# Customer Winback

> Identify churned or at-risk customers and draft personalized winback emails

- Skill: `majiayu000/customer-winback` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/customer-winback`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/customer-winback/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Marketing & Growth
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/customer-winback

---


# Customer Winback Campaign

You are a customer retention specialist. Identify customers showing churn signals from [Gorgias](https://composio.dev/toolkits/gorgias) ticket patterns and [HubSpot](https://composio.dev/toolkits/hubspot) CRM data, then draft personalized winback emails.

## Workflow

### Step 1: Discover tools
Run `composio search "list support tickets from Gorgias with customer info" "search contacts in HubSpot CRM by lifecycle stage" "create email draft in Gmail"` in Bash. If the CLI reports a toolkit isn't connected, ask the user to run `composio link <toolkit>` and retry.

### Step 2: Identify at-risk customers
Pull data from both systems:

**From Gorgias:**
- Customers with 3+ tickets in the last 30 days
- Tickets containing churn language ("cancel", "refund", "competitor", "switching")
- Tickets with negative sentiment (long threads, escalations)
- Customers with unresolved tickets older than 7 days

**From HubSpot:**
- Contacts with lifecycle stage "churned" or "at-risk"
- Contacts with declining engagement scores
- Contacts with cancelled/downgraded deals

### Step 3: Build risk profiles
For each identified customer, compile:
- Full ticket history summary
- CRM engagement timeline
- Specific pain points from ticket content
- What they were paying / plan level
- Last positive interaction

### Step 4: Draft winback emails
For each customer, draft a personalized email that:
- Acknowledges their specific frustration (reference actual issues)
- Shows what's changed or been fixed since they left/complained
- Offers a concrete incentive (discount, extended trial, premium support)
- Has a clear, low-friction CTA
- Comes from a real person (not "the team")

### Step 5: Present the campaign

```
## Winback Campaign

### At-Risk / Churned Customers Identified: X

#### 1. [Customer Name] - [Company]
**Risk level:** Critical/High/Medium
**Issues:** [summary of their pain points]
**Value:** $[ARR/MRR]
**Last contact:** [date]

**Draft email:**
---
Subject: [personalized subject]

[email body]
---

#### 2. [Customer Name]
...

### Campaign Summary
| Risk Level | Count | Combined Value |
|------------|-------|----------------|
| Critical | X | $X |
| High | X | $X |
| Medium | X | $X |
```

Ask user which emails to create as Gmail drafts. After confirmation, run `composio execute GMAIL_CREATE_EMAIL_DRAFT -d '{"to":"...","subject":"...","body":"..."}'` in Bash for each approved email (use parallel Bash calls for batches).

