← Capital One Interview Insights

Capital One·Data Scientist·Technical Phone Screen·Senior

Senior
May 2026

Summary

Capital One data scientist interview that went deep into experiment design for a small business credit pilot. The whole thing was basically one long case question that kept branching into more sub-questions every time I thought I'd landed somewhere safe.

Questions Asked (8)

Q1

How would you design a rigorous A/B test for a $100 credit pilot targeting new small-business accounts, including randomization unit, stratification, and how to handle day-of-week and regional seasonality?

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

This is where I spent most of my time and still felt like I left stuff on the table.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the business goal and defining a clear success metric (e.g., activation rate or spend lift). Then outline the experimental design: randomization at the account level, stratification by key covariates like region and business size, and methods to handle day-of-week and regional seasonality such as blocking or time-based randomization. Finally, discuss analysis considerations like power, multiple testing, and heterogeneous treatment effects.

Pro tip: Emphasize that with a $100 credit pilot, the treatment effect may be small and short-lived, so you need to pre-register the primary metric and duration to avoid p-hacking. Also, consider using a holdout group to measure long-term effects and ensure the credit isn't cannibalizing organic spend.

1. Define Objective and Metrics

Clarify the pilot's goal (e.g., increase new small-business account activation, spend, or retention) and select a primary metric (e.g., 30-day activation rate) plus guardrail metrics (e.g., credit abuse, support calls).

2. Choose Randomization Unit and Stratification

Randomize at the account level to avoid contamination, and stratify by region, business size, and acquisition channel to balance covariates and increase power.

3. Handle Seasonality and Time Effects

Use a randomized block design with time as a blocking factor (e.g., randomize within each day-of-week) or include day-of-week and region as covariates in the analysis. Consider running the test for full weeks to capture weekly cycles.

4. Determine Sample Size and Duration

Conduct power analysis to determine required sample size per arm, accounting for expected effect size, variance, and intra-cluster correlation if applicable. Set duration to cover at least one full business cycle (e.g., 4-6 weeks).

5. Analyze and Interpret Results

Use appropriate statistical tests (e.g., regression with fixed effects for strata and time) to estimate treatment effect, check for heterogeneous effects, and ensure results are robust to seasonality.

Key Points to Mention

  • Randomization unit: account-level to prevent spillover and ensure independence.
  • Stratification: by region, business size, and other relevant covariates to improve balance and precision.
  • Seasonality: use blocking or time-based randomization to control for day-of-week and regional patterns.
  • Power analysis: calculate sample size based on minimum detectable effect and variance.
  • Analysis: include covariates in regression models, check for interactions, and adjust for multiple comparisons.
  • Guardrail metrics: monitor for unintended consequences like credit abuse or cannibalization.

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

Q2

Walk through the sample size calculation for 80% power and alpha of 0.05 to detect a +2 percentage point lift over a 12% baseline activation rate, given a budget of 10,000 treatment accounts over 14 days.

A/B Testing & Experimentation
Author's notes

I actually felt okay here.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the test design (one-tailed vs two-tailed, equal allocation) and then compute the required sample size per group using the standard formula for comparing two proportions. After obtaining the required sample size, check whether the budget of 10,000 treatment accounts over 14 days is sufficient to achieve 80% power; if not, discuss options like extending the test duration or relaxing power.

Pro tip: Always state your assumptions explicitly (e.g., two-tailed test, equal variance, no novelty effect) and mention that in practice you would use a power analysis tool or simulation to validate the calculation. Also, consider practical constraints like daily traffic and whether the budget allows for the required sample size within the given timeframe.

1. Clarify the test parameters

Confirm the baseline rate (12%), minimum detectable effect (2 percentage points), significance level (alpha=0.05), power (80%), and allocation ratio (assume 1:1). Also determine if the test is one-tailed or two-tailed; typically two-tailed is used unless specified otherwise.

2. Compute required sample size per group

Use the formula for comparing two proportions: n = (Z_{α/2} + Z_β)^2 * (p1(1-p1) + p2(1-p2)) / (p2 - p1)^2, where p1=0.12, p2=0.14. Plug in Z_{α/2}=1.96 and Z_β=0.84 to get the required sample size per group.

3. Calculate total sample size and compare to budget

Multiply the per-group sample size by 2 to get the total required sample size. Compare this to the available 10,000 treatment accounts (and assuming equal control, total 20,000 accounts). If the required sample size exceeds the budget, the test is underpowered.

4. Assess feasibility and discuss trade-offs

If the budget is insufficient, discuss potential adjustments: increasing the MDE, extending the test duration to accumulate more accounts, or accepting lower power. Also consider if the 14-day period provides enough traffic to reach the required sample size.

Key Points to Mention

  • Use of the standard formula for two-proportion z-test and correct Z-values for alpha=0.05 (two-tailed) and power=80%.
  • Assumption of equal allocation and independence of observations.
  • Calculation of required sample size per group and total sample size.
  • Comparison with the available 10,000 treatment accounts and 14-day timeframe.
  • Discussion of practical implications: underpowered test, risk of false negatives, and potential remedies.
  • Mention of alternative approaches like simulation or sequential testing if applicable.

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

Q3

Define your ITT and TOT estimands for this experiment, and explain how you'd use pre-verification activity to reduce variance in the analysis.

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

ITT was easy.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clearly defining ITT and TOT in the context of the experiment, including how treatment assignment and actual treatment receipt are determined. Then explain how pre-verification activity (e.g., pre-experiment behavior) can be used as a covariate to reduce variance in the analysis, likely through CUPED or regression adjustment. Emphasize the importance of pre-registration and avoiding post-treatment bias.

Pro tip: Mention that pre-verification activity must be measured before randomization and be unaffected by the treatment to avoid bias. Also, note that while ITT is the gold standard for causal inference, TOT requires additional assumptions like exclusion restriction and monotonicity, which should be justified.

1. Define ITT and TOT

ITT is the effect of treatment assignment regardless of compliance; TOT is the effect of actually receiving the treatment. Clearly state the estimands in the context of the experiment.

2. Identify pre-verification activity

Select a pre-experiment metric (e.g., user activity before randomization) that is correlated with the outcome and unaffected by treatment.

3. Explain variance reduction technique

Describe how to use pre-verification activity as a covariate in regression adjustment or CUPED to reduce variance and increase power.

4. Address assumptions and limitations

Discuss assumptions for TOT (e.g., exclusion restriction, monotonicity) and ensure pre-verification activity is pre-treatment to avoid bias.

5. Conclude with practical implications

Summarize how this approach leads to more precise estimates and better decision-making, aligning with business goals.

Key Points to Mention

  • ITT measures the effect of assignment, preserving randomization; TOT measures the effect of treatment on compliers.
  • Pre-verification activity must be measured before randomization and be prognostic of the outcome.
  • CUPED (Controlled-experiment Using Pre-Experiment Data) or regression adjustment can reduce variance by 50% or more.
  • TOT requires instrumental variable assumptions: exclusion restriction and monotonicity.
  • Always pre-register the analysis plan to avoid p-hacking and ensure validity.
  • Variance reduction increases power, allowing detection of smaller effects or smaller sample sizes.

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

Q4

How would you set up guardrail monitoring and kill-switch thresholds for metrics like refund rate and support tickets during the pilot?

A/B Testing & ExperimentationProduct Analytics & MetricsStakeholder Management
Author's notes

I talked about sequential testing for the guardrails and setting absolute thresholds before launch rather than reacting ad hoc.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the guardrail metrics and their acceptable thresholds based on historical baselines and business impact. Then outline a monitoring system that tracks these metrics in real-time with automated alerts and a kill-switch mechanism. Emphasize collaboration with stakeholders to align on thresholds and escalation procedures.

Pro tip: Set thresholds slightly above the natural variance to avoid false alarms, but low enough to catch meaningful regressions. Also, pre-define the kill-switch criteria and automate the process to minimize manual intervention.

1. Define guardrail metrics and thresholds

Identify key metrics like refund rate and support tickets, and establish acceptable thresholds using historical data and business impact analysis.

2. Set up real-time monitoring

Implement dashboards and alerting systems to track these metrics continuously during the pilot, ensuring data quality and timely detection.

3. Establish kill-switch criteria and automation

Define specific conditions that trigger the kill-switch (e.g., metric exceeds threshold for a sustained period) and automate the shutdown to prevent further harm.

4. Align with stakeholders and document process

Collaborate with product, engineering, and business teams to agree on thresholds, escalation paths, and communication plans.

5. Review and iterate

After the pilot, analyze the effectiveness of the guardrails and adjust thresholds and processes for future experiments.

Key Points to Mention

  • Use statistical methods (e.g., confidence intervals, control charts) to set thresholds that account for natural variance.
  • Consider both absolute and relative thresholds, and account for seasonality or external factors.
  • Automate alerts and kill-switch to reduce response time and human error.
  • Define clear ownership and escalation paths for when thresholds are breached.
  • Document the rationale for thresholds and get stakeholder buy-in to ensure alignment.
  • Plan for post-pilot analysis to refine guardrails for future experiments.

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

Q5

What's your decision rule for this test, including whether to use a one-sided or two-sided test, how you'd handle multiplicity, and what superiority margin you'd pre-specify?

A/B Testing & Experimentation
Author's notes

Went one-sided since we care about improvement, not harm detection (guardrails cover the harm side).

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by framing the decision rule as a pre-registered plan tied to the business objective and risk tolerance. Then walk through the choice of one- vs two-sided test, multiplicity control, and superiority margin, justifying each with statistical and practical reasoning. Emphasize that all choices are made before data collection to avoid p-hacking and ensure valid inference.

Pro tip: In fintech, regulators and stakeholders care about false positives (e.g., launching a harmful feature) more than false negatives, so default to a two-sided test unless there's a strong directional hypothesis and a clear cost asymmetry. Also, pre-specify the superiority margin based on the minimum effect size that would justify the implementation cost, not just statistical significance.

1. Define the business objective and hypotheses

Clarify the metric (e.g., conversion rate, revenue per user) and whether the goal is to detect any difference or only improvement. State the null and alternative hypotheses explicitly.

2. Choose one-sided vs two-sided test

Use a two-sided test by default to guard against unexpected negative effects. Opt for one-sided only if there is a strong prior and the cost of missing a positive effect far outweighs the risk of a false positive.

3. Pre-specify the superiority margin

Set a minimum detectable effect (MDE) that is practically meaningful, based on business impact (e.g., ROI, cost savings). This margin defines the threshold for superiority and informs sample size calculation.

4. Plan for multiplicity

If testing multiple metrics or variants, control the family-wise error rate (FWER) using Bonferroni or Holm-Bonferroni for a few comparisons, or the false discovery rate (FDR) (Benjamini-Hochberg) for many. Pre-register which metrics are primary vs secondary.

5. State the decision rule

Conclude with a clear rule: reject the null if the p-value < alpha (adjusted for multiplicity) and the effect size exceeds the superiority margin. Otherwise, fail to reject and consider practical implications.

Key Points to Mention

  • Two-sided test is conservative and protects against negative effects; one-sided only with strong directional hypothesis and cost asymmetry.
  • Superiority margin should be based on business relevance (e.g., minimum ROI threshold), not just statistical significance.
  • Multiplicity adjustments: Bonferroni/Holm for FWER control with few comparisons; Benjamini-Hochberg for FDR control with many metrics.
  • Pre-registration of the analysis plan prevents p-hacking and ensures valid inference.
  • Sample size and power calculations depend on the chosen alpha, margin, and variance.
  • Consider sequential testing or alpha-spending if interim analyses are planned.

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

Q6

How would you check data quality issues like event lag, duplicate accounts, and bot activity before analyzing the results?

A/B Testing & ExperimentationRoot Cause Analysis
Author's notes

Pretty standard data integrity stuff.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by outlining a systematic pre-analysis validation process that covers each data quality dimension: completeness, consistency, and validity. Then, for each issue (event lag, duplicate accounts, bot activity), describe specific detection methods and how you would mitigate their impact on the analysis. Emphasize the importance of documenting and automating these checks to ensure reproducibility and scalability.

Pro tip: Frame your answer around a 'data quality gate' that must be passed before any analysis, and mention that you would quantify the potential bias introduced by each issue to decide whether to exclude, adjust, or flag affected data.

1. Define data quality dimensions and metrics

Identify the key dimensions relevant to the experiment (e.g., timeliness, uniqueness, authenticity) and define quantifiable metrics for each (e.g., event lag in seconds, duplicate rate, bot score distribution).

2. Detect event lag

Compare event timestamps with server logs or expected ingestion times to measure lag; check for patterns (e.g., by platform, region) and set thresholds to flag or exclude delayed events.

3. Identify duplicate accounts

Use deterministic (e.g., email, device ID) and probabilistic (e.g., behavioral fingerprinting) methods to find duplicates; analyze their impact on metrics and decide on deduplication or exclusion.

4. Detect bot activity

Apply heuristics (e.g., unusually high activity, uniform timing) and machine learning models (e.g., isolation forests) to flag bots; validate with manual review and exclude or segment them.

5. Mitigate and document

Implement data cleaning steps, quantify bias, and document all checks and decisions in a reproducible pipeline; monitor data quality continuously.

Key Points to Mention

  • Event lag: timestamp validation, time zone issues, and impact on conversion windows
  • Duplicate accounts: deterministic vs. probabilistic matching, and effect on user-level metrics
  • Bot activity: behavioral signals, rate limiting, and use of bot detection services
  • Data quality dimensions: completeness, consistency, timeliness, validity, uniqueness
  • Automation and reproducibility: using tools like Great Expectations or custom scripts
  • Bias quantification: sensitivity analysis to assess how data issues affect experiment results

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

Q7

How would you measure heterogeneity of treatment effects across industry, region, and spend propensity segments?

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

Subgroup analysis with interaction terms in a regression, pre-specified segments only to avoid p-hacking, and I mentioned that spend propensity could be estimated from pre-signup signals as a continuous moderator.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by framing heterogeneity as effect modification and the need to test whether treatment effects vary across pre-specified segments. Describe a hierarchical modeling approach that estimates segment-level effects while borrowing strength across segments, and emphasize validation and business implications. Conclude with how you would prioritize segments for targeting or further experimentation.

Pro tip: Pre-register your segments and interactions to avoid data dredging, and use a holdout set or cross-validation to confirm that any heterogeneity you find is reproducible. Also, consider practical significance: a statistically significant interaction may not be worth acting on if the effect size difference is small.

1. Define segments and hypotheses

Clearly specify the industry, region, and spend propensity segments based on business relevance and data availability. Pre-register which interactions you will test to avoid false discoveries.

2. Choose modeling approach

Use a hierarchical (mixed-effects) model with random slopes for treatment by segment, or a Bayesian model that allows partial pooling. This estimates segment-specific treatment effects while controlling for multiple comparisons.

3. Estimate and test heterogeneity

Compute the treatment effect for each segment and test the interaction terms (e.g., treatment × segment) using appropriate corrections (e.g., Bonferroni, FDR). Report credible intervals or confidence intervals for each segment effect.

4. Validate and check robustness

Use cross-validation or a holdout set to confirm that detected heterogeneity replicates. Check sensitivity to model specifications and outliers.

5. Interpret and act

Assess practical significance: are the differences large enough to warrant targeted strategies? Prioritize segments with the largest uplift and consider follow-up experiments to confirm.

Key Points to Mention

  • Interaction terms (treatment × segment) and the difference between overall and segment-level effects
  • Hierarchical/mixed-effects models or Bayesian partial pooling to handle multiple comparisons and small sample sizes
  • Multiple testing corrections (e.g., Bonferroni, Benjamini-Hochberg) when testing many segments
  • Power analysis for subgroup detection and the risk of false negatives
  • Cross-validation or holdout validation to ensure reproducibility of heterogeneity findings
  • Business implications: targeting, personalization, and resource allocation based on segment-level uplift

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

Q8

Given an observed lift of +1.6pp with a 95% CI of [+0.2pp, +3.0pp] and a 12% increase in support tickets, how would you frame your rollout recommendation?

A/B Testing & ExperimentationProduct StrategyStakeholder Management
Author's notes

This was the closer and it's genuinely a hard call.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging the positive lift but immediately address the wide confidence interval and the support ticket increase as risks. Frame the recommendation as a phased rollout with guardrails, emphasizing the need to balance statistical significance with operational impact. Propose next steps to investigate the ticket increase and potentially segment the analysis.

Pro tip: Quantify the trade-off: estimate the cost of support tickets and compare it to the expected revenue lift to show business acumen. Also, mention that the lower bound of the CI is close to zero, so the effect might be smaller than observed.

1. Summarize the results

State the observed lift and confidence interval, noting that the result is statistically significant but the interval is wide, indicating uncertainty. Also, highlight the 12% increase in support tickets as a potential negative side effect.

2. Assess business impact

Estimate the financial impact of the lift (e.g., revenue increase) and the cost of the additional support tickets. Consider whether the net effect is positive and aligns with business goals.

3. Evaluate risks and uncertainty

Discuss the wide confidence interval and the possibility that the true lift could be as low as 0.2pp. Also, consider the root cause of the support tickets and whether they are sustainable or can be mitigated.

4. Propose a phased rollout

Recommend a limited rollout to a larger segment while monitoring support tickets and other guardrail metrics. Suggest setting up a dashboard and alerts for key metrics.

5. Define next steps

Outline further analysis: segment the results by user cohorts, investigate the support ticket reasons, and potentially run a follow-up experiment to confirm the effect and optimize the feature.

Key Points to Mention

  • Statistical significance vs. practical significance: the lower bound of the CI is close to zero, so the effect might be marginal.
  • Cost-benefit analysis: quantify the support ticket increase in monetary terms and compare to the lift.
  • Guardrail metrics: support tickets are a guardrail; consider other metrics like customer satisfaction or churn.
  • Phased rollout: start with a small expansion, monitor closely, and have a rollback plan.
  • Root cause analysis: understand why support tickets increased and whether it's fixable.
  • Stakeholder communication: align with product, engineering, and customer support teams on the decision.

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