PwC · Primly Community

PwC data scientist interview (SQL + case + stats): what the loop actually looks like in 2026

de_derek · 5 replies

went through the PwC data scientist interview loop earlier this year for a role on their Deals Analytics team, so dropping this while it's fresh.

the process was four rounds total:

round 1: recruiter screen (30 min) normal stuff. what tools do you use, why PwC, are you ok with travel. nothing technical.

round 2: technical phone screen (60 min) this was half SQL half stats. SQL questions were not easy junior-level stuff. they gave me a schema with three tables (clients, engagements, billing records) and asked me to write queries to identify clients with declining year-over-year revenue and rank them within industry segments using window functions. if you're not comfortable with ROW_NUMBER, RANK, LAG and self-joins you'll struggle here.

the stats piece: they asked me to explain p-values without using the words "probability" or "null hypothesis" (genuinely a good question, tripped me up), then walked through a scenario about A/B test results showing lift in one region but not another and asked what could explain that. interaction effects, segment size differences, different base rates. they wanted to see if you actually think about confounders.

round 3: case interview (45 min) this surprised me. PwC is still consulting at heart so expect a mini-case even for DS roles. mine was about a retail client seeing higher churn in their loyalty program after a pricing change. you're expected to structure your analysis, not just say "build a model." frame the business question, identify the data you'd need, describe the analysis approach, then what you'd recommend. they're evaluating structured thinking as much as technical depth.

round 4: behavioral + partner interview (45 min) STAR format throughout. questions about navigating ambiguity, dealing with a stakeholder who didn't trust your analysis, delivering findings that changed a client's direction. no more SQL here, this is purely about communication and fit.

a few things that seemed to matter: Python was fine but they asked about R too (several team members use it). Tableau came up a lot for visualization. having client-facing or presentation experience helped, they're not looking for a pure research DS.

total timeline from recruiter outreach to offer was about 6 weeks. turnaround between rounds was faster than I expected, maybe 5-7 days each.

5 replies

analyst_ana

the case interview thing for a DS role caught me off guard when I was prepping for PwC last year too. I kept practicing coding problems and then the actual loop had this whole structured-problem-solving component. do they give you data to look at during the case or is it purely verbal?

brand_ben

verbal for mine. no laptop, no data. they describe the scenario and you work through it on a whiteboard (or just talking through it on a video call). it's more McKinsey-style than Google-DS-style, which makes sense given the firm's DNA.

consultant_cam

this tracks exactly with what I saw when I was at MBB and later interviewed a few data people who came from PwC. the case round is non-negotiable for advisory-track DS roles. they need people who can stand in front of a CFO and explain their analysis, not just hand off a notebook. the SQL is hard but the case is what screens people out.

for prep: practice case interviews like you're interviewing at McKinsey, not like you're interviewing at Airbnb.

ml_mike

the p-value question without using 'probability' or 'null hypothesis' is actually something I'm going to steal for my own interviews. forces people to reveal whether they actually understand the concept or just memorized the phrase.

ops_omar

6 weeks is pretty fast for a Big 4 process. my friend went through the PwC audit tech loop last year and it dragged 10+ weeks. maybe Deals moves faster because client timelines don't wait.