Wrong timing?

Step 6: Update Feedback Insights (CRITICAL)

Fetch the current Feedback Insights page, then update it with new learnings.

For Replies (What's Working)

Add to "What's Working" section:

- [Pattern]: "[Example from message]" → Reply ([Date])

Example:

- Binary ask + explicit weakness: "is it X or Y? honestly stuck on..." → Reply (2024-01-15)

For No Replies (What's Failing)

Add to "What's Failing" section if you identify an anti-pattern:

- [Anti-pattern]: "[Example]" → No Reply ([Date])

Update Stats

Update the Stats section:

| Metric | Value |
|--------|-------|
| Total Messages | [+1] |
| Reply Rate | [recalculate] |
| Avg Rating | [recalculate] |

Use mcpnotionnotion-update-page with replacecontentrange to update specific sections.

Step 7: Report Results

Logged: [Reply/No Reply] from [Name]

Message Rating: [X]/10
Lead Status: [Old] → [New]

Analysis:
[What likely worked / What might have failed]

Pattern noted:
[Key learning added to Feedback Insights]

Feedback Insights updated: [Notion URL]

---

Stats:
• Total Messages: [N]
• Reply Rate: [X]%
• This message: [Above/Below] average

Batch Analysis

If user says "analyze my last 5 messages" or "what's working?":

1. Fetch recent messages from Messages database
2. Filter to those with outcomes (Reply or No Reply)
3. Analyze patterns across the batch:

- Reply rate by element strength - Common patterns in successful messages - Common issues in failed messages

1. Update Feedback Insights with aggregated learnings
2. Present summary with actionable recommendations

Example Flow

User: Sarah replied: "hey thanks for reaching out, actually yeah we've been thinking about this"

Skill Response:

Logged: Reply from Sarah Chen

Message Rating: 8/10
Lead Status: Contacted → Replied

Analysis:
✓ Strong weakness element ("honestly stuck on...")
✓ Binary ask made it easy to respond
✓ Specific pedestal (referenced her Series A)
→ Her response confirms the pain point is real

Pattern noted:
Added to "What's Working":
- Binary ask + Series A reference → Reply (2024-01-15)

Feedback Insights updated: [Notion URL]

---

Stats:
• Total Messages: 12
• Reply Rate: 33% (4/12)
• This message: Above average (rated 8/10, avg 6.5)

Next: Would you like to book a meeting? Update with:
/rate-message Sarah meeting booked

Error Handling