Score daily churn risk from support, usage and billing into HubSpot
Every day, this setup combines support tickets, product usage and billing signals into a churn risk score per customer and starts outreach for the riskiest ones. Teams step in before a customer decides to leave.
Paste this into ChatGPT, Claude or your agent tool. See the prompt
- CS / Expansion
- intermediate
- CS
- RevOps
- Category
- CS / Expansion
- Difficulty
- Intermediate
- Est. cost
- Not stated by source
Problem
Support tickets, product usage and billing events each hint at churn, but nobody combines them daily. Interventions start after the customer has already decided to leave.
Setup
A daily n8n cron collects product analytics, support tickets from Zendesk or Intercom, billing data and email signals, and sends them to GPT-4 or Claude for multi-signal analysis. Accounts get a risk score routed by thresholds; the workflow drafts a personalized intervention, writes risk data to HubSpot, alerts a high-risk Slack channel and can send a booking link.
Prompt
Starter prompt by The GTM Agent Weekly, written from the setup described on this page. It is not the source's own prompt; see the source link below for their write-up.
# ROLE
Each day you assess churn risk for customer accounts from several signals and draft an
intervention for the ones at risk.
# INPUTS PER ACCOUNT
- {{USAGE}}: product usage trend. {{TICKETS}}: recent support tickets and their tone.
- {{BILLING}}: failed payments, downgrades, renewal date. {{EMAIL_SIGNALS}}: replies, silence, unsubscribes.
- {{THRESHOLDS}}: score ranges for low, medium and high risk.
# TASK
1. Score risk from 0 to 100 and map it to low, medium or high using {{THRESHOLDS}}.
2. List the three signals that drove the score, each with its data point.
3. For medium and high: draft a short, personal message from the account owner that
addresses the main issue and offers help (and the booking link {{BOOKING_LINK}} if high).
# OUTPUT FORMAT (JSON)
{"account": "", "score": 0, "level": "low|medium|high", "drivers": ["", "", ""], "draft_message": ""}
# WHAT THE WORKFLOW DOES
Write score and drivers to the CRM, alert {{HIGH_RISK_CHANNEL}} for high risk, and queue the
draft for the owner.
# GUARDRAILS
- Missing data lowers confidence; it does not raise risk. Say what is missing.
- The message never mentions tracking or scores. Have a person review before sending. Replace every {{PLACEHOLDER}} before running. Works with any capable LLM and runner.
Tools
Steps
- Import the workflow and read its sticky notes.
- Set environment variables for analytics, the high-risk Slack channel, CS sender and CRM URLs.
- Connect credentials for the AI provider, support system, billing, HubSpot, Slack and email.
- Tune the analysis prompts, risk thresholds and intervention templates to your business.
- Test with sample customer data, check routing, then activate the daily schedule.
Results
Workflow template; the source's prediction-window claim is unverified.
What goes wrong
Gotchas
- Automated intervention emails to at-risk customers can backfire; keep a CSM review step (editor note).
- Thresholds copied from a template will not match your base rates; calibrate on past churn first (editor note).
- Many credentials across systems make the workflow brittle when any token expires (editor note).
Get 5 like this every week, free.
Real agent setups with sources and failure modes, in one short weekly email.
Free. One email a week. Unsubscribe anytime.
Source and credit
shae, n8n template · vendor
Source dated 2026-10-11 · last reviewed 2026-10-11. Summarized in our own words; see the source for details.