Telecom Prompt

Analyze the following telecom customer service call and extract:

1. Classification of the issue: Billing, Connectivity, Customer Service Issue, General Inquiry, Cancellation.  
     - If the classification is “Cancellation”, then identify the core cancellation reason (e.g., Price – Too Expensive, Price – Better Offer, Not Using the Service, Customer Service Experience).

2. Core issue reported by the customer.  
3. Specific resolution steps taken by the agent (e.g., diagnostics, refunds, technician scheduling) in only a few words.  
4. Degree of customer satisfaction at the end (choose: Satisfied, Dissatisfied, Angry, Frustrated, Neutral).  
5. Follow‑up offered or promised: (Yes/No, with brief details in only a few words).  
6. List any personally identifiable information (PII) mentioned (e.g., phone number, email, account ID, customer name).  
7. Summarize the emotional tone of the customer throughout the call using natural language. Use phrases like:  
   - "Calm and cooperative throughout"  
   - "Initially upset, became satisfied after support"  
   - "Started cordial but became increasingly frustrated"  
8. Churn Risk Level: {churn_risk_level}  
   - If your churn model level is provided, echo it exactly.  
   - If the placeholder is empty or “None”, estimate the risk yourself using behavioral cues and justify briefly.

9. Next Best Offer: {next_best_offer}  
   - If your NBO model recommendation is provided, echo it exactly.  
   - If the placeholder is empty or “None”, choose the best offer based on predicted retention uplift.

10. If churn risk is High, recommend a specific retention tactic the agent or supervisor could apply (e.g., account credit, follow‑up call, escalation to loyalty team).  
11. Actionable insights for a supervisor to follow‑up with the customer and de‑escalate the situation (this should be based on the retention tactic).  
12. **Suggested script snippet for follow‑up call**: Provide one or two concise agent lines the supervisor can read verbatim, for example:  
   - "Hello [Customer Name], this is [Agent] from Spectrum. I’m really sorry you ran into [core issue] on your account. I’ve gone ahead and [resolution action], and I’ve added your {next_best_offer} to make this right. I want to make sure you’re completely satisfied—what else can I do for you today?”
12. Suggested script snippet for follow‑up call:
   - "Hi {customer_name}, this is {agent_name} from Spectrum. I’m truly sorry you’ve had to deal with {core_issue}—that must have been really {emotional_tone}. I’ve just {resolution_action} and added your {next_best_offer} to your account so you can get back on track. How does that sound? Is there anything else I can do right now to make this easier for you?"