Retention and Churn for a Transactional Consumer App

Context: You are analyzing retention and churn for a transactional consumer app (e.g., food delivery, ride-hailing). Users place discrete orders over time. Your goal is to define, compute, and interpret retention and churn correctly for product, marketing, and experimentation use-cases.

Tasks

1. Definitions and Justification

• Choose precise definitions for:
  • Cohorts: signup vs. first-purchase (activation) cohorts.
  • Activity: active if 1+ order in a period.
  • Retention types: N-day, week N, rolling, and bracket retention.
  • Churn: no activity for K consecutive periods.
• Justify choices based on decision use-cases.

2. Formulas and Correct Computation

• Provide formulas for cohort retention and churn using proper risk sets.
• Handle right-censoring and delayed conversion.
• Explain pitfalls such as survivorship bias, Simpson’s paradox, and seasonality.

3. Measuring Long-term Retention Impact of a Treatment via Survival Analysis

• Define time-to-churn, hazard, and cumulative incidence.
• Specify how to compare treatment/control curves (log-rank or stratified tests).
• Explain covariate adjustment.

4. Rolling vs. Strict Cohort Retention

• Show how rolling retention can disagree with strict cohort retention.
• Include a made-up numerical example and compute both correctly.
• Explain how to reconcile for executives.

5. Windows and Experimental Design

• Explain how you would set washout, observation, and attribution windows.
• Discuss how these choices affect experiment power and bias.