← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

Senior
Apr 2026

Summary

A technical screen for a DS role at Meta that went deep fast. The whole thing was basically one sprawling experiment design question about fake account removal, and they wanted specifics on everything from cluster construction to quasi-experimental fallbacks. Felt like a take-home crammed into a live session.

Questions Asked (8)

Q1

Your team is removing detected fake accounts and wants to measure the causal impact on real users' experience. Design a full experiment plan that handles network interference and product trade-offs, covering randomization strategy, metrics, power calculations, spillover diagnostics, noncompliance, rollout safety, and quasi-experimental fallbacks.

A/B Testing & ExperimentationProduct Analytics & MetricsTechnical Trade-offs
Author's notes

This was essentially seven questions in a trenchcoat.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by framing the causal question and the interference challenge, then propose a cluster-randomized design with clear metrics and power analysis. Address practical issues like noncompliance, spillover, and safety, and outline quasi-experimental fallbacks for when randomization isn't feasible.

Pro tip: Emphasize that fake account removal can create spillovers through social ties, so cluster randomization by network communities is often necessary; also highlight the importance of pre-registering the analysis plan to avoid p-hacking.

1. Define the causal estimand and randomization unit

Clarify the treatment (removing fake accounts) and outcome (real user experience). Choose a randomization unit that minimizes interference, such as clusters of connected users or geographic regions.

2. Select metrics and design power analysis

Identify primary metrics (e.g., engagement, retention, reports of fake accounts) and guardrail metrics (e.g., user trust). Conduct power calculations accounting for cluster randomization and intra-cluster correlation.

3. Plan for spillover, noncompliance, and safety

Design diagnostics to detect spillover (e.g., compare treated and control clusters' exposure). Monitor noncompliance (e.g., fake accounts not removed) and implement safety checks with sequential testing or stopping rules.

4. Outline quasi-experimental fallbacks

If randomization is infeasible, propose alternatives like difference-in-differences, synthetic control, or instrumental variables, and discuss their assumptions and limitations.

Key Points to Mention

  • Cluster randomization to handle network interference
  • Power analysis with intra-cluster correlation
  • Spillover diagnostics (e.g., exposure mapping, SUTVA violations)
  • Noncompliance (e.g., intent-to-treat vs. treatment-on-treated)
  • Rollout safety (e.g., sequential testing, guardrail metrics)
  • Quasi-experimental methods (DiD, synthetic control, IV)

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

Q2

Between user-level, ego-network cluster, and geography-level randomization, which would you choose for this fake account removal experiment and why? How would you build clusters to limit cross-treatment contamination without killing statistical power?

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

Went with ego-network clustering.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the experiment's goal and the nature of the fake account removal intervention, then evaluate each randomization unit against the trade-off between contamination control and statistical power. Recommend a hybrid or cluster-based approach that balances both, and explain how you would construct clusters to minimize spillover while preserving power.

Pro tip: Acknowledge that perfect isolation is often impossible in social networks; instead, propose a design that quantifies and bounds contamination, such as using ego-network clusters with a buffer zone, and pre-register sensitivity analyses to check robustness.

1. Clarify the intervention and outcome

Understand what 'fake account removal' entails (e.g., removing accounts, flagging, reducing reach) and the primary metric (e.g., user engagement, spam reports). This determines the likely spillover mechanisms.

2. Evaluate randomization units

Compare user-level, ego-network, and geography-level randomization on contamination risk and power. User-level has high contamination but high power; geography-level has low contamination but low power; ego-network clusters offer a middle ground.

3. Choose a unit and justify

Select ego-network cluster randomization as the best balance, explaining that it groups connected users to limit spillover while retaining more clusters than geography-level, thus preserving power.

4. Design clusters to limit contamination

Build clusters using community detection algorithms (e.g., Louvain, Leiden) on the interaction graph, then merge small clusters and add buffer zones between treatment and control clusters to reduce cross-treatment interference.

5. Address power and analysis

Use cluster-level randomization inference, account for intra-cluster correlation, and consider variance reduction techniques (e.g., CUPED) to maintain power. Pre-register sensitivity analyses for contamination.

Key Points to Mention

  • Trade-off between contamination (SUTVA violations) and statistical power
  • Ego-network clusters as a compromise between user-level and geography-level
  • Community detection algorithms (Louvain, Leiden) for cluster construction
  • Buffer zones or graph partitioning to reduce cross-treatment interference
  • Cluster-level analysis and randomization inference to handle correlation
  • Variance reduction methods (CUPED, stratification) to preserve power

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

Q3

What primary and guardrail metrics would you define for this experiment? Be specific about metric formulas, time windows, and whether they are exposure-based or calendar-based.

Product Analytics & MetricsA/B Testing & Experimentation
Author's notes

Comments per view as primary felt obvious.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the experiment's goal and the product surface, then define one primary metric that directly measures success, and 2-3 guardrail metrics that protect user experience and long-term health. For each metric, specify the exact formula, the time window (e.g., 7-day), and whether it's exposure-based (per user exposed) or calendar-based (daily aggregated).

Pro tip: Always tie guardrails to the specific risks of the change—e.g., if testing a new ad format, include ad load and user-reported satisfaction—and mention that you'd monitor them with sequential testing to catch early harm.

1. Clarify experiment context and goal

Ask about the product area, the change being tested, and the primary objective (e.g., increase engagement, revenue, or retention). This ensures metrics align with business goals.

2. Define the primary metric

Choose one metric that directly measures the experiment's success. Specify its formula (e.g., clicks per user), time window (e.g., 7 days post-exposure), and whether it's exposure-based (per exposed user) or calendar-based (daily active users).

3. Identify guardrail metrics

Select 2-3 metrics that capture potential negative side effects, such as user retention, satisfaction, or system performance. Define each with formula, time window, and exposure/calendar basis.

4. Specify time windows and aggregation

For each metric, state the exact time window (e.g., 1-day, 7-day, 28-day) and whether it's measured per user (exposure-based) or aggregated daily (calendar-based). Explain why the window fits the metric's sensitivity.

5. Summarize and justify choices

Briefly recap the primary and guardrail metrics, and explain how they balance measuring success with protecting user experience and long-term goals.

Key Points to Mention

  • Primary metric should be directly tied to the experiment's hypothesis and business objective.
  • Guardrail metrics must cover different risk dimensions: user experience, ecosystem health, and technical performance.
  • Exposure-based metrics (e.g., per user) are often preferred for A/B tests to avoid dilution from users not exposed.
  • Calendar-based metrics (e.g., daily totals) are useful for detecting system-wide or temporal effects.
  • Time windows should align with the metric's expected latency (e.g., immediate clicks vs. 7-day retention).
  • Consider using ratio metrics (e.g., clicks per session) and ensure they are statistically powered.

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

Q4

Walk through a sample size calculation for detecting a 0.5% relative change in comments per view, assuming a baseline of 0.12, overdispersion, and an intra-cluster correlation of 0.02.

A/B Testing & Experimentation
Author's notes

This is where I embarrassed myself a little.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the metric definition and assumptions, then derive the sample size formula for a two-sample proportion test with overdispersion and clustering adjustments. Walk through the calculation step-by-step, explaining how each adjustment (overdispersion factor and design effect) inflates the required sample size.

Pro tip: Always discuss the practical implications of the adjustments—overdispersion and clustering can dramatically increase sample size, so it's crucial to consider whether the experiment is feasible and to explore variance reduction techniques like CUPED.

1. Clarify metric and assumptions

Define comments per view as a ratio metric, confirm baseline rate (0.12), relative change (0.5% => absolute change = 0.0006), and specify significance level (e.g., 5%) and power (e.g., 80%).

2. Compute base sample size without adjustments

Use the standard formula for two proportions: n = (Zα/2 + Zβ)^2 * (p1(1-p1) + p2(1-p2)) / (p2-p1)^2, where p1=0.12, p2=0.1206, to get the per-variant sample size.

3. Adjust for overdispersion

Multiply the base sample size by an overdispersion factor (e.g., variance inflation factor) to account for extra variability in the metric beyond binomial assumption.

4. Adjust for clustering (ICC)

Compute the design effect = 1 + (average cluster size - 1) * ICC, and multiply the sample size by this factor to account for intra-cluster correlation.

5. Final sample size and sensitivity

Combine adjustments to get the final required sample size, and discuss sensitivity to assumptions (e.g., ICC, overdispersion) and potential variance reduction methods.

Key Points to Mention

  • Metric definition: comments per view is a ratio metric, often modeled with a Poisson or negative binomial distribution.
  • Overdispersion: variance exceeds mean; adjust using a dispersion parameter or quasi-Poisson approach.
  • Intra-cluster correlation (ICC): accounts for correlation within clusters (e.g., users within groups); design effect formula.
  • Sample size formula for proportions: two-sample z-test for proportions, with continuity correction if needed.
  • Power and significance level: typically 80% power and 5% significance, but clarify with stakeholders.
  • Practical considerations: feasibility, duration, and variance reduction techniques like CUPED or stratification.

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

Q5

Propose two specific tests to detect and quantify spillover effects in this experiment. What would the null hypothesis look like for each?

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

I suggested a ghost exposure analysis where you look at real users connected to removed fake accounts in the control group and check if their behavior shifts, and an edge-cut A/A where you compare users on the boundary of treatment clusters to interior users.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining what spillover effects mean in the context of the experiment, then propose two distinct tests: one for detecting spillover via network or geographic proximity, and another for quantifying it using a difference-in-differences or instrumental variable approach. For each test, clearly state the null hypothesis of no spillover effect. Emphasize the importance of randomization unit and potential interference.

Pro tip: Mention that spillover can bias treatment effect estimates, and propose using cluster randomization or saturation design to mitigate it. Also, highlight the need to pre-register the analysis plan to avoid p-hacking.

1. Define spillover and context

Clarify what spillover means in this experiment (e.g., users influencing each other through social ties or geographic proximity) and why it matters for causal inference.

2. Propose detection test

Suggest a test to detect spillover, such as comparing outcomes of control users who are connected to treatment users versus those who are not, using network data. State the null hypothesis: no difference in outcomes between these groups.

3. Propose quantification test

Suggest a test to quantify spillover, such as a regression with treatment intensity of neighbors as a covariate, or a difference-in-differences design comparing areas with varying treatment saturation. State the null hypothesis: the coefficient on spillover term is zero.

4. Discuss assumptions and limitations

Acknowledge assumptions like no unmeasured confounding, and limitations such as imperfect network data. Mention potential biases and how to address them.

5. Conclude with implications

Summarize how these tests inform experiment design and decision-making, and suggest mitigation strategies if spillover is detected.

Key Points to Mention

  • Interference between units (SUTVA violation)
  • Network analysis or geographic proximity for spillover detection
  • Difference-in-differences or instrumental variables for quantification
  • Null hypothesis: no spillover effect (e.g., coefficient = 0)
  • Cluster randomization or saturation design to mitigate spillover
  • Pre-registration and power analysis for spillover tests

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

Q6

Detection of fake accounts is imperfect. How would you handle noncompliance and misclassification when estimating the treatment effect? Walk through an instrumental variable or difference-in-differences approach.

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

Used removal intensity (fraction of flagged accounts actually removed in a cluster) as the instrument for actual fake account exposure reduction.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging that imperfect detection creates measurement error and noncompliance, which bias naive estimates. Then propose using an instrumental variable (IV) or difference-in-differences (DiD) design to isolate the treatment effect, depending on the setting. Walk through the assumptions, implementation, and sensitivity checks for each approach.

Pro tip: Emphasize that the IV must affect treatment take-up but not the outcome directly (exclusion restriction), and for DiD, the parallel trends assumption is key—suggest placebo tests and robustness checks to build credibility.

1. Define the causal estimand and sources of bias

Clarify the treatment effect of interest (e.g., LATE) and explain how misclassification and noncompliance introduce bias in naive comparisons.

2. Choose an identification strategy

Decide between IV and DiD based on available data and context. For IV, identify a valid instrument; for DiD, define treatment and control groups with pre/post periods.

3. State and test assumptions

For IV: relevance, exclusion, and monotonicity. For DiD: parallel trends. Discuss how to test or probe these assumptions (e.g., first-stage F-stat, pre-trends).

4. Estimate and interpret

Use two-stage least squares for IV or regression with interaction terms for DiD. Interpret the effect in light of the assumptions and potential violations.

5. Conduct sensitivity and robustness checks

Perform placebo tests, alternative specifications, and bounds analysis to assess how violations of assumptions might affect conclusions.

Key Points to Mention

  • Measurement error and noncompliance bias naive estimates
  • Instrumental variable: relevance, exclusion restriction, monotonicity
  • Difference-in-differences: parallel trends assumption
  • Local average treatment effect (LATE) for IV
  • Sensitivity analysis and robustness checks (e.g., placebo tests, bounds)
  • Practical challenges: finding valid instruments, staggered adoption in DiD

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

Q7

How would you design a staged rollout with kill-switch criteria? How do you prevent label leakage through feeds or notifications during the experiment?

A/B Testing & ExperimentationProduct Strategy
Author's notes

Staged rollout was pretty standard: start at 1%, check guardrails, expand.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by outlining a phased rollout plan with clear success metrics and kill-switch thresholds, then address label leakage by describing how to isolate treatment effects from social influence. Emphasize the importance of pre-registering criteria and using techniques like cluster-based randomization or holdout groups to prevent contamination.

Pro tip: At Meta, where social networks are dense, label leakage is a major concern; propose using ego-network clustering or graph-aware randomization to minimize interference, and always have a pre-defined kill-switch that triggers automatically if guardrail metrics degrade beyond a threshold.

1. Define Objectives and Metrics

Clearly state the primary success metric (e.g., engagement, revenue) and guardrail metrics (e.g., user satisfaction, latency). Establish minimum detectable effect and required sample size.

2. Design Staged Rollout

Plan phases: internal test, small percentage (e.g., 1%), then gradual increase (5%, 10%, 50%). At each stage, monitor metrics and have pre-defined kill-switch criteria (e.g., if guardrail metric drops by >2% with statistical significance).

3. Mitigate Label Leakage

Use cluster-based randomization (e.g., by user ego-networks or geographic regions) to reduce interference. Alternatively, employ a holdout group or switchback testing to isolate treatment effects.

4. Monitor and Automate Kill-Switch

Set up real-time dashboards and automated alerts. If kill-switch criteria are met, automatically halt the experiment and revert changes. Document learnings for future experiments.

5. Analyze and Iterate

After rollout, analyze results accounting for interference (e.g., using causal inference methods). If successful, continue scaling; if not, iterate on design or abandon.

Key Points to Mention

  • Pre-registration of hypotheses, metrics, and kill-switch thresholds to avoid p-hacking
  • Use of guardrail metrics to detect unintended negative impacts
  • Cluster randomization or ego-network clustering to handle interference in social networks
  • Holdout groups and switchback testing as alternatives to traditional A/B testing
  • Automated monitoring and alerting systems for real-time kill-switch activation
  • Consideration of network effects and spillover in metric analysis

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

Q8

If randomization is not feasible in certain markets, what quasi-experimental approach would you use as a backup? What specific covariates from logs would you need, and what product recommendation would you make if engagement drops short-term but abuse reports fall significantly?

A/B Testing & ExperimentationProduct StrategyProduct Sense & Ideation
Author's notes

Synthetic control or staggered rollout with two-way fixed effects were the two options I laid out.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging that randomization isn't always possible and propose a quasi-experimental design like difference-in-differences or synthetic control, using pre-treatment trends to validate. Then specify the covariates you'd pull from logs (e.g., user demographics, historical engagement, abuse-related signals) and how they'd be used to control for confounding. Finally, weigh the short-term engagement drop against the significant reduction in abuse reports, and recommend a product change that balances both metrics, such as a targeted intervention or a phased rollout.

Pro tip: Show that you understand the trade-off between engagement and abuse is not just a metric conflict but a business decision—tie your recommendation to Meta's mission and long-term user trust, and suggest a follow-up experiment to measure the intervention's impact on both metrics.

1. Choose a quasi-experimental design

Select an appropriate method such as difference-in-differences, synthetic control, or regression discontinuity, depending on the market and data availability. Explain why it's valid and how you'd test its assumptions.

2. Identify necessary covariates from logs

List specific covariates like user demographics, historical engagement metrics, abuse report rates, and time-based features. Explain how they'd be used to adjust for confounding and ensure comparability between treatment and control groups.

3. Analyze the trade-off between engagement and abuse

Quantify the short-term engagement drop and the reduction in abuse reports, and assess whether the abuse reduction is significant enough to justify the engagement loss. Consider segment-level impacts and long-term effects.

4. Formulate a product recommendation

Propose a concrete product change, such as a targeted intervention that mitigates engagement loss while preserving abuse reduction, or a phased rollout with monitoring. Justify it with data and business goals.

5. Suggest validation and next steps

Recommend a follow-up experiment or analysis to validate the quasi-experimental findings and measure the long-term impact of the product change on both engagement and abuse.

Key Points to Mention

  • Difference-in-differences or synthetic control as quasi-experimental methods
  • Covariates: user demographics, historical engagement, abuse report rates, time-based features
  • Trade-off analysis: short-term engagement vs. long-term trust and safety
  • Product recommendation: targeted intervention or phased rollout
  • Validation: follow-up experiment or A/B test where feasible
  • Business impact: alignment with Meta's mission and user well-being

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