Original practice scenario: Design a hypothetical grocery fulfillment workflow where a picker proposes a substitute, the customer may respond late, and the original item becomes available again.
Read model answer +
Model the order line as a versioned decision workflow with a clear fulfillment cutoff. Stock availability, customer preference and picker action are related events rather than interchangeable truth.
Work through it
Assign an order-line identity and record the proposed substitute, its price basis and the line version against which approval is requested.
Accept a response only if it still matches the active proposal and cutoff; make repeated responses idempotent and expose conflicts to the picker.
Record the actual picked item separately, reconciling it with the accepted decision and payment adjustment before finalizing fulfillment.
Tradeoffs to discuss
Waiting longer for customer input improves choice but can delay picking; defaults need a disclosed policy and an explicit unresolved state.
Common pitfalls
A late approval must not overwrite a newer rejection or an already completed pickup.
Handle quantity changes and price differences directly rather than assuming a substitute is always one equivalent unit.
Original practice scenario: A hypothetical retail team compares promoted stores with other stores and sees higher sales. Propose an evaluation that distinguishes incremental demand from store selection and stock availability.
Read model answer +
Define the promotion effect and eligible store-product population before comparing totals. High-demand stores and well-stocked locations may have been more likely to receive the promotion.
Work through it
Prefer randomized assignment with comparable periods; otherwise document the selection mechanism and examine pre-promotion trends and relevant store differences.
Measure sales alongside availability, price, returns and neighboring-product substitution, preserving stores with poor execution instead of quietly excluding them.
Estimate effects with uncertainty and relevant store or product clusters, checking whether gains persist after the promotion and whether nearby sales were displaced.
Tradeoffs to discuss
A tightly controlled pilot strengthens attribution but may miss rollout constraints; a broader observational study needs more assumptions.
Common pitfalls
Do not compare only stores that successfully kept the product in stock after treatment.
A higher promoted-item total is not necessarily higher category demand or durable customer value.
