S18 · lifecycle simulation

A churn model from intake to retirement

People, data and outputs in this scenario are simulated. Each stage is recorded when you commit; the reveal comes after.

5stages
45 minabout
0stages committed

The brief

A B2B software company (simulated) has 6,000 customer accounts: 70% on monthly plans, 30% on annual contracts. The customer-success team can run proactive outreach for about 40 accounts a week. The head of customer success (simulated) asks: "Can you predict who will churn?"

All data, results and opinions are synthetic or simulated. Carry your earlier decisions forward: later stages assume the choices from the reveals.

Stage 1 of 5

Available data: monthly subscription snapshots, product usage events, support tickets, invoices. Churn is recorded when a subscription ends. Annual accounts can only churn at renewal dates.

Which framing do you propose?

Weekly scoring date; target = monthly-plan account cancels within 60 days; features only from before the scoring date; rank to fill 40 outreach slots; annual accounts treated separately.Classify every account ever seen as churned or not, using all data up to today, so the model learns from the full history, every account contributes a label, and there is no need to choose a scoring date or a horizon.Predict the exact churn date for each account with a survival model.