AI automation case study
Churn Early-Warning
A scheduled n8n workflow that scores customer-health signals with explainable rules and prepares high-risk accounts for human review.
- Status
- Prototype
- Origin
- n8n
- Date
- 2026-08-23
Scheduled churn-risk scan with validation, explainable scoring, human-review queue, portfolio record, and executive alert branches
Open full-size imageProject proof
A closer look
Explore the interface and the decisions built into it. Open any capture at full size to read the details.
Earlier synthetic churn validation capture with high-risk, watchlist, and no-action branches
Open full-size imageProject proof
Watch the proof film
A short local film made from privacy-reviewed project captures. No contact records or client identities are shown.
Local privacy-reviewed proof film
Challenge
What needed to change
Usage, support, billing, and renewal signals can be scattered across systems, while an opaque risk score gives an account manager little to act on.
System
How the system works
A weekday schedule reads customer-health signals from Google Sheets, quarantines invalid or duplicate rows, applies an explainable 0-100 churn-risk score, prepares high-risk review rows, records the portfolio analysis, and sends one aggregated executive alert through Google Chat.
- Input
- Usage, login, support, payment, NPS, renewal, and MRR signals
- Output
- High-risk review queue, portfolio record, and aggregate alert
- Control
- No automatic outreach, discount, refund, or billing change
Outcome
What is supported by the record
The public captures show the current scheduled review workflow and the earlier three-branch validation demo. The system recommends human follow-up only and never contacts customers or changes billing automatically.
Screenshot
Current privacy-reviewed n8n canvas capture
Validation
Risk branches tested with synthetic customer signals
Tools
Connected tools and methods
- n8n
- Google Sheets
- Google Chat
- Scheduled trigger
- Explainable scoring
- Code nodes
- Shared error handler
Next step