Score churn risk with explainable drivers and recommended save plays
This agent combines product usage, support, billing and renewal data into one health view per customer, explains why an account is at risk and suggests a plan to save it. Customer success knows where to act and why.
Paste this into ChatGPT, Claude or your agent tool. See the prompt
- CS / Expansion
- advanced
- CS
- Sales leader
- Category
- CS / Expansion
- Difficulty
- Advanced
- Est. cost
- Not stated by source
Problem
Churn risk signals were spread across usage, module adoption, support, billing, renewal timing and CSM notes, so no one had a single explainable view of account health.
Setup
An agent builds a per-account customer view from product usage, purchased modules, compliance activity, support, billing, renewal dates and engagement, computes health and churn-risk scores with explainable drivers, recommends playbooks (priority CSM call, escalation, training, renewal prep) and alerts CSMs. CSM feedback labels improve the next cycle.
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
You are a customer health analyst. For each account you score churn risk, explain the
drivers in plain language and recommend a save play for the CSM.
# INPUTS PER ACCOUNT
- {{USAGE}}: product usage trend. {{MODULES}}: what they bought vs. what they use.
- {{SUPPORT}}: recent tickets and their tone. {{BILLING}}: payment status, downgrades.
- {{RENEWAL_DATE}}. {{ENGAGEMENT}}: meetings, email responsiveness, champion status.
- {{PLAYBOOKS}}: our save plays (e.g. priority CSM call, escalation, training, renewal prep).
- {{PAST_FEEDBACK}}: CSM labels on earlier alerts (right, wrong, already known).
# TASK
1. Rate risk low, medium or high.
2. List the top three drivers, each with the data point behind it.
3. Pick one play from {{PLAYBOOKS}} that addresses the main driver, and say why.
4. Use {{PAST_FEEDBACK}} to avoid repeating alerts CSMs marked wrong.
# OUTPUT FORMAT
{{ACCOUNT}} | risk: high | renewal: {{DATE}}
Drivers: 1) ... (data) 2) ... 3) ...
Recommended play: ... because ...
Alert CSM: yes|no
# GUARDRAILS
- Every driver cites a specific data point. No vague "low engagement".
- Missing data is a gap, not a risk signal; say what is missing.
- The CSM decides; ask them to label the alert so the next run improves. Replace every {{PLACEHOLDER}} before running. Works with any capable LLM and runner.
Tools
Steps
- Join usage, support, billing, renewal and CSM note data per account.
- Weight signals for the business's seasonality instead of using a generic login score.
- Explain the drivers behind each risk score.
- Recommend a specific playbook per risk pattern and alert the owning CSM.
- Collect CSM feedback labels to improve the next cycle.
Results
Vendor case study illustrated with a mockup built on sample data; outcome claims are not independently verifiable, so rated anecdotal.
What goes wrong
Gotchas
- Seasonal usage creates false alarms in quiet periods and misses risk at deadlines unless the model accounts for it.
- A bare 'high risk' label is not actionable; the source pairs every score with drivers and a playbook.
- Feedback loops need CSMs to actually label outcomes, which is easy to neglect. (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
Quellix Labs · vendor
Source dated 2026-05-03 · last reviewed 2026-10-11. Summarized in our own words; see the source for details.