Customer Churn & Survival Analysis: Hazard Rates, Kaplan-Meier & Retention Curves

What you'll build

What you will be able to do

Each lab ends with a concrete result that you build yourself and the checker confirms, cell by cell.

• Turn start and cancellation dates into a churn flag and a tenure in whole months, censor the active customers at the cut-off date and calculate the churn per customer-month.
• Follow a cohort of 2,400 subscribers through their first 12 months, work out how many are still at risk each month and turn cancellations into a hazard rate per tenure month.
• Chain 24 monthly hazards into a survival curve, read off the median customer lifetime and measure how front-loaded the losses are.
• Build a Kaplan-Meier table straight from a customer list with COUNTIF and COUNTIFS, keep the censored customers in the risk set, read off the median lifetime and draw the step-shaped survival curve.
• Draw the survival curves of the Basic and the Plus plan, work out how many Basic customers would have churned if both plans were equally risky, and test the gap with a simplified log-rank chi-square.
• Turn the survival curve into expected paid months and a discounted 24-month customer lifetime value, and see how much of it arrives in the first half-year.
• Cut 30 customers by plan, acquisition channel and onboarding call, calculate the churn rate of every segment and its lift over the average, and find the strongest churn signal.
• Price a save offer for three target groups: how many customers it really saves, what they are worth, the net value and ROI of each option, and the churn reduction the offer needs just to break even.