← Uber Interview Insights

Uber·Data Scientist·Technical Phone Screen·Senior

Senior
May 2026

Summary

Uber data scientist interview that went deep into experiment analysis. The whole session was basically one extended stats question about how you'd actually run and interpret an A/B test end to end, not just the setup but the messy aftermath.

Questions Asked (6)

Q1

Walk through how you would analyze an A/B test after it ends, specifically how you handle the difference between session-level and user-level analysis.

A/B Testing & ExperimentationProduct Analytics & Metrics
Author's notes

This tripped me up a bit because I started talking about sessions and they pushed back fast.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by outlining the end-to-end analysis process: validate data quality, compute both session-level and user-level metrics, and reconcile differences by explaining the unit of analysis and its implications. Emphasize that the choice depends on the business question and the randomization unit, and discuss how to handle variance and dependencies.

Pro tip: Always check whether the randomization unit matches the analysis unit; if not, use clustered standard errors or delta method to account for within-user correlation. Also, consider the novelty effect and long-term impact by looking at trends over the experiment period.

1. Validate experiment setup and data quality

Confirm the randomization unit (user or session), check for sample ratio mismatch (SRM), and ensure data completeness and consistency across groups.

2. Compute metrics at both levels

Calculate session-level and user-level metrics (e.g., conversion rate, revenue per session/user) and their variances, noting that session-level metrics treat sessions as independent while user-level metrics aggregate within users.

3. Reconcile differences and choose appropriate analysis

Explain why results may differ: session-level analysis can inflate significance due to within-user correlation. Use user-level analysis if randomization is at user level, or apply methods like clustered standard errors or bootstrap for session-level data.

4. Assess practical significance and business impact

Beyond statistical significance, evaluate effect sizes, confidence intervals, and impact on key business metrics. Consider segment analysis and guardrail metrics.

5. Communicate findings and recommendations

Summarize results clearly, highlighting the chosen unit of analysis and its rationale, and provide actionable next steps (e.g., rollout, iterate, or stop).

Key Points to Mention

  • Randomization unit vs. analysis unit: ensure alignment or adjust for clustering.
  • Within-user correlation: sessions from the same user are not independent, leading to underestimated variance in session-level analysis.
  • Sample ratio mismatch (SRM) check to validate experiment integrity.
  • Use of delta method or bootstrap to compute correct standard errors when analysis unit differs from randomization unit.
  • Consideration of novelty effects and long-term trends by analyzing metrics over time.
  • Business context: choose the unit that aligns with the decision (e.g., user-level for retention, session-level for engagement).

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

How would you handle ratio metrics and skewed outcomes in an experiment, and what methods would you use to get valid standard errors?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

Delta method versus Fieller for ratios, I knew delta method but blanked on Fieller by name.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining why ratio metrics (e.g., CTR, conversion rate) and skewed outcomes (e.g., revenue, session time) violate standard assumptions of normality and independence. Then outline a structured approach: define the metric, choose appropriate statistical methods (e.g., delta method, bootstrap, or regression adjustment), and validate assumptions. Emphasize the importance of robust standard errors and practical trade-offs.

Pro tip: Mention that Uber often deals with heavy-tailed metrics like trip fares or wait times, so using methods like the delta method or bootstrap with stratification can be crucial. Also, highlight the need to check for practical significance, not just statistical significance, given large sample sizes.

1. Define the metric and its distribution

Clarify whether the metric is a ratio (e.g., clicks/impressions) or a skewed continuous variable (e.g., revenue). Discuss its distributional properties and potential issues like zero-inflation or heavy tails.

2. Identify challenges for standard errors

Explain why naive methods (e.g., t-test on ratios) fail: ratio metrics have correlated numerator and denominator, and skewed data violate normality. This leads to biased or underestimated standard errors.

3. Choose appropriate methods

Describe methods like the delta method for ratios, bootstrap (especially stratified or clustered), or regression adjustment (e.g., CUPED). Mention when to use each based on sample size and computational constraints.

4. Validate and diagnose

Discuss checking assumptions via simulation or diagnostics (e.g., QQ plots, bootstrap distributions). Consider sensitivity analyses to ensure robustness.

5. Interpret results with business context

Emphasize that statistical significance must be paired with practical significance. Discuss how to communicate uncertainty and make decisions in the presence of skewed outcomes.

Key Points to Mention

  • Delta method for ratio metrics: approximates variance using Taylor series expansion.
  • Bootstrap resampling: non-parametric approach that handles skewness and complex dependencies.
  • Regression adjustment (e.g., CUPED): reduces variance by using pre-experiment covariates.
  • Clustered standard errors: account for correlated data within users or sessions.
  • Practical significance vs. statistical significance: large samples can make tiny effects significant.
  • Stratification and blocking: can improve precision and reduce skewness impact.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Some users in your experiment never actually engaged with the treatment. How do you estimate the effect on users who actually received it, and what's the statistical framework for that?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

Two-stage least squares using assignment as the instrument.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the distinction between intent-to-treat (ITT) and local average treatment effect (LATE), then explain that non-compliance requires instrumental variables (IV) or CACE analysis. Describe how random assignment serves as an instrument for actual treatment receipt, and outline the assumptions and estimation steps.

Pro tip: Mention that you would first check the first-stage F-statistic to ensure the instrument is strong, and discuss the exclusion restriction assumption—showing awareness of its untestable nature and how to argue for its plausibility in an experiment context.

1. Define the causal estimand

Clarify that the goal is to estimate the effect on compliers (those who engage when assigned to treatment), known as the Complier Average Causal Effect (CACE) or Local Average Treatment Effect (LATE).

2. Set up the instrumental variables framework

Use random assignment as an instrument for actual treatment receipt. The IV estimate is the ratio of the ITT effect on the outcome to the ITT effect on treatment uptake.

3. State and assess assumptions

Discuss the four IV assumptions: relevance (first-stage), exclusion restriction, independence, and monotonicity. Check the first-stage F-statistic and argue for the plausibility of the exclusion restriction.

4. Estimate and interpret

Compute the CACE using two-stage least squares (2SLS) or the Wald estimator. Interpret the result as the effect for compliers, noting it does not generalize to always-takers or never-takers.

5. Consider alternatives and sensitivity

Mention alternative approaches like per-protocol analysis (with caveats) or principal stratification. Discuss sensitivity analyses for violations of assumptions.

Key Points to Mention

  • Intent-to-treat (ITT) vs. local average treatment effect (LATE) / complier average causal effect (CACE)
  • Instrumental variables (IV) and two-stage least squares (2SLS) estimation
  • The four IV assumptions: relevance, exclusion restriction, independence, monotonicity
  • First-stage F-statistic to test instrument strength
  • One-sided non-compliance vs. two-sided non-compliance and the role of monotonicity
  • Limitations: LATE applies only to compliers, not the full population

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q4

Your experiment returns a large p-value. How do you decide whether that means there's no effect, and what tests would you run?

A/B Testing & ExperimentationAdaptability & Ambiguity
Author's notes

You can't just say 'no effect,' that's the classic mistake.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying that a large p-value does not prove the null hypothesis; it only indicates insufficient evidence against it. Then outline a systematic approach to assess whether the experiment was adequately powered, check for practical significance, and run additional tests to rule out alternative explanations like novelty effects or segment-specific impacts.

Pro tip: Emphasize that at Uber, where experiments run at massive scale, even tiny effects can be statistically significant, so a large p-value often signals a need to check for implementation issues or heterogeneous treatment effects rather than concluding 'no effect'.

1. Check experiment validity and power

Verify that the experiment was properly randomized, had sufficient sample size, and the intended treatment was actually delivered. Calculate the minimum detectable effect (MDE) to see if the experiment was powered to detect a meaningful effect.

2. Assess practical significance and confidence intervals

Look at the effect size and its confidence interval. A large p-value with a wide interval suggests uncertainty; a narrow interval around zero suggests the effect is likely negligible in practice.

3. Explore segment-level and heterogeneous effects

Run subgroup analyses (e.g., by city, user type, device) to see if the treatment had an effect in specific segments that is masked in the overall average. Use interaction tests to check for significant heterogeneity.

4. Test for novelty and primacy effects

Examine the treatment effect over time (e.g., daily or weekly) to detect if early spikes or dips are due to novelty or learning effects. If present, consider extending the experiment or analyzing later periods separately.

5. Consider alternative tests and sensitivity analyses

Apply non-parametric tests (e.g., Mann-Whitney U) if normality is questionable, or use bootstrapping for robust inference. Also check for metric sensitivity and consider multiple testing corrections if many metrics were analyzed.

Key Points to Mention

  • Difference between statistical significance and practical significance
  • Importance of power analysis and minimum detectable effect (MDE)
  • Confidence intervals and effect size interpretation
  • Heterogeneous treatment effects and subgroup analysis
  • Novelty and primacy effects in online experiments
  • Multiple testing corrections (e.g., Bonferroni, Benjamini-Hochberg) when analyzing many metrics

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q5

How would you use pre-experiment data to reduce variance in your experiment results?

A/B Testing & Experimentation
Author's notes

CUPED.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining that pre-experiment data can be used to create covariates that explain a portion of the outcome variance, thereby increasing the sensitivity of your experiment. Then describe specific techniques like CUPED, stratification, or regression adjustment, and emphasize how these methods reduce variance without introducing bias. Finally, discuss practical considerations such as choosing the right pre-period and validating assumptions.

Pro tip: Mention that using pre-experiment data is especially powerful when the pre-period is long enough to average out noise, but be careful not to include data that could be affected by the treatment (e.g., pre-period must end before experiment start). Also, highlight that CUPED is a standard at many tech companies, including Uber, so showing familiarity with it signals practical experience.

1. Identify the goal: variance reduction

Explain that the primary goal is to reduce the variance of the treatment effect estimate, which increases statistical power and allows detection of smaller effects.

2. Choose relevant pre-experiment metrics

Select pre-experiment covariates that are highly correlated with the outcome metric, such as the same metric measured before the experiment or related behavioral metrics.

3. Apply variance reduction techniques

Describe methods like CUPED (Controlled-experiment Using Pre-Experiment Data), stratification, or regression adjustment to remove the variance explained by pre-experiment data.

4. Validate and adjust

Check that the pre-experiment covariates are balanced across treatment and control groups, and ensure the adjustment does not introduce bias (e.g., by using only pre-treatment data).

5. Measure impact and iterate

Quantify the variance reduction achieved (e.g., via R-squared) and consider combining multiple covariates or using more advanced methods like ML-based adjustments.

Key Points to Mention

  • CUPED (Controlled-experiment Using Pre-Experiment Data) as a standard technique
  • The importance of using pre-treatment data only to avoid bias
  • How variance reduction increases statistical power and reduces required sample size
  • Stratification or regression adjustment as alternative methods
  • The need to validate that covariates are balanced and not affected by the treatment
  • Practical considerations: choosing the pre-period length, handling missing data, and avoiding overfitting

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q6

If you want to look at heterogeneous treatment effects across subgroups, how do you control for multiple testing?

A/B Testing & ExperimentationProduct Analytics & Metrics
Author's notes

Short answer: Bonferroni if you're being conservative, Benjamini-Hochberg if you care more about false discovery rate than family-wise error.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging the multiple testing problem when analyzing heterogeneous treatment effects across many subgroups. Then describe a structured approach: pre-specify hypotheses, use appropriate corrections (e.g., Bonferroni, Benjamini-Hochberg), and consider hierarchical models or false discovery rate control. Emphasize balancing statistical rigor with practical insights, especially in a product analytics context.

Pro tip: At Uber, where experiments run at scale, it's crucial to distinguish between exploratory and confirmatory subgroup analyses. Pre-register your subgroups and use FDR control for exploratory analyses to avoid false positives while still uncovering actionable insights.

1. Define subgroups and hypotheses

Clearly specify the subgroups you're interested in and whether the analysis is exploratory or confirmatory. Pre-registering subgroups reduces the risk of data dredging.

2. Choose a multiple testing correction method

Select an appropriate method based on the number of tests and desired control: Bonferroni for strict family-wise error rate (FWER), or Benjamini-Hochberg for false discovery rate (FDR) when many tests are involved.

3. Apply correction and interpret results

Adjust p-values or use a hierarchical model to shrink estimates. Interpret the results in the context of the business, focusing on effect sizes and confidence intervals, not just statistical significance.

4. Validate findings

If possible, validate significant subgroup effects in a holdout set or a follow-up experiment to ensure they are not spurious.

Key Points to Mention

  • Multiple testing problem: increased risk of false positives when testing many subgroups.
  • Family-wise error rate (FWER) vs. false discovery rate (FDR) and when to use each.
  • Bonferroni correction: simple but conservative, especially with many tests.
  • Benjamini-Hochberg procedure: controls FDR, more powerful for large-scale testing.
  • Hierarchical models or Bayesian shrinkage to borrow strength across subgroups.
  • Pre-registration of subgroups and distinguishing exploratory vs. confirmatory analyses.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.