Zomato Analytics Engineer Interview Questions (2026)

1. 1Coding Screen45 minMedium

   Medium DSA problems focused on arrays, strings, and hashmaps with working code expected quickly.

2. 2Technical Round 250 minMedium

   A harder DSA problem plus practical engineering discussion drawn from your projects and real Zomato features.

3. 3Design / Product-Thinking Round55 minHard

   Design a Zomato or Blinkit feature end to end (e.g. live order tracking or dark-store picker flow), balancing tech design with product judgment.

4. 4Hiring Manager / Culture Round45 minMedium

   Blunt conversation on hunger, ownership, why Zomato, and how you handle chaos and hard feedback.

5. 5HR Round25 minEasy

   Compensation, ESOPs, notice period, and setting expectations on pace and in-office work.

1. Q1

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

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

   Context: Context: Zomato wants to improve discovery and restaurant monetization without hurting customer experience.

   How to answer: A strong answer will define a clear, testable hypothesis, such as 'The new ranking algorithm will increase successful order conversions.' The primary metric should directly reflect this hypothesis, like 'Conversion Rate (Orders/Session)' or 'Revenue Per User'. Guardrail metrics are crucial to detect negative side effects, such as 'Average Delivery Time' or 'Cancellation Rate'. The randomization unit should be carefully chosen, typically 'User ID' or 'Session ID', to minimize contamination. Finally, a clear launch decision rule, based on statistical significance and practical impact across primary and guardrail metrics, must be articulated.

2. Q2

   For Dining, should randomization happen at customer, session, device, restaurant, or city level? Explain the tradeoffs

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

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

   How to answer: Randomization for Zomato Dining should primarily happen at the customer level to ensure independent user experiences and minimize contamination, especially for features impacting individual user behavior or recommendations. Session or device level randomization could be considered for very short-term, non-persistent UI changes, but risks user inconsistency. Restaurant level randomization is suitable for changes impacting all users of a specific restaurant (e.g., menu display logic, booking availability), but requires careful consideration of network effects. City level randomization is generally too broad and introduces significant noise unless the experiment is specifically designed to measure city-wide market dynamics or supply-side changes.

3. Q3

   Choose primary and guardrail metrics for a Zomato Gold experiment aimed at improving menu-to-order conversion. What metrics would prevent a harmful launch?

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

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

   How to answer: The primary metric should directly measure menu-to-order conversion, such as 'Menu View to Order Placement Rate' or 'Orders per Menu View'. Guardrail metrics are crucial to prevent negative side effects. Key guardrails would include 'Average Order Value (AOV)', 'Order Cancellation Rate', 'Restaurant Partner Satisfaction Score', and 'Customer Retention Rate'. These guardrails ensure the experiment doesn't cannibalize revenue, degrade user experience, or harm relationships with restaurant partners.

4. Q4

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

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

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

   How to answer: To diagnose sample ratio mismatch (SRM), I would first check for data integrity issues like incorrect logging or ETL failures. Next, I'd analyze pre-experiment metrics and user characteristics (e.g., signup date, location, device) to see if the randomization unit (e.g., user ID) was properly assigned across groups. I would perform a chi-squared test on the observed vs. expected group counts to statistically confirm SRM. Finally, I'd investigate potential causes such as implementation bugs in the randomization logic, user exclusion criteria applied post-randomization, or issues with how the experiment platform assigns users.

5. Q5

   The Hyperpure 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 TestingZomato-specific

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

   How to answer: Explain that peeking early can lead to inflated Type I error rates, increasing the chance of false positives. Discuss the need to pre-define stopping rules and sample size based on a power analysis before the experiment begins. Mention sequential testing methods (e.g., A/B testing with O'Brien-Fleming boundaries or Always Valid Inference) as a way to allow for continuous monitoring while controlling the Type I error. Advise against stopping early unless pre-specified criteria are met, emphasizing the importance of statistical rigor and avoiding p-hacking.

6. Q6

   A new Reviews feature shows a large week-1 lift in menu-to-order conversion, but the effect fades by week 4. What could explain this and how would you design the test duration?

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

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

   How to answer: The fading effect suggests novelty effect or selection bias. Users might initially engage more due to the newness of the feature, or early adopters (who are more likely to convert) might be overrepresented in the initial lift. To design the test duration, consider the user lifecycle and the time it takes for user behavior to stabilize. A minimum of 4-6 weeks is often necessary to observe long-term effects and account for weekly seasonality, with longer durations for features impacting infrequent actions.

7. Q7

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

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

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

   How to answer: Network effects in a food delivery marketplace can bias A/B tests by causing treatment effects to spill over into the control group, or vice versa. This interference typically leads to an underestimation or overestimation of the true treatment effect, depending on the nature of the interaction. For example, if a treatment reduces driver availability for control users, the control group's experience worsens, making the treatment appear more effective than it truly is. Conversely, if a treatment improves driver efficiency that benefits all users, the control group might also see improvements, diluting the perceived treatment effect.

8. Q8

   menu-to-order conversion is a low-frequency event for Dining. How would you set up an experiment with enough power without waiting too long?

   MediumStatistics & Experimentation RoundA/B TestingZomato-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 menu-to-order conversion in Dining, a strong candidate would propose using a proxy metric that is a leading indicator and occurs more frequently, such as 'menu view to dish view conversion' or 'dish view to add to cart conversion'. They would also suggest increasing the sample size by broadening the experiment's scope (e.g., more restaurants, longer duration if feasible without significant risk, or targeting a wider user segment) or by leveraging a more sensitive statistical test if appropriate. Additionally, they might consider a sequential testing approach to potentially stop the experiment early if a significant effect is observed, or a Bayesian approach which can be more flexible with smaller sample sizes and low event rates. Finally, they should discuss the trade-offs of using a proxy metric and the importance of validating its correlation with the ultimate business metric.

9. Q9

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

   MediumStatistics & Experimentation RoundA/B TestingZomato-specific

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

   How to answer: A geo/city-level experiment for Zomato Gold would involve randomly assigning entire cities or geographical regions to either a control group (no Zomato Gold changes) or a treatment group (new Zomato Gold feature/pricing). This is better than user-level randomization when there's a high risk of 'spillover effects' where the treatment of one user influences others, such as through word-of-mouth or competitive pricing. Analytical downsides include lower statistical power due to fewer experimental units, increased sensitivity to selection bias if randomization isn't perfectly balanced, and longer experiment durations to observe significant effects, making it harder to isolate the impact from other concurrent city-level events.

10. Q10

    The Restaurant Ads experiment lifts menu-to-order conversion overall, but only for new users and only in one cuisine. How would you evaluate heterogeneous treatment effects?

    HardStatistics & Experimentation RoundA/B TestingZomato-specific

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

    How to answer: To evaluate heterogeneous treatment effects (HTE), first define specific subgroups based on user tenure (new vs. existing) and cuisine type. Then, perform separate A/B test analyses for each subgroup, focusing on the menu-to-order conversion metric. Utilize statistical methods like interaction terms in regression models or post-hoc subgroup analysis with appropriate multiple comparison corrections to identify significant differences in treatment effect across these segments. Finally, interpret the practical implications of these differential effects for product strategy and further experimentation.

11. Q11

    Treatment improves menu-to-order conversion but worsens order cancellation rate for Hyperpure. Walk through a launch recommendation

    HardStatistics & Experimentation RoundA/B TestingZomato-specific

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

    How to answer: A strong candidate would first clarify the magnitude of the positive and negative impacts, potentially through a weighted average or a monetary value assigned to each. They would then propose a phased rollout strategy, perhaps starting with a smaller, less sensitive Hyperpure region or a specific customer segment, to gather more data and observe long-term effects. Crucially, they would suggest investigating the root cause of the increased cancellation rate to identify potential mitigations or refinements to the treatment. Finally, they would recommend defining clear success metrics and a rollback plan before proceeding with any launch.

12. Q12

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

    HardStatistics & Experimentation RoundA/B TestingZomato-specific

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

    How to answer: For ramp-up, I'd start with a small, low-risk user segment (e.g., 1-5% of eligible users) and monitor key operational metrics (latency, errors, database load) before gradually increasing exposure. Holdback would involve reserving a small, representative control group (e.g., 5-10% of the original control group) that never sees the new feature, even after launch, to measure long-term novelty effects and ensure no negative impact on core metrics. Post-launch monitoring requires establishing a comprehensive dashboard tracking primary success metrics (review submission rate, quality, user engagement), guardrail metrics (app crashes, uninstalls), and operational health, with automated alerts for significant deviations.

13. Q13

    Midway through the Food Delivery test, tracking for Zomato Gold changed. How would you decide whether the experiment results are still usable?

    HardStatistics & Experimentation RoundA/B TestingZomato-specific

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

    How to answer: A strong candidate would first identify the specific change in Zomato Gold tracking and its potential impact on the experiment's primary or secondary metrics. They would then analyze the timing of the change relative to the experiment's start and the observed impact on key metrics for both control and treatment groups post-change. The decision hinges on whether the change introduced bias, specifically if it affected one group more than the other, or if it significantly altered the underlying user behavior being measured. If the impact is uniform or negligible, results might be usable with caveats; otherwise, the experiment may need to be restarted or the affected period excluded.

14. Q14

    Two overlapping experiments on Dining both affect commission revenue per order. How would you detect and manage interaction effects?

    HardStatistics & Experimentation RoundA/B TestingZomato-specific

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

    How to answer: Detecting interaction effects between two overlapping A/B tests on Dining commission revenue requires careful experimental design and post-analysis. Key steps include ensuring proper randomization and segmentation to allow for factorial analysis, where each experiment's variants are crossed. Post-experiment, analyze the commission revenue per order for each of the four resulting groups (Control-Control, Exp1-Control, Control-Exp2, Exp1-Exp2) using ANOVA or regression to identify statistically significant differences beyond the sum of individual effects. If interactions are found, prioritize the experiment with the larger impact or redesign future tests to be sequential or non-overlapping.

15. Q15

    Zomato's Food Delivery 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 CasesZomato-specific

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

    How to answer: A strong business case would begin by segmenting the revenue drop by key dimensions such as geography, customer type (new vs. existing), restaurant type, and time of day to localize the problem. Next, investigate potential internal factors like app performance issues, pricing/discounting changes, or delivery logistics problems (e.g., driver availability). Concurrently, analyze external factors including competitor activity, local events, or significant weather changes. Finally, propose data-driven hypotheses for the most likely drivers and outline a plan for validation and mitigation.

Practice these with instant AI feedback in a live mock interview → Start a Zomato Analytics Engineer mock

Practise DSA and system design for scale; prepare product-thinking; expect a strong culture-fit round

Indicative Analytics Engineer pay in India: ~₹9–40 LPA (role-level range, not a Zomato-specific figure).

Other roles at Zomato

Analytics Engineer interviews at other companies