Capital One Data Scientist Interview Experience — Three Case Rounds, a Scheduling Mishap, and a Rejection

Round 1 had two interviewers — one leading, one silent the whole time.

The BQ was a standard "help others" question. They followed up with a lot of detail, including what materials or records existed for the help I gave, and what I had to sacrifice or give up something important to help someone. A handful of other detail questions too.

Then came the first case. It started by bringing up the revenue and cost of a credit card. They gave a table showing this card's capture rate across different user types:

• A: 30,000 people, 20% capture rate, average spend 100,000 per person
• B: forgot the count, 0% capture rate, average spend 50,000 per person
• C: 150,000 people, 5% capture rate, average spend 20,000 per person

Q1: calculate revenue.
Q2: what are the potential costs?
Q3: given a cost of 200 per card and a fixed cost of 10M, calculate profit — I got 9.8M.
Q4: are there any other costs? Answer: default. Default cost is 1% of total spend. Final profit came out to 2.3M.
Q5: now partnering with another company — given new customer volume, total increase of 5000, each new person spending 40,000, calculate the new revenue. I got 4.375M.
Q6: now each person gets a $500 bonus in year one — how many years to break even? I got 0.46. Then draw the breakeven graph, time on the x-axis, revenue on the y-axis.

Round 2 was with product. The BQ was an accomplishment question. Then: ten ways to design a water bottle, my favorite app and its profit model, its competitors, and if I were CEO, what three metrics I'd track and what I'd improve.

Round 3's BQ was a failure question. The case was about an internet company, though not quite the same as the ones I'd seen before. They gave: $40/month, free for the first 3 months, cost per person $25, installation fee per person $35, overhead per person $20, fixed cost $1M. I calculated average profit per person as -50.

Q2: what would you do if it's not profitable?
Q3: the answer was to extend the contract lock-in period, requiring a 21-month lock, with a $100 penalty for early termination and an expected 10% default rate. How many users are needed to break even? Answer: 6,667.

Then the case closed.

There's a bit of mixed Chinese and English in here and I'm still pretty worked up about it; I haven't recovered from this loss yet, so bear with me. This interview was honestly a disaster. I'd confirmed the interview time with HR over the phone, but then the confirmation email showed the time had been moved up several days. I emailed to ask about it, but HR was on vacation for a week and there wasn't time to fix it, so I just gritted my teeth and did the interview anyway. The interview itself actually went pretty smoothly — the only rough moment was at the end of the first case, on the breakeven question, where I froze for a second before working it out with a hint from the case interviewer. I still didn't get an extra round, and in the end I failed. I'm still pretty upset about it — this was a position I really wanted. Everything went so well up until that point, and then it ended like this. Six-month freeze period. I guess it's just fate, nothing to be done, just have to move on.