Churn Prevention Specialist

This skill splits your churn before trying to fix it — voluntary versus involuntary, early-tenure versus established, by segment — because averaging across all of them hides every meaningful pattern. You share your product description, current churn rate, and customer segments; the skill builds a signal taxonomy calibrated to your product (login decay, feature abandonment, champion departure, support sentiment), diagnoses your largest churn driver by cohort, and maps a specific intervention to each driver rather than applying a blanket discount. Dunning and failed-payment recovery are addressed separately, since involuntary churn is frequently the highest-ROI fix available.

A realistic starting prompt looks like: 'B2B project-management SaaS, 450 accounts, 4.2% monthly churn, heaviest losses in months 2-4 and again at annual renewal. We do a generic 20%-off save offer at the cancel screen. No dunning system yet.' From that, the skill identifies which of the four or five canonical churn drivers is dominant for your profile and where the earliest detectable signal sits — weeks before the cancel click — so intervention can happen before the decision is made.

The output for that scenario would open with a ranked driver diagnosis — for example: (1) Activation failure / month-2 drop, primary signal: <3 core-feature touches in first 14 days; (2) Renewal-price shock, primary signal: no usage growth in 60 days before renewal date; (3) Involuntary churn, current dunning: none. Each driver then gets its own intervention plan: the trigger condition, the action, the owner, and the metric that confirms it is working. Save-flow copy, win-back email sequences, and a dunning cadence are included where they apply.