← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

Senior
Sep 2025Remote

Summary

A Meta DS interview that was essentially one long A/B testing case study on a hashtag recommender system. The depth they expected was real, covering everything from randomization unit choice to power calculations with intraclass correlation adjustments. Left feeling like I'd taken an exam.

Questions Asked (5)

Q1

How would you choose the randomization unit for an A/B test on a hashtag recommender, and how does cross-session interference affect that choice?

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

This is where I stumbled first.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the randomization unit options (user, session, hashtag, etc.) and the trade-offs between them, emphasizing the need to align with the metric of interest and minimize interference. Then discuss how cross-session interference (e.g., user learning, network effects) can bias results and why user-level randomization is often preferred in such cases. Conclude with practical considerations like sample size, power, and implementation feasibility.

Pro tip: Mention that while user-level randomization is standard, if the recommender's effect is expected to be short-lived, session-level randomization might be acceptable—but always check for carryover effects. Also, consider using cluster randomization if interference is severe.

1. Identify the metric and unit of analysis

Determine what you're measuring (e.g., click-through rate, hashtag usage) and the natural unit for that metric (user, session, hashtag). This guides the randomization unit.

2. Assess interference and carryover effects

Evaluate how one user's exposure might affect another's (network effects) or how a user's past exposure affects future behavior (carryover). Cross-session interference is a key concern.

3. Choose randomization unit balancing bias and variance

User-level randomization minimizes interference but may reduce power if effects are short-term. Session-level increases power but risks contamination. Consider cluster randomization if needed.

4. Validate assumptions and run power analysis

Check for interference via A/A tests or holdout groups. Calculate required sample size based on chosen unit and expected effect size.

5. Monitor and adjust if needed

During the experiment, monitor for unexpected interference or novelty effects. Be prepared to switch units or extend the test if issues arise.

Key Points to Mention

  • User-level randomization is the gold standard for social products due to network effects and cross-session interference.
  • Session-level randomization can be used if the effect is immediate and no carryover, but it risks contamination if users are exposed to both variants.
  • Hashtag-level randomization is rarely appropriate because hashtags are not independent units and users can see multiple hashtags.
  • Cross-session interference occurs when a user's behavior in one session is influenced by their exposure in a previous session, violating the independence assumption.
  • Consider using a cluster randomized design (e.g., by social network clusters) if interference is strong.
  • Always run A/A tests to detect interference and ensure the randomization unit is valid.

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

Q2

How would you structure the treatment and control groups, plan the rollout ramp, and prevent cross-over or contamination in this experiment?

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

Talked through sticky bucketing to lock users into arms across sessions, and flagged that you might want a holdout on certain high-volume hashtags to avoid spillover through the hashtag graph itself.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the experiment's goal, unit of randomization, and key metrics. Then outline a randomization strategy that ensures comparable groups, a phased rollout to monitor early signals, and specific measures to prevent contamination. Emphasize trade-offs and practical constraints at Meta's scale.

Pro tip: At Meta, experiments often involve social networks, so contamination is a major concern. Mentioning techniques like cluster randomization or ego-network isolation shows you understand the platform's unique challenges.

1. Define the Experiment and Unit of Randomization

Clarify the hypothesis, primary metric, and the unit of randomization (e.g., user, session, cluster). Consider network effects and choose the appropriate unit to minimize interference.

2. Design Treatment and Control Groups

Ensure groups are statistically equivalent via random assignment. For network experiments, consider cluster-based randomization or graph partitioning to reduce spillover.

3. Plan the Rollout Ramp

Start with a small percentage (e.g., 1%) to detect early issues, then gradually increase while monitoring guardrail metrics. Use a ramp schedule that balances risk and speed.

4. Prevent Cross-over and Contamination

Implement isolation techniques such as geo-based randomization, time-based separation, or network clustering. Monitor for spillover effects and adjust as needed.

5. Monitor and Validate

Continuously check for sample ratio mismatch (SRM), contamination metrics, and guardrail metrics. Be prepared to pause or adjust the experiment if issues arise.

Key Points to Mention

  • Randomization unit and its implications for network effects
  • Cluster randomization or graph partitioning to mitigate interference
  • Sample ratio mismatch (SRM) checks and guardrail metrics
  • Phased rollout with pre-defined stopping rules
  • Techniques like geo-based or time-based isolation
  • Trade-offs between statistical power and contamination control

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

Q3

Define a primary success metric and at least three guardrail metrics for this hashtag recommender experiment.

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

Primary metric was straightforward: 24-hour follow-through rate per exposed user.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the experiment's goal: to increase user engagement with hashtags. Propose a primary success metric that directly measures this goal, such as hashtag click-through rate or usage. Then, identify guardrail metrics that ensure the change doesn't harm other key aspects like user retention, content quality, or system performance.

Pro tip: Choose guardrail metrics that are sensitive to potential negative side effects of the recommender, such as decreased diversity of hashtags or increased reports of spam. Also, consider both short-term and long-term guardrails to catch delayed effects.

1. Clarify the Experiment Goal

Restate the purpose of the hashtag recommender: to help users discover and use relevant hashtags, thereby increasing engagement. Confirm with the interviewer if needed.

2. Define Primary Success Metric

Select a metric that directly measures the desired outcome, such as the number of hashtag clicks per user or the percentage of posts that include recommended hashtags. Ensure it is aligned with the overall product objective.

3. Identify Potential Risks

Brainstorm possible negative consequences of the recommender, such as users feeling spammed, decreased content diversity, or technical issues like increased latency.

4. Select Guardrail Metrics

Choose at least three guardrail metrics that monitor these risks. Examples include user retention, hashtag diversity, report rate, and system performance metrics.

5. Justify and Prioritize

Explain why each metric was chosen and how it relates to the experiment. If needed, prioritize them based on potential impact.

Key Points to Mention

  • Primary metric should be directly tied to the experiment's goal, e.g., hashtag click-through rate or usage rate.
  • Guardrail metrics should cover different aspects: user experience (e.g., retention, satisfaction), content ecosystem (e.g., diversity, quality), and system health (e.g., latency, error rates).
  • Consider both leading and lagging indicators; for example, immediate engagement vs. long-term retention.
  • Use statistical power analysis to ensure metrics can detect meaningful changes.
  • Avoid metrics that are too correlated with the primary metric to prevent redundancy.
  • Mention the importance of monitoring guardrails throughout the experiment and setting thresholds for alerting.

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

Q4

Walk through the sample size calculation for this experiment: baseline follow rate 4%, two-sided alpha 0.05, power 0.80, minimum detectable effect of +5% relative, with user-level randomization and an intraclass correlation of 0.02 across repeated sessions. Show the formulas and adjustments.

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

This was the hardest part and I knew it going in.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the metric and randomization unit, then compute the required sample size for a simple two-proportion z-test using the given baseline and relative MDE. Next, adjust for the intraclass correlation by inflating the variance using the design effect (1 + (m-1)*ICC), where m is the average number of sessions per user. Finally, present the final sample size in terms of users and discuss any additional considerations like multiple testing or novelty effects.

Pro tip: Always state your assumptions explicitly (e.g., average sessions per user, equal variance) and note that the ICC adjustment assumes a common correlation structure; in practice, you might use historical data to estimate m and ICC more precisely.

1. Clarify metric and randomization

Confirm that the primary metric is a binary follow rate, randomization is at user level, and each user can have multiple sessions. Define the unit of analysis (user) and the unit of randomization (user).

2. Compute base sample size without clustering

Use the standard formula for two-proportion z-test: n = (Z_{α/2} + Z_β)^2 * (p1(1-p1) + p2(1-p2)) / (p2 - p1)^2, where p1=0.04, p2=0.04*1.05=0.042, α=0.05 (two-sided), β=0.20. Calculate n per arm.

3. Adjust for intraclass correlation (ICC)

Compute the design effect (DE) = 1 + (m - 1) * ICC, where m is the average number of sessions per user. Multiply the base sample size by DE to account for correlation among repeated sessions within users.

4. Finalize sample size and discuss implications

Present the adjusted sample size per arm and total. Mention that this assumes equal cluster sizes and a common ICC; if m varies, use the average or a more complex model. Also note any additional adjustments (e.g., for multiple comparisons, if needed).

Key Points to Mention

  • Two-proportion z-test formula for sample size calculation
  • Relative vs absolute MDE: +5% relative means p2 = p1 * 1.05
  • Design effect formula: 1 + (m - 1) * ICC
  • Assumption of equal cluster sizes and common ICC
  • Need to estimate average number of sessions per user (m) from historical data
  • Potential need for further inflation due to multiple testing or non-compliance

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

Q5

What bias controls would you put in place for this experiment, including variance reduction techniques, novelty effect handling, sequential monitoring, and network interference diagnostics?

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

Ran through pre-period covariate adjustment to reduce variance, flagging that you'd want at least a week of pre-period data per user.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Structure your answer around the four areas the question explicitly asks about, but frame them as a coherent bias-control strategy rather than a checklist. For each area, briefly state the technique, why it matters, and how you would implement it at Meta's scale. Emphasize trade-offs and practical constraints.

Pro tip: Show that you understand bias controls are not one-size-fits-all: tailor them to the experiment's specific risks (e.g., social network experiments need different interference handling than UI changes). Mention that you would pre-register the analysis plan and use holdouts to validate long-term effects.

1. Variance reduction techniques

Use stratification, covariate adjustment (CUPED), and paired designs to increase sensitivity without introducing bias. Explain how these reduce variance while maintaining valid inference.

2. Novelty effect handling

Detect and mitigate novelty effects by monitoring treatment effect over time, using holdout groups, and analyzing early vs. late periods. Consider running longer experiments or using switchback designs.

3. Sequential monitoring

Implement sequential testing with alpha spending (e.g., O'Brien-Fleming boundaries) or always-valid inference to allow early stopping without inflating false positives. Discuss the trade-off between speed and power.

4. Network interference diagnostics

For social or marketplace experiments, diagnose interference via cluster randomization, graph-based methods, or ego-network designs. Use diagnostics like spillover detection and consider cluster-level analysis.

5. Integrate and validate

Combine these controls into a pre-registered analysis plan, run A/A tests to validate, and use holdouts to measure long-term effects. Continuously monitor for bias and adjust as needed.

Key Points to Mention

  • CUPED (Covariate Adjustment Using Pre-Experiment Data) for variance reduction
  • Novelty effect detection via time-series analysis and holdout groups
  • Sequential testing with alpha spending functions (e.g., O'Brien-Fleming) or always-valid p-values
  • Network interference: cluster randomization, ego-network designs, and spillover diagnostics
  • Pre-registration of analysis plan and A/A tests for validation
  • Trade-offs: sensitivity vs. validity, speed vs. rigor, and scalability

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