Set up real-time sentiment monitoring prompts for ongoing customer conversations.
## Real-Time Customer Sentiment Monitor
Set up sentiment monitoring for live customer interactions:
**Alert Thresholds:** [THRESHOLDS]
**Escalation Rules:** [ESCALATION_RULES]
**Channels Monitored:** [CHANNELS]
**Sentiment Analysis Prompt:**
Analyze the following customer message for sentiment:
**Message:** [CUSTOMER_MESSAGE]
**Context:** [CONVERSATION_HISTORY]
**Output:**
```json
{
"overall_sentiment": "positive|neutral|negative|very_negative",
"confidence": 0.0-1.0,
"emotions_detected": ["frustration", "anger", "confusion", "satisfaction"],
"intensity": 1-10,
"escalation_needed": true/false,
"trigger_words": ["cancel", "lawsuit", "terrible"],
"recommended_tone": "empathetic|reassuring|celebratory|professional",
"churn_risk": "low|medium|high|critical"
}
```
**Alert Rules:**
- Sentiment drops to "very_negative" → Alert supervisor immediately
- Intensity > 8 → Flag for manual review
- Trigger words detected → Auto-escalate if policy-listed
- 3+ negative messages in sequence → Offer human takeover
- Churn risk "critical" → CSM notification
**Dashboard Elements:**
- Real-time sentiment heatmap across all active conversations
- Trending negative topics (last hour, last shift)
- Agent sentiment management scores
- Comparison: AI sentiment vs agent assessment
**Continuous Learning:**
- Log agent corrections to sentiment assessments
- Retrain monthly on false positive/negative patterns
- Industry-specific lexicon updatesFree to copy and use. Compatible with Claude 4 Opus, GPT-5, Gemini 2.5 Pro.
Define alert thresholds in [THRESHOLDS] and escalation rules.
Initial release
Sign in and download this prompt to leave a review.