just wrapped the McKinsey data scientist interview loop in early 2026, figured i'd write this up while it's fresh.
the process was four rounds over about three weeks. recruiter screen, then a technical phone screen, then a full onsite (virtual in my case), then a final partner call. not all DS roles are the same at McKinsey, so YMMV by team, but this was for an analytics/data science role embedded in their QuantumBlack AI practice.
SQL showed up in two of the four rounds. not LeetCode-hard SQL, but also not trivial. think: multi-table joins with aggregation, window functions, filtering on aggregated values. one question had me write a query to compute a moving average and flag outliers. they gave me a whiteboard/shared doc to type in, no actual database to run it against. so you need to be fluent enough that your code is obviously correct on read.
Stats more conceptual than computational. they asked me to explain what p-value actually means (watch out, this is a trap question where a lot of people trip up). probability questions, one on A/B testing design, and one where i had to describe how i'd detect and handle selection bias in an observational dataset. no coding on the stats section.
The case this is the McKinsey thing that's unique and people underestimate. even for a data science role, you get a light version of a consulting case. it's structured as: 'a client in X industry has problem Y, you have this dataset (shared as a short table). what do you do?' they're not expecting perfect consulting polish, but they do want to see structured thinking: frame the problem, state assumptions, propose an analysis, interpret results.
i practiced this literally zero times and it showed. i stumbled a lot. still got through but it was not my best moment.
PEI / behavioral standard McKinsey PEI (personal experience interview). pick two or three good stories and know them cold. they drill down hard: 'what exactly did you say', 'what was the outcome in numbers', 'what would you do differently'. vague stories do not survive the follow-up questions.
level was mid-senior IC, offer was competitive for consulting, base in the $155-175k range for NYC if you're curious. PM me if you want more detail.