Capital One Data Scientist Interview Experience — Power Day With Three Back-to-Back Case Studies

Mini Case
About streaming subscriptions: a new streaming service is launching, calculate the break-even point. Asked whether the break-even user count I calculated was reasonable — if not, what could be done to make it happen.

Power Day

• Case #1 was the original weather insurance case
  • Asked what factors to consider when it comes to the target customer
  • Premium prepaid for 12 months: $30/month. Servicing cost: $3/month. Cost of benefits paid if a failure occurs: $8000. Regulatory expense: $4/quarter + $300 if a benefit is paid. Asked for the maximum likelihood.
  • Shared the risk for groups A, B, C, D plus the cumulative risk, asked which ones to pick to maximize profit (A+B) — draw a chart and explain the reasoning
  • How to find a way to sell to groups C and D
• Product case — pick your own favorite digital app
  • Basically: introduce the app, why you like it, what the revenue streams are, who the competitors are, what the advantages are by comparison, and the three key metrics you'd care about as a product manager
  • Write out six ways to optimize the app
• Case #2 was about a co-branded credit card
  • Asked what type of company we'd choose to partner with
  • Gave data on three user segments (headcount, spending before and after the partnership); asked what pattern the data showed, then calculate profit from the numbers (discount rate 20% as cost, revenue...) — didn't manage to write down all the numbers in time. Each new user brings in $300 in revenue; asked how many new users would be needed to break even.
  • Spend $200K on a marketing campaign: option 1 keeps the original discount rate with a 0.12 acquisition rate; option 2 cuts the discount rate in half with a 0.07 acquisition rate. Asked to compare the two options, then asked whether there was any red flag in our assumptions.