Katkat
Back

Segment 500k users into three groups

Capital OneData Scientist · Technical Screen
MediumData ScientistTechnical ScreenAnalytics & Experimentation

Last updated

This question evaluates a candidate's competency in data-driven customer segmentation, expected-revenue modeling, and trade-off analysis between churn risk and monetization.

You must segment 500,000 users into three contiguous groups (High, Medium, Low) along a churn-risk score where Q1 is least risky and Q4 most risky. Low must include at least 20% of users. Use the summary to choose boundaries that maximize expected 90-day revenue.

Summary by risk quartile: +----------+---------+---------------------+------------------+ | Quartile | Users | Avg_Monthly_Spend | 90d_Churn_Prob | +----------+---------+---------------------+------------------+ | Q1 | 125,000 | $40 | 0.05 | | Q2 | 125,000 | $35 | 0.10 | | Q3 | 125,000 | $25 | 0.20 | | Q4 | 125,000 | $15 | 0.40 | +----------+---------+---------------------+------------------+ Assume expected 90-day revenue per user = 3 * Avg_Monthly_Spend * (1 − 90d_Churn_Prob). Tasks:

  1. Compute expected 90-day revenue contributed by each quartile.
  2. Enumerate the feasible contiguous 3-way partitions that satisfy Low ≥ 20% and pick the one that maximizes total expected 90-day revenue in High+Medium (show brief calculations).
  3. For your chosen segmentation, report users and expected revenue in each segment and overall.
  4. Recommend actions for each segment (e.g., offers, experimentation plan), and explain how you would monitor drift that could change optimal cut points.