Chime Data Scientist Interview Experience — SQL and a Business Case in One Technical Screen

This round is split into two parts: one hour of SQL and one hour of a case study.

SQL: They give you a transaction table and a user table. If you're fast you'll probably get through 5-6 questions. The time field is a timestamp, so you need to use date_trunc to handle it. The first few questions are pretty basic and you can knock them out with a few CTEs. It involves several window functions like dense_rank(), lag(), and sum() — the sum needs a rolling window, so it's worth reviewing that.

CASE: Same type of question as the one in the attachment from another post — they give you an Excel file with the specific numbers for each metric: revenue, profit, LTV, COST, CPA. Then you have to give a recommendation on whether to launch the feature based on the data.
It's really about how well you understand the business and the trade-offs. For example:

1. This feature would lower your profit but also lower your CPA — how should you make the recommendation in that case?
2. They segment users into groups A and B, and group B has higher revenue. First they ask you to figure out from the data what's different between the two groups, then what your recommendation would be.
   At the end they ask you to give a summary.