Went through a Unity DS interview loop earlier this year for a role on their player engagement analytics team. Here's the breakdown for people searching for this, because there's almost no content.
Overall structure: 4 rounds after the screen. Totally remote.
Round 1: SQL + data manipulation. Live coding on a shared doc (not a specialized platform, just Google Docs or CoderPad). Two SQL questions. First was a window function problem: given a table of player events, find each player's longest consecutive daily login streak. Second was a multi-table join to compute 7-day and 30-day retention cohorts by game genre. Both are medium difficulty but they want clean, readable SQL, not just working SQL. I commented my logic and that was noted positively.
Round 2: Stats and experimental design. This was a hybrid case + conceptual round. First part: they give you a scenario where a game feature was shipped to 10% of users and show you summary stats. You walk through how you'd evaluate statistical significance, what confounders to flag, whether the sample size is sufficient. Second part: they ask you to design an A/B test for a hypothetical in-game economy change (loot box frequency). Power calculations, randomization unit, guardrail metrics. This was the most rigorous round.
Round 3: Product / business case. 'A game on Unity's platform is showing declining D30 retention. Walk me through your analysis approach.' This is less stats, more structured thinking. They want to see you segment, hypothesize, and prioritize before jumping to models.
Round 4: Behavioral. Standard but they asked specifically about communicating analysis to non-technical stakeholders, which suggests that's a real pain point for the team.
Tip: Know the gaming metrics cold. DAU/MAU, D1/D7/D30 retention, ARPU, LTV. If you've only worked in non-gaming industries, spend a day learning the vocabulary. It's not exotic, but walking in cold on it shows.