← DoorDash Interview Insights

DoorDash·Data Scientist·Technical Phone Screen·Senior

Senior
May 2026

Summary

DoorDash data science interview focused entirely on experiment design and metrics for three Dasher compensation and engagement initiatives. Pretty dense case-style questions, less coding, more product analytics and stats. Felt like a product sense round dressed up as a DS interview.

Questions Asked (3)

Q1

How would you design an experiment to evaluate a Top Dasher program that routes better orders to high-performing couriers? What metrics and success criteria would you use?

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

This one tripped me up more than I expected.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the goal of the Top Dasher program: to improve overall marketplace efficiency by incentivizing high-performing couriers with better orders. Then outline a randomized controlled experiment (A/B test) where eligible couriers are randomly assigned to either receive the Top Dasher benefits or not, and define a clear set of metrics to measure impact on courier behavior and marketplace health.

Pro tip: Emphasize the importance of analyzing heterogeneous treatment effects—especially how the program impacts different segments of couriers (e.g., part-time vs. full-time) and markets—to avoid masking important nuances with an overall average effect.

1. Define Hypothesis and Goals

Clearly state the hypothesis: routing better orders to high-performing couriers will increase their retention and overall delivery efficiency. Align with business goals such as reducing delivery times, increasing courier satisfaction, and improving order completion rates.

2. Design the Experiment

Use a randomized controlled trial (A/B test) with eligible couriers randomly assigned to treatment (Top Dasher benefits) or control (no change). Ensure proper randomization, sample size calculation, and consider stratification by market or courier tenure to balance groups.

3. Select Metrics

Choose primary metrics (e.g., courier retention, delivery time, order completion rate) and secondary metrics (e.g., courier earnings, customer satisfaction, overall marketplace efficiency). Include guardrail metrics to monitor unintended consequences (e.g., impact on other couriers, order assignment fairness).

4. Analyze Results

Compare treatment and control groups using statistical tests, and conduct subgroup analyses to understand heterogeneous effects. Check for novelty effects and ensure the experiment ran long enough to capture meaningful behavior changes.

5. Define Success Criteria and Next Steps

Predefine success criteria (e.g., a statistically significant increase in retention with no degradation in delivery time). Based on results, recommend whether to roll out, iterate, or abandon the program, and suggest further experiments to optimize.

Key Points to Mention

  • Randomization unit: courier-level randomization to avoid contamination, but consider market-level randomization if spillover effects are likely.
  • Primary metric: courier retention (e.g., 30-day retention) as a key driver of long-term marketplace health.
  • Secondary metrics: delivery time, order completion rate, courier earnings, and customer satisfaction.
  • Guardrail metrics: monitor impact on non-Top Dashers, order assignment fairness, and overall delivery costs.
  • Statistical power: calculate required sample size to detect meaningful effect sizes, accounting for multiple comparisons.
  • Heterogeneous treatment effects: analyze by courier segment (e.g., full-time vs. part-time) and market characteristics.

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

Q2

How would you precisely define Dasher response rate, what factors could move it, and how would you structure and analyze an A/B test for an extra pay incentive?

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

The definition part sounds trivial but it really isn't.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining Dasher response rate as the proportion of offered delivery opportunities that a Dasher accepts, then discuss factors that influence it such as pay, distance, time of day, and Dasher characteristics. For the A/B test, outline a clear hypothesis, randomization unit, sample size calculation, and success metrics, and explain how you would analyze the results to determine the impact of the extra pay incentive.

Pro tip: Emphasize the importance of considering the trade-off between response rate and cost per delivery, and mention that you would monitor for unintended consequences like Dasher cherry-picking or reduced earnings for non-incentivized deliveries.

1. Define the metric

Clearly define Dasher response rate as the number of accepted delivery offers divided by the total number of offers sent to Dashers, possibly segmented by region, time, or Dasher type.

2. Identify factors

List factors that could move the response rate, such as payout amount, distance, estimated time, Dasher experience, time of day, day of week, and market conditions.

3. Design the A/B test

Propose an experiment where Dashers are randomized into control (no extra pay) and treatment (extra pay incentive) groups, ensuring proper randomization and sample size to detect a meaningful effect.

4. Define success metrics

Specify primary metric (response rate) and secondary metrics (e.g., delivery completion rate, cost per delivery, Dasher earnings) to evaluate the overall impact.

5. Analyze and interpret

Use statistical tests (e.g., t-test or regression) to compare response rates between groups, check for significance, and consider practical implications and potential side effects.

Key Points to Mention

  • Definition of response rate: accepted offers / total offers
  • Factors: payout, distance, time, Dasher tenure, market supply/demand
  • Randomization unit: Dasher or delivery opportunity? Consider clustering
  • Sample size and power analysis to detect small effects
  • Primary and guardrail metrics (e.g., cost, Dasher satisfaction)
  • Statistical methods: hypothesis testing, confidence intervals, segmentation

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

Q3

Compare a per-order versus per-time compensation model for Dashers. What are the trade-offs and how would you run an empirical test to decide between them?

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

Favorite question of the three.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the two compensation models and their key trade-offs for Dashers and DoorDash. Then outline a rigorous A/B test design that measures impacts on Dasher behavior, marketplace efficiency, and costs, ensuring you address potential confounds and long-term effects.

Pro tip: Emphasize that compensation changes can have long-term effects on Dasher retention and marketplace dynamics, so consider running a holdout group to measure persistent effects beyond the test period.

1. Define models and trade-offs

Clearly define per-order (fixed payment per delivery) and per-time (hourly wage) models. Discuss trade-offs: per-order incentivizes speed and efficiency but may lead to cherry-picking; per-time provides income stability but may reduce urgency and increase costs.

2. Identify metrics and hypotheses

Select primary metrics (e.g., delivery time, Dasher utilization, cost per delivery) and secondary metrics (e.g., Dasher satisfaction, retention). Formulate hypotheses about how each model affects these metrics.

3. Design the experiment

Propose a randomized controlled trial (A/B test) with Dashers randomly assigned to per-order or per-time compensation. Ensure proper randomization, sample size calculation, and control for confounders like geography and time.

4. Analyze and interpret results

Use statistical methods to compare metrics between groups, checking for significance and practical impact. Consider heterogeneous treatment effects across Dasher segments and markets.

5. Decide and iterate

Based on results, recommend a model or a hybrid approach. Discuss potential long-term monitoring and iterative testing to adapt to changing conditions.

Key Points to Mention

  • Randomization unit: Dasher-level randomization to avoid contamination, but consider cluster randomization by market if spillovers are a concern.
  • Primary metrics: delivery time, cost per delivery, Dasher utilization, and order completion rate.
  • Secondary metrics: Dasher satisfaction, retention, and customer experience (e.g., ratings, complaints).
  • Potential confounders: time of day, day of week, market characteristics, and Dasher experience level.
  • Long-term effects: run a holdout group to measure sustained impact on Dasher behavior and marketplace health.
  • Ethical and practical considerations: ensure fair compensation during the test and communicate changes transparently to Dashers.

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