Interviewed for a DS role at Dropbox this year, L4 equivalent. Here's the honest breakdown of each component.
SQL round. This was harder than I expected for a DS interview. Not just "write a JOIN and group by", they gave me a multi-table schema (something like users, events, subscriptions) and asked me to answer a product question that required a window function, a self-join, and a subquery. I think the intent was to see if I could actually translate an analytical question into working SQL without a hint. I tripped up on one subquery and had to backtrack; they let me work through it. Time: 45 minutes, one main problem with follow-ups.
Case / product analytics. They gave me a metric that had dropped (I can't give exact detail but it was something like engagement rate on a core feature). They wanted me to walk through how I'd diagnose it: segment by user cohort, platform, region, subscription tier. Then they pushed on what I'd actually look at first and why. This part was very product-sense adjacent. You need to know Dropbox's business model well enough to know what metrics actually matter.
Stats / ML. They asked me to explain A/B testing setup for a feature experiment: how I'd determine sample size, handle network effects (since Dropbox is inherently social via shared folders), and decide if the results were conclusive. This is where Dropbox-specific domain knowledge helps because the social graph makes naive A/B randomization tricky.
Overall: the DS interview at Dropbox is legitimately technical. You need strong SQL (window functions, CTEs), solid stats intuition, and enough product context to discuss metrics in business terms. Prep for both the technical AND the "what does this number mean for the company" questions in the same conversation.