Stripe Data Scientist interview questions

20 Data Scientists have shared notes from their Stripe interviews, with the questions they were asked in each round. That is 6% of the 342 Stripe interview notes on Screna AI.

20 Data Scientist interview notes

Stripe·Data Scientist·Technical Phone Screen

Jul 2026
Stripe DS interview with a meaty marketing attribution case. Five sub-parts, all SQL-heavy, and they really wanted you to think about metric design before touching any code. Felt more like a take-home squeezed into a live session.
  • Define a primary success metric for the marketing campaign that is attributable, time-bounded, and robust to activity spikes. Include two guardrail metrics and write out the precise formulas.
  • Write SQL to compute, for each contact week and country, the 14-day post-contact purchase conversion rate and average revenue per contacted user. Use only the first contact per user and exclude purchases that happened before the contact time.
  • Write SQL to build a matched baseline: pair each contacted user to one non-contacted user in the same signup week and country, breaking ties by smallest user_id, then compute the 14-day purchase rate for the matched group.
  • On a specific date, contacted-user conversion in the US drops 20% versus the prior 7-day average. Write SQL to break this down by device and product_id, showing count contacted, 14-day conversion, and delta versus the prior period for each slice. Return the top 3 slices by approximate contribution to the drop.
  • What check would you add to make sure your metrics aren't biased by users who unsubscribed immediately after being contacted?

“I went with 14-day post-contact purchase conversion using only first contacts, which felt right.”

View Post

Stripe·Data Scientist·Technical Phone Screen

Jul 2026
Stripe DS interview with a meaty experiment design question built around a merchant subscription model. The whole thing was one big multi-part case and it went deep fast.
  • You built a ranked list of merchants likely to adopt Subscription. Design a randomized experiment to test whether targeting the top 50,000 with a sales or email nudge actually drives conversions. Walk through the RCT setup including control/treatment split, stratification by score decile and country, guardrails, and a staged ramp.
  • Compute the required per-arm sample size to achieve 80% power at a two-sided alpha of 0.05, given a baseline conversion of 2% and a minimum detectable uplift of 50% relative. Show the formula and the actual number.
  • Given a $5 outreach cost per merchant and an expected $120 gross margin over 90 days from a converted merchant, what precision threshold does the model need to break even on targeting?
  • What contamination risks exist when some customers transact across multiple merchants in the experiment? How do you handle merchants already piloting Subscription, and how do you attribute conversions when a merchant gets multiple outreach touches?

“I started with the split and stratification, which felt solid, but I fumbled a bit on the guardrails.”

View Post

Stripe·Data Scientist·Onsite - Behavioral / Leadership

Jul 2026
Behavioral round at Stripe for a Data Scientist role, basically one meaty question about disagreeing with your manager. Not a lot of fluff, they wanted a real story with stakes.
  • Tell me about a time you disagreed with your manager's decision on a project. How did you handle it, and what happened after?

“The part I underestimated was the 'what happened after' piece.”

View Post

Stripe·Data Scientist·Technical Phone Screen

Jul 2026
Stripe DS interview that went deep into dispute rate diagnostics. The whole session was basically one extended case study on a payments metric gone wrong, with real numbers to crunch and no hand-holding on methodology.
  • Given a weekly dispute rate time series that spikes then partially reverts, how would you determine whether the change is a real signal or just noise?
  • Using the provided counts (Week 34: 800 disputes out of 100,000 payments; Week 35: 1,400 out of 110,000), compute a two-sided test for the difference in proportions and a 99% confidence interval.
  • Given per-country breakdowns showing US dispute rate rose while EU dispute rate actually improved, compute the mix-adjusted overall change and explain why the aggregate rate still went up (Simpson's paradox).
  • You need to monitor 200 country-by-industry segment pairs weekly for dispute rate changes. How do you handle multiple testing, and how do you combine statistical significance with practical effect size?
  • For segments with fewer than 5,000 succeeded payments, how would you handle dispute rate estimation, and how would you report the uncertainty?

“I started with the obvious stuff, variance in small counts, week-over-week volatility, but they pushed me to be more precise about what 'real' even means here.”

View Post

Stripe·Data Scientist·Onsite - Cross-functional / Panel

Jun 2026
Stripe data scientist panel, heavy on cross-functional conflict scenarios tied to a payments analytics context. The whole thing felt like a leadership debrief more than a technical screen, lots of 'draft the actual message' and 'what artifact do you publish' type prompts.
  • Walk through a specific situation where you had competing priorities from Risk, Sales, and Engineering on a payments analytics project. Use concrete dates and metrics to structure your answer.
  • Given a one-week planning window, show your prioritization framework, name three things you're committing to and two you're explicitly dropping, and explain the trade-offs and expected impact of each.
  • Write out two actual messages verbatim: one where you push back on an executive request that would compromise a guardrail, and one where you bring a skeptical engineer onside. Also describe your escalation path if you don't reach agreement within 24 hours.
  • A manager from an adjacent team challenges your methodology in a design review. How do you de-escalate, actively look for evidence that you might be wrong, and ultimately make a call? What do you publish afterward and how do you measure whether the resolution actually worked?
  • Looking back at the project you described, what's one thing you'd do differently and what specific measurable outcome would that change have produced?

“This is the one I over-prepared for and still fumbled the metrics part.”

View Post

Stripe·Data Scientist·Technical Phone Screen

Jun 2026
Stripe DS interview with a single deep-dive case on predicting merchant subscription adoption. The whole thing was one long multi-part ML design question and it went places I did not fully anticipate.
  • Design a production-ready classification system to predict which non-subscription merchants will adopt Stripe's Subscription product within the next 60 days, covering labeling strategy, leakage prevention, feature engineering, modeling choices, evaluation, and post-deployment monitoring.

“This was basically six questions wrapped in one and I did not realize how deep it went until I was already three sub-parts in.”

View Post

Stripe·Data Scientist·Onsite - Behavioral / Leadership

Jun 2026
Stripe data scientist interview with a pretty intense behavioral/project question that required way more prep than I gave it credit for. The structure they wanted was exhaustive and I definitely undersold the quantitative side of my story.
  • Walk me through a project you led that measurably moved a core business metric. Cover the context, your specific goal and constraints, the three most consequential decisions you made including trade-offs and stakeholders across regions, one intentional risk you took, and before/after numbers with dollar impact and how you validated attribution. Then tell me what you'd do differently to double the impact in the same timeframe, and what you'd cut if you had to ship in three hours.

“This question is basically four questions stapled together and I did not pace myself well.”

View Post

Stripe·Data Scientist·Technical Phone Screen

Jun 2026
Stripe DS interview, technical screen format, one big SQL problem that looked manageable until it really wasn't. The kind of question where you think you understand it and then re-read it three times and realize you missed half the requirements.
  • Given a merchant table and a transaction table, write a single SQL query to find the top 10 merchants who have no succeeded Subscription transactions in the last 180 days but show recurring payment behavior. Define a recurring customer as someone with at least two succeeded payments to the same merchant in that window, same amount and card fingerprint, with gaps between 28 and 35 days. Output includes merchant ID, recurring customer count, a repeat transaction rate for the last 30 days, first seen date, and a flag for whether they currently use subscriptions. Count each customer at most once, use window functions where appropriate.

“This wrecked me a little.”

View Post

Stripe·Data Scientist·Take-home Assignment

Jun 2026
Stripe take-home for a DS role. One week to build a purchase propensity model end-to-end, slides plus runnable code. The prompt was dense and clearly designed to see if you actually know ML in production, not just theory.
  • How would you define the prediction target and snapshot time to avoid label leakage, including late-arriving events?
  • What baseline and main model would you propose for purchase propensity, and how do you justify the choice between logistic regression and gradient-boosted trees?
  • Which metric would you optimize under class imbalance, PR-AUC or ROC-AUC, and what business metric would you report on the slides?
  • Describe a temporal cross-validation scheme that gives an honest performance estimate and explain how you'd tune hyperparameters quickly within the one-week time budget.
  • How would you detect and mitigate data leakage, target leakage, and train-test contamination? Give at least two concrete checks you would code.
  • What is your calibration plan, how would you select a threshold for an email targeting use-case with a cost per send, and how do you communicate this on slides?
  • What would your ablation plan look like if time runs short, and what are the 5 to 7 slide headlines that tell a compelling story to a hiring manager?

“This is the part I almost got wrong.”

View Post

Stripe·Data Scientist·Technical Phone Screen

Jun 2026
Stripe data scientist interview, heavy on experimentation stats. Five questions back to back, all quantitative, no fluff. The kind of loop where you either know your sample size formulas or you don't.
  • For a two-arm conversion experiment with a baseline rate of 12% and a target absolute uplift of 0.5 percentage points (two-sided, alpha 0.05, power 80%), derive the required per-variant sample size using the normal approximation for a difference in proportions. Show the formula and walk through the calculation.
  • You ran the experiment for a week. Control had 150k users with 18,000 conversions, treatment had 150k users with 18,900 conversions. Calculate the point estimate, a 95% confidence interval, and a p-value. Then explain whether the result is statistically significant, and whether it's actually meaningful for the business.
  • Suppose you apply CUPED using a pre-experiment covariate that explains 30% of the variance in the outcome (R-squared = 0.30). How does this change the effective variance, and what is the revised minimum detectable effect? Show the math.
  • You're monitoring four guardrail metrics with unadjusted p-values of 0.03, 0.01, 0.20, and 0.04. Apply Holm-Bonferroni correction at a familywise error rate of 5% and state which guardrails are still significant. Show the ordering and the adjusted thresholds at each step.
  • How would you detect and correct for overdispersion or miscalibration in conversion rate estimates when users are clustered by geographic region?

“I knew the formula but fumbled the pooled variance piece.”

View Post