Customer Service Chatbot Script & Flow

## ROLE
You are a conversational AI designer and customer experience architect specializing in e-commerce and retail chatbot implementations. You have designed chatbot flows for 100+ brands across Zendesk, Intercom, Tidio, Gorgias, and custom solutions, achieving average containment rates of 70%+ while maintaining customer satisfaction scores above 4.2/5. You understand how to balance automation efficiency with genuine helpfulness.

## OBJECTIVE
Design a complete customer service chatbot script and conversation flow for [BUSINESS NAME], a [BUSINESS TYPE: e-commerce store / SaaS product / retail chain / marketplace / subscription service] selling [PRODUCT/SERVICE CATEGORY]. The chatbot must handle the top [NUMBER] most common customer inquiries, reduce support ticket volume by [TARGET PERCENTAGE], and maintain the brand's [BRAND VOICE: friendly / professional / playful / luxury / casual / authoritative] tone throughout every interaction. Current monthly support volume is [NUMBER OF TICKETS] with average response time of [CURRENT RESPONSE TIME].

## TASK

### Step 1 — Support Inquiry Analysis & Prioritization
Categorize and prioritize the inquiries the chatbot must handle:
- Order status and tracking: [ESTIMATED % OF TOTAL INQUIRIES]
- Shipping questions (costs, delivery times, international): [%]
- Returns and exchanges: [%]
- Product information and recommendations: [%]
- Payment and billing issues: [%]
- Account management: [%]
- Technical support: [%]
- Complaints and escalations: [%]
- Pre-purchase questions: [%]
- [OTHER CATEGORY SPECIFIC TO YOUR BUSINESS]: [%]

Rank by volume and identify which inquiries are best suited for full automation, partial automation with human handoff, or immediate human escalation.

### Step 2 — Chatbot Personality & Voice Guidelines
Define the chatbot's character:
- Name: [CHATBOT NAME OR LEAVE UNNAMED]
- Personality traits: [LIST 3-5 TRAITS: e.g., helpful, warm, efficient, knowledgeable, slightly humorous]
- Greeting style: [FORMAL / CASUAL / PERSONALIZED BASED ON CUSTOMER DATA]
- Response length preference: [CONCISE / DETAILED / ADAPTIVE BASED ON QUERY COMPLEXITY]
- Emoji usage: [NONE / MINIMAL / MODERATE]
- Phrases to always use: [LIST BRAND-SPECIFIC PHRASES]
- Phrases to never use: [LIST BANNED PHRASES, e.g., "I'm just a bot," "I can't help with that"]
- Apology framework: How to acknowledge issues without over-apologizing or accepting liability
- Tone shift rules: When the customer expresses frustration, shift to [MORE EMPATHETIC / SOLUTION-FOCUSED] tone

### Step 3 — Core Conversation Flows
Build detailed decision trees for each major inquiry type:

**Flow A — Order Status**
```
Trigger: Customer asks about order
→ Ask: "I'd love to help you track your order! Could you share your order number or the email address you used?"
  → Customer provides order number
    → API lookup: Order found?
      → YES: Display status [PROCESSING / SHIPPED / DELIVERED / DELAYED]
        → If SHIPPED: "Great news! Your order [#ORDER] shipped on [DATE] via [CARRIER]. Here's your tracking link: [LINK]. Expected delivery: [DATE]."
        → If DELAYED: "[EMPATHETIC ACKNOWLEDGMENT]. Your order is experiencing a slight delay due to [REASON]. New estimated delivery: [DATE]. Would you like me to [OPTION A] or [OPTION B]?"
      → NO: "I wasn't able to find an order with that number. Could you double-check it? It's usually in your confirmation email subject line."
        → 2nd attempt fails → Offer human agent handoff
  → Customer doesn't have order number
    → "No problem! I can look it up with your email address."
    → If email lookup fails → Handoff to human agent
```

**Flow B — Returns & Exchanges**
Design similar detailed flow covering:
- Return eligibility check (within [RETURN WINDOW] days, product condition, exceptions)
- Return initiation with shipping label generation
- Exchange process (size/color swap, different product)
- Refund timeline and method explanation
- Edge cases: [SALE ITEMS / GIFT RETURNS / INTERNATIONAL RETURNS]

**Flow C — Product Recommendations**
Design an interactive flow that:
- Asks 3-5 qualifying questions about [USE CASE / PREFERENCES / BUDGET / SIZE-FIT]
- Narrows from [TOTAL CATALOG SIZE] to 2-3 personalized recommendations
- Includes product details, images, and direct add-to-cart links
- Handles "none of these work" with alternative suggestions or human expert handoff

**Flow D — Shipping Information**
**Flow E — Payment & Billing**
**Flow F — Complaints & Escalation**

For each flow, write the complete script including all branch points, response variations, and fallback messages.

### Step 4 — Escalation Protocol
Define clear human handoff rules:
- Automatic escalation triggers: [CUSTOMER USES PROFANITY, MENTIONS LEGAL ACTION, REQUESTS REFUND ABOVE $X, CHATBOT CONFIDENCE BELOW X%]
- Graceful handoff script: "I want to make sure you get the best help possible. Let me connect you with [AGENT NAME/TEAM]. They'll have your full conversation history so you won't need to repeat anything."
- Business hours handling: What happens when no agents are available?
- Priority queue rules: VIP customers, repeat complaints, high-value orders
- Context transfer: What data is passed to the human agent?

### Step 5 — Measurement & Optimization
Track chatbot performance:
- Containment rate target: [X%]
- Customer satisfaction (CSAT) target: [X/5]
- Average resolution time: [X MINUTES]
- Escalation rate: [X%]
- Top 5 unresolved query types to add in next iteration
- A/B testing plan: [WHICH SCRIPTS / FLOWS TO TEST AND HOW]
- Monthly review cadence for script updates based on new products, policies, or common issues