Google Data Scientist candidates report SQL and Python screens, deep A/B testing and statistics questions, product and ML cases, and a Googliness round. Loop structure varies between candidates, so prepare for all of these.
$196K
Avg. Base Comp
$320K
Avg. Total Comp
3-5 rounds
Typical Rounds
Not reported
Process Length
Across candidate reports, the Google Data Scientist interview is less about obscure coding and more about how you reason through experiments and product problems. A/B testing and experimentation come up repeatedly, and several candidates say the follow-up questions went deeper than the opening prompt. Reported topics include choosing a success metric, sample size, novelty effects, sample ratio mismatch, seasonality, and conflicting metrics such as click-through rate rising while session time falls.
SQL appears in many accounts, and window functions are the most commonly mentioned skill, for example a 7-day rolling active-user metric or a 7-day retention metric from event logs. Python reports vary from basic list processing to cleaning a messy dataset.
Product and ML cases are also reported. Examples include investigating a rise in Pixel return rates, improving Google Maps, and designing a YouTube recommendation or engagement model. Interviewers are described as pushing back on assumptions, so practice clarifying the problem, naming metrics and guardrails, and defending your choices aloud.
Loop structure differs between candidates. Some report two screens followed by a virtual onsite, while others describe four rounds plus a behavioral Googliness round. Ask your recruiter what each round covers. Outcomes in the reports range from offers to rejections, with some still waiting.
Synthesized from 42 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Google process.
Every metric I named got a follow-up, and that was the real theme of my Google Data Scientist loop. I would say a number, feel good about it, and then the interviewer would ask what that number might miss. Being ready for that second question mattered more than having a clever first answer.
The process started with a recruiter call and then a tech screen. The screen leaned more on statistics than I expected, with some probability questions and a small amount of coding. The second half moved to questions about a project from my resume, so I'd recommend going back through your old work and being able to explain your decisions in detail, not just summarize what you built.
The onsite was split into several separate conversations. One covered stats and probability. Another got into product metrics and how to tell whether a change was actually working. There was also an experimentation round, a round of more technical questions, and a googleyness conversation. The product portion seemed to depend on the team, so I wouldn't assume everyone gets the same set of questions.
For prep, I used Trustworthy Online Controlled Experiments for the A/B testing material. The most useful thing I did was a mock interview with a Google data scientist on Prepfully. My answers sounded solid in my head but got vague as soon as someone pushed back. She kept asking why I had picked each metric and what could make it misleading. It was frustrating at the time, but it's a big part of why I felt steady in the real loop. I accepted the offer.
Prep tip from this candidate
For the product metrics round, practice pairing every metric you propose with the specific ways it could mislead, since interviewers kept pressing past the first answer. A mock with someone who challenges your metric choices is worth doing, because answers that sound fine on their own can get vague under pushback.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Google
Write a query that returns all neighborhoods that have 0 users.
| Question | |
|---|---|
| 2nd Highest Salary | |
| Top Three Salaries | |
| First Touch Attribution | |
| First to Six | |
| Merge Sorted Lists | |
| Experiment Validity | |
| String Shift | |
| 500 Cards | |
| Last Transaction | |
| Button AB Test | |
| Top 3 Users | |
| Raining in Seattle | |
| Third Purchase | |
| Job Recommendation | |
| Minimum Change | |
| Impression Reach | |
| Lazy Raters | |
| Jars and Coins | |
| WAU vs Open Rates | |
| Find the First Non-Repeating Character in a String | |
| Network Experiment Design | |
| Daily Retention Summary | |
| Bucket Test Scores | |
| Complete Addresses | |
| RMS Error | |
| Find the Index with Equal Left and Right Sum | |
| Reducing Error Margin | |
| Find Bigrams | |
| Delivery Estimate Model |
Synthesized from candidate reports. Individual experiences may vary.
Candidates report different starts. Some had a recruiter call covering experience, role expectations and logistics, while one was invited straight to a technical screen. It is worth confirming the format of each round with your recruiter.
Candidates typically describe SQL with window functions or multi-table joins, plus Python that ranges from list processing to cleaning messy data. Statistics questions are also reported, including Bayesian reasoning and simulating a biased coin toss. Narrating your logic and edge cases appears to matter as much as the final answer.
Several candidates report that this was the most demanding portion. Expect to design an experiment, choose core and guardrail metrics, estimate sample size, and handle sample ratio mismatch, novelty effects or conflicting metrics. Follow-up scenarios may build on the same prompt, so prepare to go deep rather than just naming a framework.
Candidates report product cases such as a Pixel return-rate increase, a Google Maps feature, or a drop in engagement, along with ML framing for recommendation or engagement models. Interviewers may probe assumptions, data availability and evaluation. Clarify the goal first, then metrics, then approach.
Several candidates describe a behavioral or Googliness round on working style, leadership, collaboration and difficult decisions. Final outcomes differ by candidate, and some were still waiting to hear back when they shared their experience, so plan to keep other applications moving.
