This question evaluates a data scientist's competency in marketplace analytics, causal inference, metric engineering, and experiment design by asking for precise supply–demand balance metrics, a root‑cause analysis, and randomized interventions for repeated same‑day disablements in a regional market.
Instacart offers two fulfillment modes:
When shopper supply is insufficient to maintain service levels, same‑day can be temporarily disabled. In Miami, same‑day was disabled on 2 of the last 3 Sunday afternoons.
(a) Define precise supply–demand balance metrics: real‑time shopper‑to‑order ratio, fill rate within SLA, queueing delay distribution, shopper acceptance rate, and effective capacity after time‑to‑shop and travel constraints. Choose one leading metric to trigger disablement and justify.
(b) Build a root‑cause tree. Specify the exact data you would pull (e.g., traffic by minute, batch creation rate, shopper online/active counts, substitution complexity, store closures, pay rates, weather, local events). Explain how you would causally separate demand spikes from supply drops (e.g., instrument with exogenous weather, difference‑in‑differences vs. other markets).
(c) Propose three interventions—(1) supply incentives, (2) scheduled‑to‑same‑day rebalancing, (3) demand shaping via ETAs/fees—and for each, design an experiment: eligibility, randomization unit, guardrails (cancellations, NPS), success metrics, expected impact, and key risks (cannibalization, fairness, marketplace instability).
Login required