Boston Consulting Group Interview Questions — Page 1

Boston Consulting Group·Data Scientist·Technical Phone Screen

Apr 2026

BCG data scientist interview, one big SQL question that looked manageable until I actually tried to write it out. More of a technical screen vibe than a full onsite, and the problem had enough moving parts to trip you up if you're not careful with window functions and date filtering.

Write a SQL query that computes, per car model, the total revenue for the current 7-day window and the prior 7-day window (completed orders only), identifies the top 3 customers by spend in the current window with their share of model revenue, and calculates week-over-week revenue growth. Ties in the top-3 ranking should be broken by the customer's earliest purchase date on that model across all history.

“This one took me longer to structure than I expected.”

View Post

Boston Consulting Group·Data Scientist·Technical Phone Screen

Apr 2026

BCG data scientist interview with a pandas/data cleaning question that looked straightforward on the surface but had a few gotchas worth thinking through beforehand.

Given a customers table and an orders table, merge them in pandas keeping all customers, replace null order amounts with 0 to compute a total_spent column, fill missing country values with the mode of known countries, and return the result sorted by total_spent descending.

“I got the left merge and fillna parts pretty quickly but fumbled on the mode imputation.”

View Post

Boston Consulting Group·AI Engineer·Technical Phone Screen

Apr 2026

Interviewed for an AI Engineer role at BCG and got hit with a math/logic puzzle about moon phases. Not what I was expecting from a consulting firm but here we are.

Given a specific month and day, along with the moon phase on January 1st, calculate what moon phase it would be on that date. Assume standard month lengths and a cycle of 8 repeating phases.

“Took me a minute to even parse what they were asking.”

View Post

Boston Consulting Group·Data Scientist·Hiring Manager Screen

Apr 2026

BCG data scientist interview that was essentially one long compound question disguised as a structured conversation. Whoever designed this prompt was not messing around.

Give a 90-second pitch tailored to this specific role, then walk through a project where your analysis actually changed a business decision. Cover the objective, constraints, your responsibilities, the hardest disagreement you had to navigate, the metrics you defined upfront, the trade-offs you made under pressure, a mistake and how you handled it, and the quantifiable before-and-after outcome.
If we called your last manager and a peer separately, what one sentence would each say about something you should keep doing and something you should change?

“This is basically five questions stapled together and they expect you to hold the thread across all of it.”

View Post

Boston Consulting Group·Data Scientist·Technical Phone Screen

May 2026

BCG data scientist technical screen, basically one big pandas problem that looked manageable until you actually had to write production-quality code for it. A lot of moving parts crammed into a single question.

Given three DataFrames (customers, products, orders), write idiomatic vectorized pandas code that: filters to completed orders and computes revenue, merges the tables with appropriate validation, calculates a clipped discount percentage with median imputation for missing values, identifies each customer's first purchase per model via groupby/transform, enforces correct dtypes (categorical and datetime), avoids SettingWithCopyWarning, and returns a specific set of columns sorted by revenue descending. The solution must be idempotent.

“This was a lot.”

View Post

Boston Consulting Group·Data Scientist·Online Assessment (OA)

May 2026

BCG data scientist online assessment, two probability questions under proctoring. Not a vibe-check round, they want you to actually derive things from scratch with no formula sheet to lean on.

Given n i.i.d. Bernoulli(p) random variables, derive a closed-form expression for the probability that their sum is even.
A fair coin is flipped repeatedly until two consecutive flips show the same outcome. What is the expected number of flips required?

“I knew parity arguments were a thing but blanked on how to set them up cleanly under time pressure.”

View Post

Boston Consulting Group·Data Scientist·Technical Phone Screen

May 2026

BCG data scientist technical screen, pretty dense with ML fundamentals. They covered a lot of ground fast and expected you to derive things from scratch rather than just name-drop concepts.

Given a specific set of prediction scores and binary labels, compute ROC AUC manually using pairwise positive-negative comparisons, then verify it by drawing the ROC curve and applying the trapezoid rule.
How does severe class imbalance, say 1% positive rate, change how you interpret AUC versus Average Precision?
For each of four scenarios, choose the appropriate output activation function and loss: single-label multiclass classification, multi-label classification, regression bounded between 0 and 1, and regression with potential outliers. Justify each choice.
Why do sigmoid and tanh activations cause vanishing gradients in deep networks, and how do leaky-ReLU or GELU address this in hidden layers?
Compare MSE and MAE in terms of gradient behavior, sensitivity to outliers, and what each loss function is actually optimizing for statistically.
Contrast bagging and boosting in terms of bias and variance reduction, and explain when you'd prefer one over the other for noisy data.
Name two concrete, measurable ways to diagnose overfitting and two mitigation strategies that don't involve using validation data during training.

“This was the part I was least prepared for.”

View Post

Boston Consulting Group·Data Scientist·Online Assessment (OA)

May 2026

BCG's CodeSignal assessment for a Data Scientist role had a quick probability section that felt more like a stats refresher than anything tricky. Two questions, back to back, both classical.

You flip a fair coin three times. What is the probability of getting exactly two heads?
A bag has 4 red and 6 blue balls. If you draw two balls without replacement, what is the probability that both are red?

“Binomial setup, pretty clean.”

View Post

Boston Consulting Group·Data Scientist·Technical Phone Screen

May 2026

BCG data scientist interview that was basically one long pandas engineering problem with three sub-questions buried inside it. Heavier on implementation detail than I expected for a consulting firm.

Write a robust pandas function that merges seven dataframes with inconsistent column naming and whitespace into a single denormalized analytics table with a fixed column schema, handling missing values through a tiered imputation strategy and preserving orders even when shipper or payment data is absent.
Walk through the specific pandas operations and index choices you would use to make the multi-table merge and aggregation pipeline efficient, targeting O(N log N) or better complexity.
For order_id 10, verify the expected values of total_amount, product_count, and category_list given the sample data, including how missing quantity and unit_price are handled.
Explain how the imputation strategy you described avoids data leakage if the dataset is later split by date for train/test modeling.

“This was a lot to hold in your head at once.”

View Post

Boston Consulting Group·AI Engineer·Technical Phone Screen

May 2026

BCG AI Engineer interview with a coding question on array parity patterns. Pretty focused on algorithmic thinking, nothing too wild but it required some careful edge case handling.

Given an integer array, count every contiguous subarray where adjacent elements strictly alternate between odd and even values. Subarrays of length 1 count too.

“The length-1 edge case is the kind of thing you skip over mentally and then get burned by.”

View Post