Expedia · Primly Community

Expedia data scientist interview: SQL, case study, and stats questions from my recent loop

ds_dmitri · 5 replies

Went through the Expedia DS interview process for a mid-level DS role on the personalization team. Here's exactly how it broke down, because every breakdown I found online was vague.

OA (take-home, optional depending on role): Not all DS roles get a take-home. Mine didn't. Some roles apparently give a 48-hour case study instead. Ask your recruiter upfront.

SQL round: This was 45 minutes. Harder than I expected. Not just joins and aggregates. I got a question about booking funnel conversion that required a self-join and a window function to calculate session-over-session retention. Then a question about rolling 7-day averages. If you're not comfortable with window functions (LAG, LEAD, PARTITION BY, ROWS BETWEEN), fix that before your loop.

Stats/probability round: This surprised me with how applied it was. No "what is p-value" basics. My questions were: (1) You ran an A/B test on the search results page and conversion went up 3%, but booking revenue per session went down slightly. How do you decide what to do? (2) The Expedia homepage has global traffic. How would you design an experiment across different booking windows (someone planning 6 months out vs. same-day)? Basically they want to see if you understand experiment design, not just the test outcomes.

Case study / product analytics round: Given a dataset about hotel cancellation rates by region and asked to identify drivers and propose interventions. This is the travel-domain knowledge test. Know something about how hotel bookings work (free cancellation policies, booking windows, seasonality).

Behavioral round: Same as what others describe. Leadership, impact, dealing with ambiguity. Have STAR stories ready.

Total process was about 4 weeks from OA to offer. Feedback loop was reasonable. The SQL and stats rounds are the ones where I've seen candidates get cut. Don't underestimate them thinking it's just "analytics."

Feel free to drop questions if you have a loop coming up.

5 replies

analyst_ana

The A/B test question where conversion goes up but revenue goes down is such a good question. That's a real-world tension and a lot of junior DS people would just say "conversion went up, ship it." Did they probe on what metric to use as the primary success metric?

ds_dmitri

Yes, exactly. They wanted to know how I'd define the primary metric upfront and what guardrail metrics I'd monitor. My answer was that booking revenue per session should probably be the primary for a booking platform and conversion rate is a leading indicator, not the goal. They seemed to like that framing.

de_derek

Window functions catching people out in 2026 is a rite of passage. LAG and LEAD should be muscle memory at this point if you're going into any DS or analytics role at a travel company. Their funnel data is inherently sequential.

contractor_kai

What was the comp range for mid-level DS in Seattle if you don't mind sharing?

ds_dmitri

My offer was $155K base, $40K RSU over 4 years, 10% target bonus. That's for L3/mid-level I think, though they don't name levels the same way big tech does. Seattle cost of living adjusted it felt fair.