ByteDance · Statistics & Data Analysis

Design and evaluate a dasher bike rollout

TrueInterview

October 7, 2026 · 2 min read

DoorDash (DD) is preparing to bring back an optional program that allows current car dashers to also enroll to use their personal bicycles or e-bikes for deliveries, while still keeping access to car mode. You need to determine whether to expand this program next quarter. Answer with precision:

State the business objective and a falsifiable primary hypothesis (for example, lower median delivery ETA in dense zones without cutting dasher earnings per hour). List at least two secondary objectives (for example, cost per order, supply density).
Define one primary success metric, no fewer than three guardrails, and two driver metrics. For each one, give precise definitions (numerator/denominator), the attribution window, the unit of analysis, exclusions (such as batched orders or outliers), and directionality.
Pick a randomization unit—dasher-level, zone-day-level, or order-level—and defend that choice by addressing noncompliance (opt-in, vehicle switching), spillovers or interference (supply rebalancing, congestion), inventory constraints (for instance, only 600 bikes available for 1,200 interested dashers across 5 cities), and operational feasibility. If opt-in is required, design an encouragement RCT (invite versus no-invite) and explain ITT versus TOT; specify how you will instrument actual bike usage.
Outline your sample-size and power plan: baseline assumptions, target MDE for the primary metric, the impact of clustering/ICC, seasonality and day-of-week controls, and how you will handle unequal cluster sizes. Detailed math is not required, but be specific about the inputs you would request.
Pre-rollout plan: instrumentation to infer the actual vehicle used on each trip, data QA checks, eligibility rules, safety/compliance, and a canary plus ramp schedule with explicit stop/go thresholds.
Analysis plan: handle heterogeneous effects (hills/elevation, weather, time-of-day, restaurant wait), contamination or crossovers, and selection bias from opt-in. Describe how you would separate speed-of-travel effects from restaurant latency.
Decision framework: specify the success criteria that trigger scale-up versus rollback, and how you would monitor post-launch with long-term holdouts. Overview: This question tests experimental design, program evaluation, causal inference, metric specification, and operational analytics skills for marketplace platforms, and it falls within the Analytics & Experimentation domain.

ByteDance · Statistics & Data Analysis

Design and evaluate a dasher bike rollout

TrueInterview

October 7, 2026 · 2 min read

State the business objective and a falsifiable primary hypothesis (for example, lower median delivery ETA in dense zones without cutting dasher earnings per hour). List at least two secondary objectives (for example, cost per order, supply density).
Define one primary success metric, no fewer than three guardrails, and two driver metrics. For each one, give precise definitions (numerator/denominator), the attribution window, the unit of analysis, exclusions (such as batched orders or outliers), and directionality.
Pick a randomization unit—dasher-level, zone-day-level, or order-level—and defend that choice by addressing noncompliance (opt-in, vehicle switching), spillovers or interference (supply rebalancing, congestion), inventory constraints (for instance, only 600 bikes available for 1,200 interested dashers across 5 cities), and operational feasibility. If opt-in is required, design an encouragement RCT (invite versus no-invite) and explain ITT versus TOT; specify how you will instrument actual bike usage.
Outline your sample-size and power plan: baseline assumptions, target MDE for the primary metric, the impact of clustering/ICC, seasonality and day-of-week controls, and how you will handle unequal cluster sizes. Detailed math is not required, but be specific about the inputs you would request.
Pre-rollout plan: instrumentation to infer the actual vehicle used on each trip, data QA checks, eligibility rules, safety/compliance, and a canary plus ramp schedule with explicit stop/go thresholds.
Analysis plan: handle heterogeneous effects (hills/elevation, weather, time-of-day, restaurant wait), contamination or crossovers, and selection bias from opt-in. Describe how you would separate speed-of-travel effects from restaurant latency.
Decision framework: specify the success criteria that trigger scale-up versus rollback, and how you would monitor post-launch with long-term holdouts. Overview: This question tests experimental design, program evaluation, causal inference, metric specification, and operational analytics skills for marketplace platforms, and it falls within the Analytics & Experimentation domain.