Google Data Analyst Interview Questions (2026)

1. 1Recruiter screen30 minEasy

   Background, level calibration, and process walkthrough with a recruiter.

2. 2Technical phone screen45 minHard

   One or two DSA problems solved live in a shared editor with emphasis on optimal complexity and clean code.

3. 3Coding round (onsite)45 minHard

   Harder DSA with follow-up constraint changes; interviewer scores GCA and RRK on a rubric.

4. 4System design round45 minHard

   Design a planet-scale system (e.g. a piece of Search or YouTube) with explicit capacity estimates and tradeoffs.

5. 5Googleyness & Leadership45 minMedium

   Behavioral round on collaboration, ambiguity, and user-first judgment scored against Google's structured rubric.

6. 6Hiring Committee review30 minMedium

   No candidate interaction; the written feedback packet is reviewed and the hire/no-hire decision is made, followed by team matching.

1. Q1

   Design an A/B test for a new Search ranking or recommendation change. Define hypothesis, primary metric, guardrails, randomization unit, and launch decision rule

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Context: Google wants to improve relevance while protecting user trust and privacy.

   How to answer: A strong answer defines a clear, testable hypothesis for the Search ranking change, such as 'The new ranking algorithm will increase user engagement (e.g., clicks on top results) without negatively impacting search quality.' The primary metric should directly reflect the hypothesis, like 'Click-Through Rate (CTR) of the first 3 organic results' or 'Average time to successful search.' Guardrail metrics are crucial to detect negative side effects, such as 'Overall query success rate,' 'revenue per search,' or 'latency.' The randomization unit should be 'user-ID' to ensure consistent experience, and the launch decision rule should specify statistical significance thresholds for primary metrics and acceptable movements for guardrails, typically requiring no significant negative impact on any guardrail.

2. Q2

   For YouTube Shorts, should randomization happen at user, session, device, advertiser, or country level? Explain the tradeoffs

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Consider cross-device behavior, interference, marketplace effects, and operational feasibility.

   How to answer: Randomization for YouTube Shorts should primarily happen at the user level to ensure independent observations and avoid contamination. Session or device level randomization could be considered for very short-term, immediate impact tests where user identity across sessions/devices is hard to track, but risks user-level contamination. Advertiser or country level randomization is generally too broad and introduces significant confounding factors, making it difficult to isolate the treatment effect for a product like Shorts, which is consumed individually.

3. Q3

   Choose primary and guardrail metrics for a Google Ads experiment aimed at improving query success rate. What metrics would prevent a harmful launch?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Include user experience, partner health, revenue, reliability, and long-term retention considerations.

   How to answer: A strong candidate would identify 'Query Success Rate' (e.g., clicks on ads, conversions, or user engagement post-query) as the primary metric. For guardrail metrics, they would suggest 'Revenue per User/Query' (to ensure profitability isn't negatively impacted), 'Ad Impression Share' or 'Click-Through Rate (CTR)' (to monitor ad visibility and user interaction), and 'Latency' or 'Error Rate' (to ensure system stability and user experience aren't degraded). The key is to balance user experience improvements with business objectives and technical performance.

4. Q4

   During a Google Maps experiment, the treatment/control split is 52/48 instead of 50/50. How would you diagnose sample ratio mismatch?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Assume assignment logs, exposure logs, and eligibility filters may disagree.

   How to answer: A strong candidate would first define Sample Ratio Mismatch (SRM) and its implications for experiment validity. They would then outline a diagnostic process starting with checking the randomization unit (e.g., user ID, session ID) and ensuring consistent hashing. Next, they would verify data pipeline integrity, looking for data loss or filtering post-randomization. Finally, they would suggest analyzing the SRM across various dimensions like device type, geography, or time to pinpoint the source of the imbalance.

5. Q5

   The Google Play experiment is trending positive after two days. A PM wants to stop early and launch. How do you handle peeking and sequential testing?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Discuss pre-specified stopping rules, alpha spending, business urgency, and risk.

   How to answer: Explain that stopping an A/B test early due to positive trends (peeking) inflates the Type I error rate, leading to false positives. Discuss the need for pre-defined sample sizes or test durations based on power analysis to ensure statistical validity. Propose solutions like using sequential testing methods (e.g., Always Valid p-values, O'Brien-Fleming boundaries) or Bayesian approaches to allow for early stopping while controlling error rates. Emphasize communicating the risks of peeking to the PM and advocating for a full-duration test unless a valid sequential method is employed.

6. Q6

   A new Cloud Marketplace feature shows a large week-1 lift in query success rate, but the effect fades by week 4. What could explain this and how would you design the test duration?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Discuss novelty, learning effects, seasonality, and durable impact.

   How to answer: The fading effect could be due to novelty effect (users initially engage more with new features) or selection bias (early adopters are more tech-savvy and successful). Another possibility is a change in user behavior, where initial success leads to more complex queries that are harder to satisfy. To design the test duration, one should consider the user's learning curve and the natural usage cycle of the feature, aiming for at least 4-6 weeks to capture long-term behavior and account for novelty effects. A staggered rollout or a longer pre-period analysis could also help establish a more robust baseline.

7. Q7

   In a marketplace-like Search feature, treatment users may affect control users. How would network effects or interference bias the experiment?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Examples include advertiser supply, content inventory, delivery capacity, or pricing pressure.

   How to answer: Network effects in a marketplace A/B test can lead to interference, where the actions of treatment users impact control users, or vice-versa. This typically biases the experiment results, often underestimating the true treatment effect if the effect is positive (e.g., increased supply from treatment users benefits control users). The bias arises because the control group no longer represents a true baseline without the treatment's influence. This invalidates the core assumption of independent user groups, making direct comparison misleading.

8. Q8

   query success rate is a low-frequency event for YouTube Shorts. How would you set up an experiment with enough power without waiting too long?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Discuss proxy metrics, variance reduction, larger samples, longer windows, and risk of metric gaming.

   How to answer: To address low-frequency events like query success rate on YouTube Shorts, a strong candidate would propose using a larger sample size to achieve sufficient statistical power more quickly. They should also consider using a more sensitive metric, such as a proxy metric that occurs more frequently and is highly correlated with query success (e.g., 'query initiated' or 'time spent on search results page'). Additionally, they might suggest a longer experiment duration if the impact of the treatment is expected to manifest over time, or consider a sequential testing approach to stop early if a significant effect is observed.

9. Q9

   Design a geo or country-level experiment for Google Ads. When is this better than user-level randomization, and what are the analytical downsides?

   MediumStatistics & Experimentation RoundA/B TestingGoogle-specific

   Context: Use matched markets, pre-period balancing, spillover checks, and fewer experimental units.

   How to answer: A geo-level experiment for Google Ads involves randomizing entire geographic regions (e.g., countries, states, DMAs) into control and treatment groups, rather than individual users. This approach is superior when there are network effects, spillover effects, or when the treatment itself is geographically bound (e.g., a new ad format only available in certain regions). However, it introduces significant analytical challenges, including lower statistical power due to fewer experimental units, potential for selection bias if regions are not well-matched, and increased variance due to larger unit heterogeneity, requiring more sophisticated statistical methods like CUPED or synthetic control for analysis.

10. Q10

    The Google Maps experiment lifts query success rate overall, but only for new users and only in one device_type. How would you evaluate heterogeneous treatment effects?

    HardStatistics & Experimentation RoundA/B TestingGoogle-specific

    Context: Balance pre-planned segments with exploratory slicing and multiple testing risk.

    How to answer: To evaluate heterogeneous treatment effects (HTE) in this Google Maps experiment, I would start by segmenting the user base by relevant covariates like user tenure (new vs. existing) and device type. Then, I would perform separate A/B test analyses within each segment to calculate the treatment effect and its statistical significance. For segments showing a significant lift, I would further investigate potential confounding factors or interaction effects that might explain the localized impact. Finally, I would synthesize these segment-specific results to provide a comprehensive understanding of where and for whom the experiment is truly effective.

11. Q11

    Treatment improves query success rate but worsens page load latency for Google Play. Walk through a launch recommendation

    HardStatistics & Experimentation RoundA/B TestingGoogle-specific

    Context: Make a decision under conflicting metrics and quantify tradeoffs for stakeholders.

    How to answer: A strong recommendation would involve quantifying the trade-off between improved query success rate (QSR) and worsened page load latency (PLL) using a common metric like revenue or user engagement. This requires understanding the business impact of each metric and potentially running a follow-up experiment to isolate effects or test a hybrid solution. The recommendation should consider user segmentation, potential long-term effects on user retention, and clearly state assumptions and next steps, such as further analysis or a phased rollout.

12. Q12

    How would you design ramp-up, holdback, and post-launch monitoring for a successful Cloud Marketplace A/B test?

    HardStatistics & Experimentation RoundA/B TestingGoogle-specific

    Context: Include ramp stages, persistent holdback, alert thresholds, rollback criteria, and owner accountability.

    How to answer: A strong answer would detail a phased ramp-up strategy, starting with a small percentage of traffic (e.g., 1-5%) to detect technical issues and validate metrics, gradually increasing to 50% or full rollout. For holdback, the candidate should explain reserving a small, untouched control group (e.g., 1-2%) from the experiment to measure long-term novelty effects and ensure overall platform stability. Post-launch monitoring involves establishing real-time dashboards for key business metrics, technical health, and user feedback, alongside setting up alerts for significant deviations and defining clear rollback triggers.

13. Q13

    Midway through the Search test, tracking for Google Ads changed. How would you decide whether the experiment results are still usable?

    HardStatistics & Experimentation RoundA/B TestingGoogle-specific

    Context: Compare instrumentation versions, affected traffic share, raw logs, and sensitivity analyses.

    How to answer: A strong candidate would first identify the nature of the change (e.g., tracking pixel, attribution model, data schema) and its potential impact on key metrics like conversions or revenue. They would then propose analyzing pre- and post-change data for both control and experiment groups, looking for a significant shift or divergence in trends. Key methods include a difference-in-differences analysis or examining the change point for anomalies. If the change introduced a systemic bias affecting both groups equally and proportionally, the relative lift might still be valid; otherwise, the results are likely compromised or require advanced causal inference techniques to salvage.

14. Q14

    Two overlapping experiments on YouTube Shorts both affect ad revenue per mille. How would you detect and manage interaction effects?

    HardStatistics & Experimentation RoundA/B TestingGoogle-specific

    Context: Discuss experiment registry, factorial design, exclusion rules, and interaction terms.

    How to answer: A strong candidate would first define interaction effects and explain why they are problematic (e.g., invalidating individual experiment results, leading to suboptimal product decisions). They would then propose detection methods like factorial designs (if planned) or post-hoc analysis using regression models with interaction terms. Management strategies would include sequential experimentation, using guardrail metrics, or implementing an experimentation platform with robust conflict resolution and interaction detection capabilities. Finally, they would discuss the trade-offs of different approaches given YouTube's scale and complexity.

15. Q15

    Google's Search revenue suddenly drops 10% week over week. Structure a business case to diagnose the issue and identify the most likely drivers

    MediumProduct Analytics & Business CaseBusiness CasesGoogle-specific

    Context: Consider traffic, conversion, pricing, mix, supply/inventory, outages, marketing, and seasonality.

    How to answer: A strong candidate would structure their diagnosis by first clarifying the scope (global vs. regional, specific product lines) and confirming the data's accuracy. Then, they would categorize potential drivers into internal (product changes, algorithm updates, ad policy shifts, technical issues) and external factors (macroeconomic trends, competitor actions, seasonality, major news events). They would prioritize investigation by looking at internal changes first, then segmenting the drop by geography, device, query type, and advertiser category to pinpoint the most affected areas. Finally, they would propose specific data points and teams to consult for each potential driver.

Practice these with instant AI feedback in a live mock interview → Start a Google Data Analyst mock

Master DSA and communicate your thinking out loud; use Google's structured Explain-Clarify-Improve approach; prepare for Googleyness/behavioral

Indicative Data Analyst pay in India: ~₹6–22 LPA (role-level range, not a Google-specific figure).

Other roles at Google

Data Analyst interviews at other companies