Capital One Data Scientist Interview Experience — A Streaming Renewal Case I Didn't Pass

I interviewed for a mini case at Capital One (C1) a while back and didn't pass. Today I want to share the questions and do a bit of a retro.

The interviewer was a guy who used to work in consulting (I get the feeling C1 has a real preference for consulting backgrounds). He started with a quick self-intro, then had me introduce myself. He mentioned the official time allotted for this case is one hour, but a lot of the time it wraps up in 15-20 minutes. Then we got straight into it.

Case background

Say you work at a streaming company (something like Netflix or Hulu), and your core job is to decide whether certain shows should be renewed or cancelled.

Question 1:

There's a show called The Analyst. What factors would you look at to decide whether to renew it?

Question 2:

There are two shows: The Analyst and Sharkbank, both on two-year contracts. You need to calculate the two-year contract profit for each show. The data given was:

The Analyst

• Viewers per season: 5M
• Revenue per viewer: $15
• Annual cost: $50M

Sharkbank, if it succeeds:

• Viewers per season: 7M
• Revenue per viewer: $15
• Annual cost: $60M
• New show startup cost: $20M

Sharkbank, if it fails:

• Viewers per season: 4M
• Revenue per viewer: $15
• Annual cost: $60M
• New show startup cost: $20M

I calculated the profit for Sharkbank under both the success and failure scenarios separately. Then he asked: if success and failure are each 50% probability, what's the combined expected profit?

Question 3:

Which one would you pick — The Analyst or Sharkbank?

Question 4:

If you wanted to increase The Analyst's profit, what would you do?

Then the interview ended.

My retro:

I think my biggest problem was that I treated the case too much like a pure calculation problem. The moment I saw numbers I jumped straight into calculating profit, but what the interviewer actually cared about was:

• Whether my thinking was structured
• Whether the framework was complete
• Whether my assumptions were clear
• Whether my delivery was organized

The math was only a small part of it.

Take question 1, for example — I really should have broken it down with a framework first:

• Financial dimension (revenue, cost, margin)
• User dimension (acquisition, retention, brand impact)
• Strategic dimension (IP value, sequel potential, differentiation)
• Risk dimension (genre volatility, production risk, competition)

instead of jumping straight into calculating profit.

Question 3 also isn't just "pick whichever makes more money" — it's really about:

• Risk appetite
• Portfolio strategy
• Content diversity
• Stable cash flow vs. high risk / high reward

This time my delivery clearly wasn't structured enough — I was kind of calculating and talking at the same time, without laying out the logic beforehand.

Takeaways:

1. A case is not a math problem. The calculation is just support — the core is business judgment.
2. Framework first, details second. You need to open with something like: "I'll look at this from three angles: A, B, C."
3. Doing the expected value calculation is just the start. After that you need to discuss risk tolerance, cash flow stability, and strategic value.
4. Practice communicating more, not calculating more. The interviewer is mainly looking at whether you're someone who can express themselves clearly, reason rigorously, and make judgment calls.

This question wasn't that hard difficulty-wise — the real difficulty was in the delivery and the sense of structure. I think my case experience is genuinely on the thin side, and I need systematic practice rather than relying on instinct to answer.