This was for the Data Science, Analytics (PhD) summer intern position.
The interviewer's background was in monetization. His questions followed pretty much the standard pattern you see on the forum — business goal / metric / experiment — and then he dug into specific points based on what I said.
Roughly, he asked:
- Do you think it would be worth adding a Venmo-like peer-to-peer money transfer feature to Messenger?
- What are the possible negative effects or potential risks?
- How would you monetize it?
- How would you test it? I said you'd do a pilot test or launch a beta version of the app, that kind of thing — it felt like he was steering the conversation toward causal inference.
- What metric would you use to measure whether the feature hit its expected goal?
- How would you tell whether user engagement went up for people in the beta test compared to people not in the test: difference-in-differences.
When I answered that last question I mentioned the beta test, and the interviewer asked what happens if a friend of yours isn't in the test but also wants to use the feature.
Then he kept pushing and asked what I thought the distribution of the number of transfers would look like in the first 30 days after the app launched. It's obviously going to be skewed, so then he asked the classic follow-up — roughly where would the mean / median / mode / 95th percentile fall. And then how would that distribution change two months later?
Discussion
Loading comments…