← DoorDash Interview Insights

DoorDash·Data Scientist·Technical Phone Screen·Senior

Senior
May 2026

Summary

DoorDash data science interview focused heavily on marketplace experimentation and metric diagnostics. Two meaty questions, both requiring you to think across all three sides of the platform simultaneously. Not a casual screen.

Questions Asked (2)

Q1

How would you decide whether DoorDash should pilot a time-based compensation model for dashers instead of the current per-order model? Walk through how you'd design the experiment, what metrics you'd track, and how you'd account for spillover effects across consumers, merchants, and dashers.

A/B Testing & ExperimentationProduct Analytics & MetricsPricing & Monetization
Author's notes

This one is harder than it looks because you can't just A/B test dashers in isolation.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the business objective and the causal question: does time-based pay improve dasher efficiency and satisfaction without harming marketplace balance? Then outline a randomized experiment design with careful metric selection and spillover mitigation, and discuss how you'd analyze heterogeneous effects and make a launch decision.

Pro tip: Emphasize that dashers are not independent—they interact with each other and with the platform's matching algorithm—so you must design for interference, e.g., by randomizing at the zone level and using switchback or cluster randomization. Also, mention that you'd pre-register the analysis plan to avoid p-hacking.

1. Define the causal question and success criteria

Clarify what 'success' means: e.g., increase dasher utilization, reduce delivery time, maintain or improve consumer experience, and keep unit economics sustainable. Specify the primary metric and guardrail metrics upfront.

2. Design the experiment to handle interference

Because dashers compete for orders, individual randomization can bias results. Use cluster randomization (e.g., by city or zone) or a switchback design where the entire market alternates between compensation models over time. Ensure sufficient power by simulating under different spillover scenarios.

3. Select metrics across all stakeholders

Track dasher metrics (earnings per hour, acceptance rate, completion rate, active time), consumer metrics (delivery time, order cancellation, ratings, retention), merchant metrics (order volume, preparation time, cancellations), and platform metrics (cost per delivery, total deliveries, contribution margin).

4. Analyze and account for spillover effects

Use causal inference methods (e.g., difference-in-differences, synthetic control, or instrumental variables) to estimate direct and indirect effects. Check for equilibrium effects: if time-based pay changes dasher supply, it may affect consumer wait times and merchant order flow.

5. Make a launch decision with sensitivity analysis

Evaluate results against pre-registered thresholds, conduct subgroup analyses (e.g., by market density, dasher tenure), and simulate long-term impact. Recommend pilot expansion only if benefits are robust and spillovers are manageable.

Key Points to Mention

  • Randomization unit: dashers are not independent; use cluster or switchback randomization to avoid contamination.
  • Primary metric: define a clear north-star metric (e.g., deliveries per hour or dasher retention) and guardrails (e.g., consumer wait time, cost per delivery).
  • Spillover effects: time-based pay may change dasher supply, which affects order assignment and consumer experience; measure these equilibrium effects.
  • Heterogeneity: effects may vary by market density, time of day, and dasher experience; plan subgroup analyses.
  • Statistical power: account for intra-cluster correlation and potential interference when calculating sample size.
  • Long-term vs short-term: pilot may show novelty effects; consider running longer or using holdout groups to measure sustained impact.

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

Q2

Order completion rate just dropped significantly. Walk through how you'd diagnose what caused it and estimate the business impact.

Root Cause AnalysisProduct Analytics & MetricsA/B Testing & Experimentation
Author's notes

Felt more comfortable here.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the metric definition and scope (e.g., time period, geography, platform) to ensure alignment. Then systematically break down the funnel to isolate the drop, segment by dimensions to find patterns, and finally quantify the business impact in terms of orders and revenue. Use a hypothesis-driven approach, prioritizing the most likely causes based on data.

Pro tip: Always tie the diagnosis back to business impact early and often—interviewers want to see that you can prioritize actions based on potential revenue loss. Also, mention that you'd validate findings with A/B tests or holdout groups when possible.

1. Clarify and Validate the Metric

Confirm the exact definition of 'order completion rate' and the time frame of the drop. Check for data pipeline issues, logging errors, or changes in tracking that could cause a false alarm.

2. Segment and Localize the Drop

Break down the metric by dimensions such as platform (iOS/Android), geography, user cohort, restaurant type, and time of day. Identify which segments are driving the overall drop.

3. Analyze the Funnel and Hypothesize Causes

Map out the order completion funnel (e.g., app open → search → add to cart → checkout → payment → delivery). Compare conversion rates at each step pre- and post-drop to pinpoint where the drop occurs. Generate hypotheses (e.g., payment failures, app crashes, competitor promotion).

4. Quantify Business Impact

Estimate the number of lost orders and revenue by applying the drop rate to the affected segment's baseline volume. Consider downstream effects like customer churn and lifetime value.

5. Prioritize and Recommend Next Steps

Rank hypotheses by likelihood and impact, and propose immediate fixes (e.g., rollback a release) and further analyses (e.g., A/B test a new checkout flow).

Key Points to Mention

  • Define the metric precisely: order completion rate = completed orders / initiated orders, and ensure consistent denominator.
  • Check for external factors: seasonality, holidays, competitor actions, or marketing campaigns.
  • Use funnel analysis to isolate the step with the largest drop (e.g., payment authorization failures).
  • Segment by user demographics, device, and geography to find disproportionate impacts.
  • Quantify impact in terms of orders, revenue, and customer lifetime value (CLV).
  • Propose validation via A/B testing or holdout groups to confirm root cause.

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