CVS Health Interview Questions — Page 2

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview with a health insurance analytics SQL problem. Pretty standard stuff but the multi-part structure kept me on my toes.

Given a membership table and a claims table, write a SQL query to calculate total paid claim amounts in 2024 broken down by age band and gender.
For a given age band, show the year-over-year trend of total paid claim amounts across all years in the data.

“Straightforward join on member id, filter claim_date to 2024, then group by age_band and gender.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview, technical round, one big coding question that was essentially a full ML pipeline build. Pretty involved for a single prompt but it covered a lot of ground at once.

Build a reproducible sklearn classification pipeline in Python to predict whether a user will subscribe in the next 30 days. The dataset has user and behavioral features including country, device type, session counts, and purchase history. Requirements include: temporal train/validation split, a ColumnTransformer with separate numeric and categorical preprocessing pipelines, logistic regression with class balancing and hyperparameter search over C using stratified cross-validation, ROC AUC and PR AUC evaluation plus threshold tuning for F1, probability calibration on training data only, permutation feature importance on the validation set, and an explanation of one target leakage risk in the schema.

“This was a lot.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview that went pretty deep into statistical testing. Two-part question covering p-value communication and nonparametric vs parametric test selection. More conceptual rigor than I expected for a DS role at a pharmacy chain.

Explain what a p-value means to someone without a statistics background. Use a concrete everyday analogy and address common misinterpretations like confusing it with effect size or the probability that the null hypothesis is true.
You have 12 patients with blood pressure measured before and after a low-sodium diet. The distribution of paired differences is skewed with outliers. Do you use a paired t-test or a Wilcoxon signed-rank test? Walk through your assumptions, how you'd check them, how you handle ties and zeros, and what effect size you'd report.
You're comparing hospital length of stay between two independent clinics with unequal variances, heavy-tailed distributions, and sample sizes of 18 and 25. Compare Welch's t-test to the Wilcoxon rank-sum test. What does each actually estimate, when would you prefer one over the other, and how would you complement the result with confidence intervals and a robust effect size?

“I fumbled the analogy part more than I should have.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview with a pretty involved technical exercise covering SQL and pandas. The questions were specific enough that I had to actually think rather than just recite patterns. No fluff, just schema and specs.

Given orders, order_items, and products tables, compute the percentage of distinct orders in calendar year 2024 that contained at least one Subscription-category product. Each order should be counted only once regardless of how many subscription items it has. Return a single row with the result rounded to two decimal places.
Using the same schema, compute year-over-year order count change broken down by user location and age group (18-29, 30-44, 45+). Return orders in 2023, orders in 2024, and the YoY percent change. If 2023 count is zero, return NULL instead of dividing by zero.
Using pandas DataFrames with orders, products, and users data, find the number of unique users in 2024 who bought any product with 'Pro' in the name (case-insensitive). Return a DataFrame with location, age group, and unique user count, sorted by unique users descending then location ascending.

“The deduplication part is what gets you.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

Brutal technical screen for a DS role at CVS Health. The whole thing was basically one massive multi-part prompt about designing a flu vaccination email experiment from scratch, and they wanted you to go deep on every single layer of it.

Design an end-to-end email campaign experiment to increase verified flu vaccinations within 30 days among eligible members. Walk through targeting and exclusion rules, randomization unit, holdout structure, primary and guardrail metrics with ground-truth outcome sourcing, power analysis for a 2% baseline with 20% relative uplift at alpha=0.05 and 80% power, frequency capping and creative testing, seasonality controls, HIPAA and opt-out compliance implications, and your full analysis plan including a fallback causal method if randomization breaks.

“This was basically nine questions stuffed into one.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview that was essentially one long end-to-end experiment design case for a pharmacy flu vaccination campaign. Heavy on stats and causal inference, way more involved than I expected for a single session.

You're running a pharmacy flu vaccination campaign over SMS and Email. Would you design this as a simple A/B test or a 2x2 factorial, and why? Walk through your reasoning on interaction effects, interference, and send costs.
Define the eligibility criteria, exclusions, randomization unit, and stratification variables for this experiment.
Compute the per-arm sample size for this experiment given an 8% baseline vaccination rate, 1.5 percentage point MDE, two-sided alpha of 0.05, and 80% power. Then discuss how household clustering with ICC of 0.01 and average household size of 1.3 inflates that number.
Define the ITT estimator for this campaign. Then use deliverability as an instrument to compute the TOT, show the ITT/compliance relationship, and state the assumptions required.
In a factorial design where customers can receive both SMS and Email, how would you instrument and analyze the data to attribute incremental impact to each channel? Why is self-reported attribution biased, and is there any valid use for it?
In one market, 100 customers in the SMS arm got vaccinated but only 50 say they came because of a message. The control vaccination rate is 7.5% and the SMS arm rate is 9.0%. Compute the ITT lift and explain why the self-report number doesn't change your causal estimate. What would you test next if lift falls below 1.5 percentage points?
How would you monitor the experiment during rollout, finalize results, and recommend a scaled channel policy under a fixed budget?

“I went straight to factorial and felt pretty good about it, but I fumbled the interference piece.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview, technical round focused entirely on SQL. One long multi-part problem covering deduplication, window functions, and distribution bucketing. Pretty dense for a single question but it covered a lot of ground.

You have two tables with intentional duplicates. Write SQL to: (a) identify duplicate user_ids and produce a deduplicated users table using window functions, picking the canonical row by earliest signup date then alphabetically by name; (b) find users with at least 2 purchases after deduplicating orders so each order_id counts once; (c) build a distribution table of purchase counts per user including zeros, with each bucket's percentage of total unique users.

“This was one question but it felt like three separate interviews stapled together.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist interview with a meaty causal inference case study around flu-shot campaign targeting. The whole thing was essentially one long open-ended problem covering modeling, evaluation, policy design, and online testing. Pretty intense for a single session.

You have historical campaign logs with randomized holdouts. How would you design a treatment effect modeling approach to decide who to contact via SMS or Email for an upcoming flu-shot campaign, given demographic features, past vaccination history, engagement signals, and known treatment assignment probabilities?
How would you evaluate uplift models offline? Walk through the metrics you'd use and how you'd factor in channel costs like SMS at $0.02 versus Email at $0.001.
Given a budget that allows contacting at most 40% of eligible customers, how would you rank customers and assign them to a channel?
How would you design an online validation test to compare model-based targeting against uniform random targeting? What success metrics and stopping rules would you define?
How would you detect segments where your model is actually causing harm, meaning negative uplift, and what would you do about them in the targeting policy?

“This is where I spent most of my mental energy.”

View Post

CVS Health·Data Scientist·Onsite - Behavioral / Leadership

Jun 2026

Interviewed for a Data Scientist role at CVS Health. The question was a layered behavioral prompt that also doubled as a strategy question, which I didn't fully see coming.

What are your top one or two strengths most relevant to this role, and can you walk me through a concrete example that shows measurable impact? Also, what trade-offs did you make, how did you bring other teams along, what would you do differently, and how would you apply those strengths in your first 90 days on our experimentation and measurement work?

“This is five questions stitched into one and I did not clock that fast enough.”

View Post

CVS Health·Data Scientist·Technical Phone Screen

Jun 2026

CVS Health data scientist technical screen, PostgreSQL heavy. Five distinct tasks covering schema design, data insertion, aggregation queries, date windowing, and a DML update. No behavioral stuff at all, just straight SQL the whole time.

Design a normalized e-commerce schema in PostgreSQL with products, customers, orders, and order_items tables. Use minimal correct types and include appropriate constraints like primary keys, foreign keys, NOT NULL, UNIQUE, and CHECK constraints.
Write a query to compute per-customer gross revenue for paid orders placed in August 2025. Include customers with zero revenue as 0, and sort results by revenue descending then customer_id ascending.
List all products that appear in non-refunded, non-cancelled orders placed within the last 7 days relative to a fixed reference date of September 1, 2025. Return the product id, name, and the earliest order date within that window.
Write an UPDATE statement that sets coupon_code to 'NONE' for all paid orders from August 2025 where coupon_code is currently NULL. Then write a SELECT to verify the count of affected rows.
Given ASCII table samples, write INSERT statements that exactly reproduce the provided data across all four tables, respecting foreign key relationships and insertion order.

“The 'minimal correct types' phrasing tripped me up a bit.”

View Post