ORIGINAL COMPANY-CONTEXT PRACTICE
Test a synthetic kitchen scheduling change
Original practice scenario: A hypothetical restaurant chain changes its order-priority algorithm to reduce waiting time. Design an experiment that accounts for shared kitchen capacity.
Written for independent practice using company context; this is not a verified question from a company interview.
A REASONED APPROACH
Model answer.
Orders sharing a kitchen interfere with one another, so customer-level randomization may not isolate the scheduling effect. Evaluate the policy at a unit and time scale matching that interference.
Work through it
- Randomize stores or suitable time blocks with a planned crossover and washout where feasible, balancing baseline demand and operating patterns.
- Predefine wait-time percentiles, cancellation, error and workload guardrails; analyze by assigned policy and account for clustering in uncertainty estimates.
- Inspect peak periods, large orders and carryover effects, recording outages or protocol deviations without removing inconvenient outcomes after seeing results.
Tradeoffs to discuss
- Store-level experiments reduce interference but need enough independent stores; time-block designs face carryover and changing demand.
Common pitfalls
- Average wait improvement can hide severe tail delays.
- Do not treat orders from the same kitchen as independent observations.