CS Reply Pack

Customer-service reply library and auto-reply rule pack — intent classification, emotion de-escalation (LEAP method), objection-handling matrix, escalation red lines. Outputs a JSON rule table you can import into any helpdesk or chatbot.

Customer Service Reply Pack

What problem it solves

The three deadly sins of support replies: slow, defensive, blame-shifting. This pack hardens "how to answer every message type" into a three-layer structure — intent → strategy → script — that human agents copy-paste and bots use as their rule set.

Core structure

1. Intent classification (8 types)

Pre-sale question / price negotiation / shipping chase / logistics status / return & refund / quality complaint / venting / off-topic chit-chat — each maps to a response priority and template set.

2. Objection matrix (LEAP method)

  • L (Listen): restate the customer's ask before answering — "So you're saying…, correct?"
  • E (Empathize): emotion before facts — "Five days of waiting is genuinely frustrating."
  • A (Ask/Act): give choices, not essays — "Would a reshipment or a full refund work better for you?"
  • P (Promise): every promise carries a deadline — "Tracking number to you by 6 PM today."

3. Escalation red lines (always human handoff)

Safety issues, legal threats (regulator complaints / media), journalists, key accounts (annual spend > threshold), refunds beyond authorization — stop templating immediately, hand off with an internal summary attached.

Deliverables

  1. 30 high-frequency scenario scripts (paste-ready)
  2. Anger de-escalation flow (4-step cooldown)
  3. Review-request and bad-review recovery scripts (compliant — no fake-review inducement)
  4. Auto-reply rule table (intent → template → escalate?), JSON importable

Boundaries

No false promises ("you'll definitely get a refund"), no competitor bashing; refund commitments defer to the merchant's actual policy via placeholders.