Post-Sale Verification Agent

After a customer completes a transaction — placed an order, signed a contract, booked an appointment — call them within 24h to verify the details, confirm contact preferences for fulfillment, and capture a quick CSAT signal. This pattern works for e-commerce, healthcare bookings, real estate sales, and field service.

What it does

1. Greets by name, references the specific transaction (order ID, contract number, appointment date)
2. Confirms key details: delivery address, contact preference, scheduled date
3. Asks if they have any questions before fulfillment
4. Captures a 1–5 CSAT and any free-form feedback
5. Triggers escalation tool if CSAT ≤ 2 — routes to a human follow-up

Suggested configuration

Code
agent = client.agents.create(
    name="Post-Sale Verifier",
    brain={
        "provider": "phantom",
        "model": "phantom",
        "instructions": VERIFY_PROMPT,
        "temperature": 0.3,  # consistent, less improvising
    },
    voice={
        "provider": "elevenlabs",  # warmer for the post-sale moment
        "model": "eleven_flash_v2_5",
        "voice_id": "<your_brand_voice_id>",
    },
    transcriber={"provider": "deepgram", "model": "nova-3", "language": "en-US"},
    interruption={"enabled": True},
    timeout={"enabled": True, "trigger_seconds": 10},
    call_duration={"enabled": True, "max_seconds": 240},  # 4-min cap
    analytics={
        "recording": {"enabled": True},
        "summary": {"enabled": True},
        "success_evaluation": {
            "enabled": True,
            "rubric_type": "numeric_scale",
            "criteria": "Rate 1–5: did the agent verify the order, capture CSAT, and escalate appropriately if needed?",
        },
        "data_extraction": {
            "enabled": True,
            "fields": [
                {"name": "details_confirmed", "type": "boolean"},
                {"name": "csat_score", "type": "number"},
                {"name": "feedback", "type": "string"},
                {"name": "escalation_triggered", "type": "boolean"},
            ],
        },
    },
    transfer={
        "enabled": True,
        "mode": "warm",
        "destinations": [{
            "name": "Customer Success",
            "number": "+1...",
            "description": "Escalate dissatisfied customers",
            "keywords": ["unhappy", "wrong", "complaint", "manager"],
        }],
    },
)

Prompt template

Code
You are a customer-care representative for {{company_name}}.
You're calling {{first_name}} to verify their recent {{transaction_type}}
(reference: {{transaction_ref}}) placed on {{transaction_date}}.

GOAL: confirm details are correct, ensure they have what they need ahead of
{{fulfillment_event}} (e.g. "delivery", "appointment", "move-in"), and capture
a CSAT score.

CALL FLOW:
1. Greet by name. Confirm you're speaking with {{first_name}}.
2. State the reason for the call: "I'm following up on your {{transaction_type}}
   from {{transaction_date}}, just to make sure everything looks right."
3. Verify the key details:
   - {{verification_item_1}} (e.g. delivery address)
   - {{verification_item_2}} (e.g. preferred contact method)
   - {{verification_item_3}} (e.g. scheduled date)
   For each: read it back and ask "is that correct?"
4. Open the floor: "Any questions before we proceed?"
5. CSAT: "On a scale of 1 to 5, how would you rate your experience so far?"
6. If CSAT ≤ 2: "I'd like to connect you with a customer success specialist
   to make this right — can I transfer you now?" — call transfer tool.
7. Wrap up: "Thanks for choosing {{company_name}} — have a great day."

GUIDELINES:
- Keep it under 4 minutes.
- If they're busy, offer to call back: "I just need 2 minutes — is now okay?"
- If they say a detail is wrong, capture the correction in your notes and assure
  them you'll update the record. Don't promise specific timelines you can't keep.
- Never argue with a low CSAT. Acknowledge, escalate, end the call.

Variables to set per call

Code
client.voice_sessions.outbound(
    agent_id=agent["id"],
    to_number=customer["phone"],
    number_id="num_...",
    dynamic_variables={
        "company_name": "Your Company",
        "first_name": customer["first_name"],
        "transaction_type": "order",
        "transaction_ref": "ORD-12345",
        "transaction_date": "May 3rd",
        "fulfillment_event": "delivery",
        "verification_item_1": f"shipping to {customer['address']}",
        "verification_item_2": f"contact via {customer['contact_pref']}",
        "verification_item_3": f"delivery scheduled for {customer['delivery_date']}",
    },
)

Integration tips

• CSAT pipeline: post-call webhook delivers extracted_data.csat_score — pipe into your analytics stack.
• Escalation routing: the warm-transfer destination should be a real number reachable in business hours; outside hours, swap for a callback-request tool.
• Spam-call avoidance: dial within 24h of the transaction while context is fresh; after 72h customers often don't remember and answer rates drop.

Related

• Call transfer — warm vs blind transfer
• Post-call analysis — extracting structured CSAT
• Lead qualification — companion outbound pattern