Meta BI Analyst Interview Loop
Meta's BI roles sit in operations, integrity, risk and finance rather than product analytics. Reports on the Meta BI analyst interview match its other analyst loops. There's a recruiter call, a screen that's mostly SQL, and around four interviews: SQL and tools, a case, cross-functional, and behavioral. Ask which team the role is in, because that's where the case comes from.
What decides it
Clean SQL under time, and a recommendation without hedging. The BI version puts extra weight on knowing what a decision-maker needs to see.
Practice by round
Timed SQL on operational tables rather than product ones. You'll get joins, aggregations, date handling, a distinct count and a top-N. Meta's interviewers move on as soon as you finish, so clean and quick beats thorough.
Window functions, CTEs and multi-table joins on the kind of data Meta's operations teams hold. That includes ticket processing rates, session time, feature completion, and promotions and sales. In the real loop there's also a question about the reporting layer. What would you precompute, and how does a report break when a source table changes?
| Question | Difficulty |
|---|---|
| Products Never Sold | Medium |
| Top Two Media Types | Medium |
| Customer Consumable Sales Percentages | Medium |
| Sales Evaluation on Media Formats | Medium |
| Day 1 Common Reactions | Medium |
| Successfully Sent Messages | Medium |
| Processed Ticket Rate By Type | Medium |
| Find the total number of approved friendship requests in January and February | Medium |
A metrics case tied to the team. You define a success metric, judge whether to launch, find out why two numbers disagree, or measure the health of Groups. Then you say what you'd recommend.
| Case study | Difficulty |
|---|---|
| Returning User | Easy |
| Finding Invalid Schools | Medium |
| Meta/Facebook Nicknames | Medium |
| Double the Number of Ads | Medium |
| Too Many Friends | Medium |
Enough statistics to say whether a number can be trusted. That covers reading an A/B result, sample size, conversion rate, and mean versus median. There are no derivations.
| Question | Difficulty |
|---|---|
| Conversion Rate | Easy |
| Negative R Squared | Easy |
| Expectation of Variance | Medium |
| Mean, Median Age in Mexico | Medium |
| Used by Different Groups | Medium |