Big Tech platforms Data Analyst Interview Loop
In Big Tech data analyst interviews, the same title means different things at each company. Amazon calls the role Business Intelligence Engineer. Meta has three analyst roles that candidates often mix up. Google places analysts across Ads, Cloud, Operations and People, with a different loop in each. Underneath, the shape is shared. There's a recruiter call, a timed SQL screen, and three to five interviews: SQL and analysis, a business or product case, stakeholder communication, and behavioral. There's less statistics than in the data scientist loop, and it's about reading results.
Practice by round
Timed SQL basics, sometimes followed by a question on reading a statistic. These screens check fluency, and the interviewer moves on as soon as you finish. Be clean, quick and correct.
Window functions, CTEs and multi-table joins on campaign, customer, order or event data. Then comes the question most of them care about. What does this tell the team, and what would you look at next? Amazon also asks you to rewrite the query another way, and what happens at a hundred times the volume.
| Question | Difficulty |
|---|---|
| Top Engagements | Medium |
| Most Active Users On Messenger | Medium |
| Expensive Projects | Medium |
| Class Performance | Medium |
| New Products | Medium |
| Activity Rank | Medium |
| Common Friends Script | Medium |
| Top Posts Per Channel | Medium |
| User Activity Analysis | Medium |
| Interaction Summary | Medium |
| Finding Purchases | Medium |
| Top-Rated Support Employees | Medium |
| Top 2 Highest-Selling Items | Medium |
| Pages Currently Active | Medium |
| User Flag Performance Analysis | Medium |
The business case, with details left out on purpose. You might define a success metric for a feature, decide whether to launch, pick a north-star metric for an e-commerce product, or explain two metrics that disagree. There are estimation cases too, like how many birthdays or how many Big Macs. You're scored on scoping: what you ask, what you assume out loud, and what you recommend. In the real loop, stakeholder communication is often its own interview. The projects let you practice the whole thing end to end.
| Case study | Difficulty |
|---|---|
| Returning User | Easy |
| Search Toolbar Change | Medium |
| Twitter Feed | Medium |
| Spike in Uploads | Medium |
| Investigate the Discrepancy | Medium |
| Too Many Friends | Medium |
| Facebook Groups in the US | Medium |
| Defining a Product Success Metric | Medium |
| Launch or No-Launch Decision Framework | Medium |
| Fake News on Facebook | Hard |
Statistics at the level of interpretation: A/B tests and hypothesis tests, mean versus median versus mode, skewed distributions and outliers, sampling distributions and Type I error, and z versus t.
| Question | Difficulty |
|---|---|
| Non-Gaussian Distribution | Easy |
| Random Bucketing | Easy |
| Statistics Mean Function | Easy |
| Interpret P-value | Easy |
| Variance in Unsupervised Model | Easy |
| New Observation is Outlier | Easy |
| Box Plot and Histogram | Easy |
| Expectation of Variance | Medium |
| Mean, Median Age in Mexico | Medium |
| Used by Different Groups | Medium |
| Serving ads within Newsfeed | Medium |
| Sample Size | Medium |