What this prompt does
This prompt builds a complete churn prediction and prevention framework for your SaaS instead of just listing churn tips. You supply [product_name], [product_type], [customer_count], and your current [churn_rate], and ChatGPT works through a six-step process: defining churn signals, building a health score, setting up early warnings, designing interventions, creating win-back sequences, and specifying a retention dashboard. The framework is deliberately ordered so signals feed the score, the score feeds the alerts, and the alerts feed the playbooks.
The variables tune how the framework fits your business. [signal_categories] decides which families of leading indicators get monitored, [warning_threshold] sets how many days before predicted churn the system should fire alerts, and [intervention_count] controls how many distinct playbooks it writes. [cohort_dimensions] shapes the analytics layer, telling the dashboard how to slice churn trends. Because the health-scoring step asks for ten concrete metrics with green/yellow/red thresholds and a composite formula, the output is detailed enough to hand to an engineer or analyst.
When to use it
- You are seeing churn climb and want a structured way to catch at-risk accounts before they cancel.
- You have usage, support, and billing data but no unified health score tying them together.
- You are standing up a customer success or retention function and need ready-made intervention playbooks.
- You want a win-back sequence for recently churned customers with defined timing and offers.
- You need to design a retention dashboard and want the metrics and cohort cuts specified up front.
- You are scoping a churn-scoring pipeline and want the signal taxonomy mapped before building.
Example output
The response is a structured framework document. You get a signal taxonomy grouped by category, a health-scoring model listing ten metrics with measurement frequency, threshold values, and a composite 0-100 formula, an early-warning spec with recipients and escalation paths, a set of intervention playbooks (one per churn cause), a multi-touch win-back sequence with timing and channel, and a dashboard definition covering trends, cohort analysis by your [cohort_dimensions], and revenue-impact projections. It is a blueprint, not just advice.
Pro tips
- Set
[churn_rate]accurately — it anchors the urgency and the revenue-impact projections the model generates. - Keep
[signal_categories]aligned to data you actually collect; asking it to weight signals you can't measure produces a score you can't compute. - Tune
[warning_threshold]to your sales cycle — 30 days suits annual contracts, but a shorter window fits monthly plans where churn decisions happen fast. - Use
[cohort_dimensions]that match how you already segment customers so the dashboard output maps onto reports you maintain. - The composite health formula is a reasonable starting point, but validate the weights against your own churned-vs-retained accounts before trusting it operationally.
- Run it once per major segment if your enterprise and self-serve customers churn for very different reasons.
