Went through the HubSpot DS interview process for a mid-level data scientist role on their product analytics team. Four rounds. Here's what was in each.
Round 1: SQL + Analytics
CoderPad, 45 minutes. Two SQL questions. First was a multi-table join with a window function: something like "given a table of user sessions and a table of feature interactions, find users who used feature X in their first 7 days but stopped using it after day 30." Standard retention analysis framing.
Second question was more open: given a funnel dataset, identify where users drop off and suggest one hypothesis for why. They wanted SQL to pull the numbers AND a short narrative explanation. That second part trips people up. Know how to narrate a query output, not just write the query.
Round 2: Product case
45 minutes, no data provided. The prompt was something like: "HubSpot just launched a new inbox feature for customer service teams. How would you measure whether it was successful? How would you design an A/B test for it?"
They care about metric selection: you need to pick a north star metric and explain why it beats alternatives. I went with "tickets resolved per agent per day" with engagement rate as a guardrail. They pushed on whether engagement was a good guardrail. Be ready to defend your choices.
Round 3: Stats
Short, about 30 minutes. Covered: experiment design (power, sample size, type I/II error), "what's wrong with this A/B test" scenarios, and one question about working with imbalanced data. No deep probability theory. More applied stats than academic.
Round 4: Hiring manager behavioral
Half technical, half values. She asked about a time I had to influence a product decision with data when the PM was resistant. Classic cross-functional influence question. Also asked how I prioritize when I have three requests with competing urgency.
Offer I saw (for reference)
Mid-level DS, Boston (hybrid). My offer was around $155k base, $30k target annual bonus, RSUs vesting over 4 years. Total comp in the range of $195-200k all-in at current stock. That was competitive but below what I got from two other offers, so I declined. Market for experienced DS in 2026 is decent if your SQL is solid.