← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

Senior
Apr 2026

Summary

Meta DS interview focused almost entirely on experiment design, and it went deep fast. The A/B testing questions were layered and connected, not just isolated prompts, so you had to keep your earlier answers consistent as new constraints got added.

Questions Asked (5)

Q1

Given a baseline conversion rate of 4%, a 10% relative uplift to detect, two-sided alpha of 0.05, and 80% power, compute the minimum per-variant sample size. Show the formula and the numeric result.

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

I knew the formula but fumbled the z-values from memory mid-call.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the parameters and the formula for sample size calculation for two proportions. Then compute the baseline and treatment conversion rates, plug into the formula, and present the numeric result clearly. Mention any assumptions and discuss practical implications.

Pro tip: Always state the formula and assumptions explicitly, and round up the sample size to ensure sufficient power. Also, mention that this is a simplified calculation and real-world adjustments (e.g., for multiple comparisons, sequential testing) may be needed.

1. Clarify parameters and assumptions

Confirm the baseline conversion rate (4%), relative uplift (10%), alpha (0.05 two-sided), and power (80%). State that the test is two-sided and assume equal variance for the sample size calculation.

2. Compute absolute effect size

Calculate the treatment conversion rate: 4% * (1 + 10%) = 4.4%. The absolute difference (delta) is 0.4 percentage points (0.004).

3. Apply sample size formula

Use the standard formula for two proportions: n = (Z_{α/2} + Z_{β})^2 * (p1*(1-p1) + p2*(1-p2)) / (p2 - p1)^2. Plug in Z_{α/2}=1.96, Z_{β}=0.84, p1=0.04, p2=0.044.

4. Compute numeric result

Calculate the numerator and denominator: (1.96+0.84)^2 = 7.84; p1*(1-p1)=0.0384, p2*(1-p2)=0.042064, sum=0.080464; delta^2=0.000016. So n = 7.84 * 0.080464 / 0.000016 ≈ 39,427 per variant. Round up to 39,428.

5. Interpret and discuss

State that approximately 39,428 users per variant are needed. Discuss that this is a large sample, and mention potential adjustments for real-world constraints (e.g., multiple metrics, sequential testing).

Key Points to Mention

  • Formula for sample size calculation for two proportions
  • Z-scores for alpha (1.96 for 95% confidence) and power (0.84 for 80% power)
  • Baseline and treatment conversion rates (4% and 4.4%)
  • Absolute vs relative uplift (10% relative = 0.4 percentage points absolute)
  • Assumption of equal variance and two-sided test
  • Rounding up to the next integer to ensure sufficient sample size

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

Q2

You now have two primary metrics: conversion rate and qualified-lead rate. Apply a Bonferroni correction to the sample size calculation, then explain the trade-offs of that approach versus using a gatekeeper metric or a hierarchical testing structure.

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

Bonferroni just divides alpha by 2, so you use 0.025 instead of 0.05, which bumps the z-value to about 2.24 and inflates sample size by maybe 30%.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, explain how to apply the Bonferroni correction to the sample size calculation for two primary metrics, then compare the trade-offs of Bonferroni against using a gatekeeper metric or hierarchical testing. Focus on statistical power, false positive control, and practical implications for decision-making.

Pro tip: Emphasize that Bonferroni is conservative and can inflate sample size requirements, so in practice, consider the cost of false positives versus false negatives and the business impact of each metric. Mention that hierarchical testing preserves power for the primary metric while still allowing secondary analyses.

1. Clarify the metrics and correction

Define conversion rate and qualified-lead rate as co-primary metrics. Explain that Bonferroni correction adjusts the significance level (α) by dividing it by the number of metrics (k=2), so α' = α/2.

2. Adjust sample size calculation

Show that the required sample size per variant increases because the effective α is smaller. Use the formula n ∝ (z_{1-α'/2} + z_{1-β})^2 / Δ^2, and note that reducing α increases the z-value, thus requiring a larger n.

3. Explain trade-offs of Bonferroni

Discuss that Bonferroni controls the family-wise error rate (FWER) but is conservative, reducing power and increasing sample size. It may be impractical if metrics are correlated or if one metric is more important.

4. Compare with gatekeeper metric

Describe gatekeeper approach: test one primary metric first; only if significant, test the second. This preserves α for the primary but may miss effects on the secondary if the primary fails. It prioritizes one metric over another.

5. Compare with hierarchical testing

Explain hierarchical (fixed-sequence) testing: order metrics by importance, test sequentially without alpha adjustment, stopping when a non-significant result occurs. This controls FWER and maintains power for the primary metric, but requires a priori ordering and may not test all metrics if early ones fail.

Key Points to Mention

  • Bonferroni correction: α/k, controls FWER, conservative, increases sample size.
  • Sample size formula: n ∝ (z_{1-α'/2} + z_{1-β})^2 / Δ^2, so smaller α' increases n.
  • Gatekeeper metric: sequential testing, preserves α for primary, risk of missing secondary effects.
  • Hierarchical testing: fixed-sequence, controls FWER, maintains power for primary, requires ordering.
  • Trade-offs: Bonferroni simple but costly; gatekeeper and hierarchical more efficient but require assumptions.
  • Consider correlation between metrics and business importance when choosing approach.

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

Q3

With 1.2M eligible sessions per day, 70% passing eligibility checks, and a cap of 20% of eligible traffic for the experiment, estimate how many days the test needs to run. Then propose a ramp schedule that keeps the test statistically valid while managing operational risk.

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

The arithmetic is straightforward: 1.2M times 0.7 is 840K eligible, 20% of that is 168K per day across both arms, so 84K per variant per day.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by calculating the daily eligible traffic and the experiment traffic under the 20% cap. Then determine the required sample size per variant based on desired statistical power and minimum detectable effect, and divide by daily traffic to get the duration. Finally, propose a ramp schedule that gradually increases traffic while monitoring for guardrail metric regressions.

Pro tip: Always state your assumptions explicitly (e.g., baseline conversion rate, MDE, power) and show that you understand the trade-off between test duration and sensitivity. Mention that you would validate the ramp with a pre-period and use sequential testing or CUPED to maintain validity.

1. Calculate daily eligible and experiment traffic

Compute 70% of 1.2M to get 840k eligible sessions per day. Then apply the 20% cap to get 168k sessions per day in the experiment.

2. Determine required sample size per variant

Use power analysis: assume baseline conversion rate, minimum detectable effect, significance level (α=0.05), and power (1-β=0.8). Calculate total sample size needed (e.g., using formula for two proportions).

3. Estimate test duration

Divide the total required sample size by the daily experiment traffic (168k) to get the number of days. Round up to account for weekly seasonality (e.g., full weeks).

4. Propose a ramp schedule

Start with a small percentage (e.g., 5% of eligible traffic) for a few days to monitor guardrail metrics, then increase gradually (e.g., 10%, 20%) over 1-2 weeks until reaching the cap, ensuring each ramp step has enough data to detect issues.

5. Ensure statistical validity during ramp

Use sequential testing or always-valid p-values to allow peeking, and consider CUPED to reduce variance. Maintain consistent assignment and avoid mixing ramp periods in the final analysis.

Key Points to Mention

  • Assumptions: baseline conversion rate, MDE, α, power, and whether metrics are rates or continuous.
  • Sample size calculation formula or reference to tools (e.g., power.prop.test in R).
  • Daily traffic calculation: 1.2M * 0.7 * 0.2 = 168k sessions/day.
  • Duration: total sample size / 168k, rounded up to full weeks to handle weekly seasonality.
  • Ramp schedule: start small (e.g., 5%), monitor guardrails, then increase to 20% over 1-2 weeks.
  • Statistical techniques: sequential testing, CUPED, or fixed horizon with no peeking if using traditional methods.

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

Q4

When would you use a z-test versus a t-test for this kind of experiment, and if your 95% confidence interval barely excludes zero, what does that actually tell you?

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

Z-test is fine here because sample sizes are large and Bernoulli outcomes let you invoke CLT without much worry.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the assumptions and context of the experiment (sample size, known variance, normality). Then explain the decision rule between z-test and t-test, and finally interpret the borderline confidence interval with nuance about practical significance and uncertainty.

Pro tip: At Meta, with large samples, the z-test and t-test often yield nearly identical results; focus on the effect size and business impact rather than just statistical significance. A confidence interval barely excluding zero suggests a small effect that may not be practically meaningful—always consider the confidence interval width and the cost of a false positive.

1. Clarify assumptions and context

Ask about sample size, whether population variance is known, and the distribution of the metric. This determines the appropriate test.

2. Explain z-test vs t-test

Use z-test when population variance is known and sample size is large (n>30) or data is normal. Use t-test when variance is unknown and estimated from the sample, especially with small samples.

3. Discuss practical considerations in A/B testing

In practice, with large samples (common in tech), t-test and z-test converge. However, t-test is more robust to unknown variance and is default in many tools.

4. Interpret borderline confidence interval

A 95% CI barely excluding zero means the effect is statistically significant but likely small. Consider the lower bound: if it's close to zero, the effect could be negligible. Also, 5% of such intervals will exclude zero by chance even if true effect is zero.

5. Connect to business impact

Assess whether the effect size is meaningful for the product. A tiny but significant lift might not justify implementation costs. Also consider the confidence interval width and potential for practical significance.

Key Points to Mention

  • Difference between z-test and t-test: known vs unknown population variance, sample size considerations.
  • In large-scale A/B tests, t-test is often preferred due to unknown variance and robustness.
  • Statistical significance vs practical significance: a barely significant result may have little real-world impact.
  • Confidence interval interpretation: 95% CI barely excluding zero indicates uncertainty; the true effect could be near zero.
  • Multiple testing and false positive risk: with many metrics, some will show significance by chance.
  • Effect size and confidence interval width: narrow CI with small effect vs wide CI with large effect.

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

Q5

Before looking at experiment outcomes, how would you verify that randomization worked correctly and that treatment and control groups are balanced across dimensions like country and platform?

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

Standard SRM check first, compare observed traffic split to expected 50/50 using a chi-square test.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining that randomization checks are a pre-experiment validation step to ensure internal validity. Describe a systematic process: first verify the randomization mechanism, then compare key dimensions between groups using statistical tests, and finally investigate any imbalances. Emphasize that while some imbalance is expected by chance, systematic differences indicate a problem.

Pro tip: Use standardized mean differences (SMD) rather than p-values for balance checks, as p-values are sensitive to sample size and can flag trivial imbalances in large experiments. Also, pre-register the dimensions and thresholds for imbalance to avoid post-hoc rationalization.

1. Verify Randomization Mechanism

Check that the randomization unit (e.g., user, session) was correctly assigned to treatment/control using the intended algorithm (e.g., hash-based). Ensure no bias in assignment, such as unequal probabilities or leakage.

2. Identify Key Dimensions for Balance

List pre-experiment covariates that could affect the outcome, such as country, platform, user demographics, and historical behavior. These should be measured before the experiment starts.

3. Compare Distributions Across Groups

For each dimension, compute summary statistics (e.g., mean, proportion) for treatment and control. Use standardized mean differences (SMD) or statistical tests (e.g., t-test, chi-square) to assess balance.

4. Assess Overall Balance and Investigate Imbalances

Check if any dimension exceeds a pre-defined threshold (e.g., SMD > 0.1). If imbalances are found, investigate potential causes (e.g., randomization bug, sample ratio mismatch) and consider whether they bias results.

5. Document and Decide on Next Steps

If randomization is flawed, halt the experiment or apply statistical adjustments (e.g., CUPED, stratification). If balanced, proceed with analysis, noting any minor imbalances as limitations.

Key Points to Mention

  • Sample Ratio Mismatch (SRM) check: compare observed vs. expected group sizes using chi-square test.
  • Standardized Mean Difference (SMD) as a scale-free measure of imbalance, with a common threshold of 0.1.
  • Importance of pre-experiment covariates and avoiding post-treatment variables.
  • Multiple comparisons problem: use Bonferroni correction or false discovery rate when testing many dimensions.
  • Practical significance vs. statistical significance: small imbalances may not bias results if uncorrelated with outcome.
  • Randomization unit and analysis unit consistency (e.g., user-level randomization requires user-level analysis).

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